# Danny Zuckerman — Full Text Index > Generated at build time. The curated index lives at https://dazuck.com/llms.txt — this file is the full body of every published essay and project, for clients that want to ingest in one pass. --- # Work history My path to now. I’m currently feeling out what’s next by spending time with people and companies wrestling with what AI means for organizations, education, urban life, small businesses, and political economy. If that sounds like you, [reach out](mailto:dazuck@gmail.com)! ## Recall (2025–2026) Role: COO Post-merger I owned the integration of the two teams and investor bases and drove the pivot into AI infrastructure. I personally went all-in on AI-native work early: built a shared team memory and tooling MCP; largely automated the ops function; created agent-run workflows for operations, GTM and community; led team adoption practices with CEO and CTO. - Oversaw operations for the combined team (~40 at close) through the launches of the Recall Network, the Recall Foundation, and the RECALL token — all inside nine months. - Shipped fast with a lean team — one product went from idea to 80K live users in four days. - Recall's public launch reached [1.4M users](https://www.coindesk.com/markets/2025/10/08/decentralized-ai-market-place-recall-announces-token-generation-event), 10M transactions, and 200 active agents; the AlphaWave competition alone drew 1,000+ competing AI agent teams. - Spoke at [SuperAI Singapore 2025](https://www.youtube.com/watch?v=gzDkDK5vPYA) ("Building AI's Infinite Game: Why Competing Agents Will Accelerate Intelligence") among other conferences and interviews. Deep dive: https://dazuck.com/work/recall.md ## 3Box Labs (2018–2025) Role: Co-founder & CEO Tried to reinvent the data substrate for the web, shifting power over data from platforms to people through data control and portability. Started as COO; took over as CEO in May 2021. Proud of the hard work done right towards a huge mission, and many major achievements in product, partnerships and fundraising from leaders like USV. But ultimately did not crack the market in a meaningful way. Exited to Textile (rebranded Recall) as a natural landing spot for the mission, team and investors. - Drove adoption by 100s of customers and 1000s of developers, built a thriving community, and created deep partnerships with leading web3 companies like MetaMask and Zerion. - [Ceramic](https://blog.ipfs.tech/2021-07-13-ceramic-mainnet-launch/), our decentralized event-based data network, is open sourced and still in use today as a leading technology for open and verifiable collaboration. - Won numerous awards, including BuiltIn's 'Best Startup to Work For' and web3 industry product awards. - Raised $35M in venture capital from [Union Square Ventures and other leaders](https://www.coindesk.com/business/2022/02/16/usv-multicoin-lead-30m-raise-for-3boxs-ceramic-data-network). Led the company through a complex merger with Textile (later Recall), closing what USV partners called the most complicated deal they'd ever seen. Deep dive: https://dazuck.com/work/3box.md ## ConsenSys (2017–2018) Role: Strategy & Operations Joined ConsenSys in 2017 to help reshape uPort, its flagship decentralized-identity venture, from an R&D project into a product company. - Led strategy for decentralized identity at uPort; the team shipped the world's first live government-issued self-sovereign identity pilot, with [the city of Zug, Switzerland](https://interoperable-europe.ec.europa.eu/collection/egovernment/document/swiss-city-zug-issues-ethereum-blockchain-based-eids). - Led the reorg that broke uPort into independent units — three spun out into successful independent companies. Deep dive: https://dazuck.com/work/consensys.md ## Identity Theft (2017–2019) Role: Co-Author Co-wrote _Identity Theft: Rediscovering Ourselves After Stroke_ with my mom, Stanford professor Debra Meyerson — two years examining what makes up identity by studying what happens when trauma takes it apart. - Published by [Andrews McMeel](https://publishing.andrewsmcmeel.com/book/identity-theft-debra-meyerson/) in 2019; won the [Silver Nautilus Book Award](https://www.nautilusbookawards.com/_files/ugd/01ac16_649fcab142214a45b8c368ccc14835e6.pdf). - Adam Grant: "The most powerful book I've read on the psychological experience of having and recovering from a stroke… a testament to the strength of the human spirit." Deep dive: https://dazuck.com/work/identity-theft.md ## Zearn (2014–2016) Role: Product Manager First operator role. Joined an early-stage K-5 math platform as its first PM. Shipped excellent software; helped build and run the team of engineers, designers, teachers and more; drove rapid growth in organic teacher-led usage and district-led customers. - Joined at 12 people; helped grow the team to 50 and the platform from zero to ~2M active students. - Designed and shipped Zearn's core adaptive learning engine and first iOS app. - Created number sense progression IP on the recently released Common Core standard. Deep dive: https://dazuck.com/work/zearn.md ## Bain (2011–2014) Role: Associate Consultant → Consultant Taught me how to think, communicate, and work strategically and analytically. A wide-angle lens on how technology, finance, and manufacturing companies actually work. Loved the teamwork and rigor, but left when I realized I wanted to build systems rather than advise them. - Owned data analysis, financial models, surveys, and client communications across six cases for Fortune 50 companies. - Offered early direct promotion to Consultant — a rare track in SF. Multiple clients requested my specific continuation. Deep dive: https://dazuck.com/work/bain.md ## Stanford (2007–2011) Role: BA Political Science Studied political science, behavioral economics, and technology in society. - Concentration in Political Philosophy, minor in Economics. - President of Stanford Political Union and Club Baseball. Deep dive: https://dazuck.com/work/stanford.md --- # Writing ## Which Model is All the Way Down? URL: https://dazuck.com/writing/which-model-is-all-the-way-down Date: 2026-09-29 Type: essay Summary: Multi-model workflows, specialized models like Jev, and enterprises owning their intelligence all cut against everything collapsing into one lab's OS. So where does AI value live? As recently as early August I wrote a group chat, "I've assumed for a while that all functionality will collapse into the harness, and harnesses will collapse into the OS." This was met with general consensus. It seemed models were eating everything, everyone was playing to expand their stack, and more access to more OS was going to be the primary differentiator. But that was a whole month ago. Three things have happened, or accelerated, to force a rethink. **First, the benefits of multi-model workflows persist**, and as we look to AI to do more they matter more. Any model family shares an architecture, training approach, and data set - so blind spots and weaknesses. Combining model families in execute/verification or other agent swarm approaches reduces gaps and combines strengths. This is obvious in my own work, and research backs it: [Sakana's Fugu](https://arxiv.org/html/2606.21228v1) beat the best single model on all four benchmarks it tested. Enterprises are moving the same way, with [81% now using 3+ model families](https://a16z.com/leaders-gainers-and-unexpected-winners-in-the-enterprise-ai-arms-race/), up from 68% a year earlier. If mixing models is essential, can one OS (dominated by one company and its own model) hold everything? **Second, [Jev arrived](https://typesafe.ai/blog/introducing-system-one-models-and-jev).** It excels at efficient classification, probabilities, routing, judging, and many other use cases. [Tom Tunguz found](https://tomtunguz.com/ai-comes-for-the-if-statement/) it beat a frontier model 80% to 47% on real email threads, at 82x lower cost. And it hints that after a period of large LLMs eating everything, we may head into a period of differentiation by specialization, and speciation (aka more model types). Maybe Anthropic, OAI, Google, etc. can replicate Jev quickly ([Kev](https://github.com/jaredpalmer/kev) already did), but then won't that type commoditize, and every other type besides? Either way model types multiply, and a harness (consumer) or platform (enterprise) that can orchestrate independent model architectures has a growing case for independence. **Third, [Pat Grady's great video](https://x.com/gradypb/status/2103536289438130670) makes the case for enterprises wanting to own their intelligence.** Lab APIs offer a handful of points on the cost/performance curve, and real workflows fall between them. Owning and tuning your own models is how you get on the Pareto frontier for every workflow (and role in a multi-model workflow, presumably) you run. This would have seemed crazy in January, except to the most advanced organizations. Most large enterprises were just learning how to use this stuff, and the conventional wisdom was that models are commoditizing, so why train your own? The labs still capture most of the dollars today; open models were [under 4% of spend](https://trendingtopics.eu/open-weight-models-from-china-are-capturing-a-growing-share-of-ai-usage/) on Vercel's gateway even at nearly 30% of tokens. But as we shift rapidly from tokenmaxing to a consensus that inference spend is going to increasingly dominate costs, maximum efficiency and performance is a natural business imperative. That suggests a collapse into the dominant models and OS's is far less likely, on the enterprise side at least. On the surface, consumer looks like a different story, with a recent onslaught of do-everything agents in a good interface. [Instinct](https://techcrunch.com/2026/09/28/viral-ai-agent-instinct-raises-1b-series-c-at-a-10b-valuation/) is a $10B company growing 10% per day; Muse [hit #1 on the App Store](https://www.cnbc.com/2026/09/21/meta-muse-personal-ai-agent-downloads.html) ten days after launch. The narrative today is *most users just want something easy and simple*, not Claude Code or OpenClaw to battle with. But that mirrors the enterprise/developer narrative nine months ago, at the "one model for everything" stage. Given the pace and cycles of AI, it seems likely consumer is just at an earlier stage, and differentiation comes there too. Many people, e.g. [Dennison Bertram](https://x.com/DennisonBertram/status/2104712154800632108), think so. Which all leads back to the original question: **does everything collapse into the OS, and what is the OS for all these agents, whether one model for all or many used together?** Software-defined harnesses were all the rage just a few short months ago, but as apps build in agents and platforms build in agents and labs build in agents, will they be squeezed out? Or will enterprises insist on owning their harness as the control plane of intelligence? Or will the OS settle closest to the hardware - Android/iOS for consumer, and the data center itself for enterprise? That buys maximum access, scale and raw intelligence. And it seems an implicit hypothesis for Google as they prioritize Cloud over Gemini, and for Nvidia as they [buy Hugging Face](https://www.sec.gov/Archives/edgar/data/0001045810/000104581026000078/nvda-20260902.htm) and add to their developer platform. Or can vertical players carve off segments from the hyperscalers? Thomson Reuters shipped a model it says is competitive with the strongest frontier models, and Harvey post-trained its own on an open-weight base. Vivek Ramaswami and Sabrina Albert call these ["neo neo labs"](https://aspiringforintelligence.substack.com/p/the-rise-of-the-neo-neo-lab), ready to serve their segments with more custom intelligence. ![models-all-the-way-down.png](https://dazuck.com/images/models-all-the-way-down.png) In most of these scenarios, we get the proverbial "models all the way down," with a model trained to help you manage, observe, and continuously tune the other models as needed. **Which model will live at the bottom of that stack?** ## If Software Is the New Marketing, What's the New Product? URL: https://dazuck.com/writing/software-is-the-new-marketing Date: 2026-03-20 Type: essay Summary: Software has no moat, but intelligence may. Four forms of intelligence — in the model, the data, the people, and the network — with very different defensibility. Recall Labs CEO Andrew [recently pointed out](https://x.com/i/status/2029652435413913609) that "software is the new marketing." As software's cost to produce rapidly falls, moats for pure software are disappearing. This is increasingly true for even sophisticated software. At the same time, the cost to produce content is falling even faster, so marketing channels are absolutely overloaded with AI slop. Software is becoming marketing because it's cheap to produce but not free, and with a far easier install path for even technical and niche tools ("Claude, install this") shipping OS tools or skills or full products is becoming the best way to cut through the noise of AI content slop and get attention for a person or company's capabilities. This begs the question: if software is the new marketing, what's the new product? For most companies in the knowledge economy, the product is shifting to intelligence. Software is a layer to access or leverage intelligence for these companies. (For others that touch financial transactions, physical goods, transport, etc. the dynamics are different). Software has no moat, but intelligence may. The key question is: where will intelligence live, and who will control it? We're seeing the emergence of four forms of intelligence, with major questions still about how distributed or concentrated it will be. ### Intelligence in the model Foundation labs ship consumer products — ChatGPT, Claude Code, Gemini — to create leverage for what the underlying models can do. Given the capex and scale involved, this will likely be concentrated among a handful of players and a long tail of open source alternatives that keep the big labs honest on pricing power. We may yet see a proliferation of foundation models in specialized domains (the 'medicine model'), but each new generation suggests that generalized models may eat more intelligence. It's clear either way that models are eating a lot of intelligence. It's less clear what could be held out and keep value as a complement. ### Intelligence in data This is where a lot of software companies want to plant their flag: proprietary data, insights and intelligence. The pitch is that models are trained on the open internet, and certain additional data, but curated, proprietary, hard to recreate data can augment that in highly valuable ways. It's clear there's some value here, but less clear how much pricing power and defensibility there will be. There is evidence on both sides of this. Tales from the "models ate my business" side (and there are many many more): - When Anthropic launched a Meta ads connector, Ryze AI's deal close rate [collapsed from 70% to 20% in weeks](https://finance.yahoo.com/news/claude-just-killed-startup-sf-173105543.html). - Jasper AI's revenue roughly halved after its marketing-copy moat turned out to be one context window away from commoditized. - Chegg's converted visits dropped 89% year-over-year once Google's AI Overviews absorbed the queries it monetized. Meanwhile, the big labs are sourcing domain data directly — Anthropic and OpenAI both launched healthcare-specific products in January, integrating medical coding databases and clinical records. At the scale these labs are operating at, they can find training data extremely widely. For inference data & context, data portability regulation is accelerating: the EU Data Act took effect in September 2025 (US CFPB's personal financial data rules hit the largest institutions in April 2026), giving customers the right to port data from connected devices, and the labs are also racing to connect to enough of this that it becomes table stakes for any piece of software — people simply won't keep using software that refuses to share because the bar has been raised for context awareness. But models won't eat all data. The most clear example is that companies whose data intelligence is generated through physical processes (or other hard to re-create processes), seems defensible for now. OpenAI and Anthropic are not going to immediately compete on self-driving with Tesla and Waymo, who have been building their training sets for years. Fero Labs predicts product quality from individual manufacturing line telemetry — each steel mill's blast furnace has unique dynamics no foundation model has access to. CAPE Analytics layers years of labeled insurance loss data over aerial imagery to produce property-level actuarial risk scores that a general vision model can't replicate. Recursion Pharmaceuticals runs 2.2 million samples per week through automated wet labs, generating 50+ petabytes of biological data that emerges from physical experimentation, not text corpora. Data that's fundamentally textual, public, or derivable from public information is being largely absorbed because it can be relatively easily replicated in an age of AI agents, abundant data, and synthetic data. Data generated through physical-world interaction — sensors, labs, regulatory processes, verified relationships — resists absorption (at least until intelligent robotics scales). ### Intelligence in the people (services) a16z recently [framed the shift](https://a16z.com/services-led-growth/) as "services-led growth" — selling outcomes to clients rather than tools to workers. Forward-deployed engineer roles are up 800-1000% year-over-year. Decagon prices per conversation handled, not per agent seat. The inversion from Software as a Service to Services by Software goes deeper than pricing, though. Software is becoming the marketing of choice for highly capable people and teams to advertise their hands-on services. Trail of Bits, the security firm, built and open-sourced Slither and Echidna. The tools prove they understand the attack surface better than anyone — so they get hired for the audits. thoughtbot, the Rails consultancy, open-sourced Factory Bot and dozens of other tools — collectively over a billion downloads. They don't cold-call. Their GitHub is their sales deck. The audits fund better tools. This arc is accelerating and reaching wider. Patrick McKenzie built small SaaS products for years — but the real business was always consulting, where he went from $100/hour to $30,000/week. The software was proof of work. Simon Willison co-created Django, built Datasette, and now consults on AI tooling — with a readership in the millions that came from building in public. The tools and experience with software are the pipeline. When anyone can build the tool, what matters is the judgment behind it. Knowing what to build for a specific client, how to weave it into their messy organization, and where it creates leverage they couldn't see themselves. That's the intelligence. The software proves you have it. In the age of AI, everyone has more leverage, and the best people and teams are differentiated even more than before from the rest. They'll command higher prices, and while software as a service margins will fall, services enabled by software margins will rise. But how do you know who is the best, in a world filled with slop? **Working, effective software is the proof of work.** ### Intelligence in the network The last form is the trickiest. It's about keeping multiple parties intelligent with respect to each other — making sense of information scattered across organizations, systems, and people who don't naturally talk. Jamin Ball's recent writing on systems of record [makes the case well](https://cloudedjudgement.substack.com/p/clouded-judgement-121225-long-live): as automation increases, the importance of a canonical source of truth increases, not decreases. _"The more we automate,"_ Ball writes, _"the more important it becomes that someone has done the unglamorous work of deciding what the correct answer is and where it lives."_ But a source of truth is, at the end of the day, a database. Within a single enterprise, coordination software is often a well-designed application on top of shared data — a pretty thin moat. Enterprises providing systems of record are going to have a much tougher time charging exceptional margins when CTOs and CIOs believe they can replicate much of the functionality and interoperability built atop the core database easily, and just do it on their fully owned and controlled DBs. The moat gets thicker when coordination crosses organizational boundaries, where you need governance rules, dispute resolution, and trust protocols between competing interests. That's intelligence embedded in how parties stay coherent — not just what they store. Visa operates a rule book across billions of parties — dispute resolution, fraud scoring, settlement logic built over decades. Surescripts handles 95% of US e-prescribing by coordinating trust between EHRs, pharmacies, and benefit managers — coordination so effective the FTC sued them for monopolization. The joint decision making to migrate here is a bigger moat. But even strong moats have limits: Ball's own Amadeus example is telling. Amadeus still owns the system of record for travel, but Booking and Expedia captured seven times the value by building the front door on top. Top companies are racing to introduce stablecoin systems to compete with Visa. And regulations + technology are making interoperability more of a standard, reducing the strength of switching costs for cross-boundary coordination layers. There's still very real value in anchoring multi-party trust, but the moat is probably far more in the trust than the technology. ### What intelligence do you offer? The knowledge economy world is being repriced. For the vast part of the economy that we used to call the software industry, the product is shifting from the code to the intelligence underneath it. The major question for companies and investors navigating this transition is what intelligence do generalized models eat, and what intelligence can effectively complement it. Companies that figure this out will create mind-blowing software at mind-blowing speeds to market just how powerful their particular intelligence is as a complement to SOTA models. Companies that don't figure this out may find themselves waiting in fear of each new Anthropic release. ## If AI Builds, Do Humans Shape? URL: https://dazuck.com/writing/if-ai-builds-do-humans-shape Date: 2026-03-19 Type: essay Summary: When autonomous agents write and ship the code, the human job isn't review, specs, or management. It's shaping: deciding what should exist next. At Recall we've been running a 'dark software factory': autonomous AI agents pick up goals, write the code, open pull requests, run tests, fix failures, and ship. We've had hundreds of goals executed, hundreds of commits per day. Most of the code in our newest repos was never typed or reviewed by a human hand. This has pushed us to live one of the more interesting questions in software teams right now: **if AI writes the code, what do humans actually do?** ### It's not review, specs, or supervision These were the theses last year. AI does this fine now, at a pace humans can't keep up with. We'll share more about how in a future post, but for now we'll just say it's pretty clear humans spec'ing just get in the way, and humans reviewing is an untenable bottleneck. A more recent theme has been to treat the human role as management. Set priorities, assign work, track progress. But this is just project management with a faster workforce. You end up in Jira-with-robots, which is exactly as depressing as it sounds. ### Jamming, shaping, taste-ifying Increasingly we are spending our time "jamming" at a virtual whiteboard, shaping what the AIs will work on. ![The Studio's shaping canvas: goals ranked by priority, each with a readiness ring](https://dazuck.com/images/studio-constellation.png) Shaping is the work of defining what should exist next in our product. It's not spec'ing — we _do not touch_ how things will be built, any implementation details, quality gates, none of it. (We actually have to train ourselves to stay away from these, because the AI software factory does worse when we interject there.) Shaping is making sure that a Goal is well understood, defined, and aligned to the taste and stakeholder needs we're trying to serve. It's ensuring we know the outcome we want, the scope that makes sense, and how it relates to our strategy. It's the input to the software factory. To shape effectively, we built ourselves an app we call "The Studio" — a virtual whiteboard to make this new job as full-flow as possible. ![A goal in The Studio: outcome, why, acceptance criteria and deliverables, with supporting evidence on the left and AI analysis on the right](https://dazuck.com/images/studio-goal-detail.png) Goals are just GitHub issues, but we definitely don't want to spend our time in the GitHub UI. So we pull each issue into a rich card, enriched with context from our internal systems — a tool called Atlas that curates context from across all our data — and project out the impact, risks, and strategic considerations of each goal. And because we know what makes a factory-ready goal, the UI shows us where we need to tweak things. ### This is...fun? This has reduced our team's dread about the future of software jobs. Shaping IS the most fun part — the strategizing at the whiteboard with all the creativity and lightbulbs, before you have to go do the tedious parts of making something real. Shaping IS the thinking. We sit with a spatial canvas of goals and see them in relationship to each other — which ones reinforce, which ones conflict, where the gaps are. ![Resonance view: goals the AI has clustered into themes such as Auth and Access Control, Data and Compliance, and User Experience](https://dazuck.com/images/studio-resonance-demo.png) ![Tension view: pairs of goals the AI flagged as conflicting, each with a one-line reason](https://dazuck.com/images/studio-tension-demo.png) We do NOT spend time tagging and organizing and updating properties of goals. AI does this for us. We asked our Studio for layouts that show us which goals are in tension with each other, which amplify each other, and which conform (or don't) to our stated strategic direction. The AI generates these views dynamically — they aren't manually maintained. When you remove the implementation burden, the remaining work is the most interesting part. Deciding what should exist. Understanding why. Defining the boundaries. Spotting the tensions between competing priorities. These are deeply human capabilities — spatial reasoning, judgment under ambiguity, strategic sense-making. The factory can't do these. We can't do what the factory does. The division is surprisingly clean. ### The shape of what's coming I think every team that adopts autonomous AI execution will converge on some version of this. Not immediately — the first wave is supervision (review all the PRs), the second wave is management (prioritize the backlog), and the third wave is shaping (define outcomes with enough precision that supervision becomes unnecessary). The tools for shaping don't exist yet, mostly. We're building one — a spatial canvas where goals, context, and relationships can be viewed dynamically. AI assists at every layer: scoring readiness, discovering relationships between goals, coaching on strategic alignment, and surfacing what the factory's execution patterns reveal about your goal quality. But the tool is less important than the realization: the human job in an AI-native team isn't going to be managing the AI — that will just get in its way after the next round of harnesses. It's going to be learning, thinking, and shaping what you want your dark software factory to bring into existence. ## Chief Repeat Officer: Reminders and AI System Design URL: https://dazuck.com/writing/chief-repeat-officer Date: 2025-01-12 Type: note Summary: Why AI teams require repetition even more than human teams, and how to build structural reminders into your system design. ## Jan 12, 2025 — Chief Repeat Officer Executives are Chief Repeat Officers. Say the same priorities over and over until they stick. This is just as true when your team includes AIs, but it looks very different. With humans, you're a repetition engine. You say goals and strategy in the all-hands. You say it in the 1:1. You say it reviewing the doc. You make a video saying it and share a doc saying it. AIs need repetition even more than humans. Give the goals in the first prompt and it might be forgotten by message 10. The context window may be long, but attention drifts. In one of my projects this weekend I have super tight persona and responsibility definitions and flows that ensure they are seen every session. I had documented that a Product Officer Agent should "own the positioning, packaging and persona - don't worry about immediate next builds." A few sessions later, that agent kept directing all its work towards the micro items on the roadmap. AIs like tangible, concrete work in front of them and tend to forget meta things as they get buried in more tokens focused on detailed things. So you have to build reminders into the system - not brute repetition but system design. Skills, loops, system prompts, context files—structures that keep surfacing what matters most. In this case I created a few hooks that bring back the primary directives within sessions to prevent drift. You're still the Chief Repeat Officer. You need to make sure the key things come back into your team's context at runtime. With humans, you repeat until it lands (at which point it's probably about to change). With AI, you design the system so it repeats for you. --- ## Getting AI to Improve Itself URL: https://dazuck.com/writing/getting-ai-to-improve-itself Date: 2025-01-08 Type: essay Summary: How a custom handoff command and project diary made my AI collaboration compound over time. Over the holidays I built a family app—meal planning, fitness tracking, household coordination. Ten days, about $50 in API costs, and it works well enough that my wife uses it too. I built this to test a theory: **building exactly what I need with AI is now better, faster and cheaper than adopting professionally built products** for a massively expanded set of things, even as a non-engineer. This isn't exactly controversial right now. But the reason it's true isn't just that building is easy—it's that maintaining, extending, and improving what you build is easy too. The interesting part of my holiday project wasn't the app. It was how easy it was to build a system that improves itself—and therefore keeps getting better at building and maintaining my apps over time. I tried ~a dozen different new tools and strategies across my Claude Code-centric approach, from prompting strategies to claude.md configurations to executor-evaluator patterns (which I highly recommend). Two tactics really lifted my system: a custom /handoff command and an enforced project diary. Between the two, my agents started to work, learn and improve together over time in a way that made my job easier and easier as we went. ## The /handoff command Until this project I mostly relied on claude's built-in compaction. And while I knew it was faulty and degraded performance, I didn't want to go through the effort of context curation and didn't trust myself to capture everything I needed. I resolved to buck that in this project. After a few sessions and some inconsistent git committing in between dashing to family dinners, I decided to create a custom `/handoff` command that I could run as a work session was winding down. It would 1. **Run tests and commit changes** — Making sure we wrapped work vs. left small threads hanging. 2. **Update the project diary** — Capturing decisions, learnings, open questions to pair with the commit history (see next section) to reduce fear & downside of closing this session. 3. **Generate a prompt for the next session** — What to pick up, what context matters, what might go wrong - this made my default switch from preserve the session to handoff, restarting my context more frequently. 4. **Reflect on the session and suggest improvements to system instructions** — Based on what happened this session, what should change in root and project claude.md, what hooks or skills to add, etc. That last step did a lot of heavy lifting early. After one session where a sub-agent produced code with lint errors, the handoff suggested: > "Always run lint after integrating subagent code." After another where an evaluator agent caught a feature silently using mock data, it suggested: > "Periodic product-level reviews catch integration gaps that pass technical tests." These weren't generic tips. They were specific learnings from actual work - _my_ actual work and prompting habits - proposed back as durable rules. I'd review the suggestion, say "yep, do it," and the instruction set would update. The first batch of handoffs all produced several meaningful improvements. After ~5 rounds, the suggestions tapered and sessions usually ended with "the system is working well, no improvements suggested." About every 8-10 sessions a tweak would be suggested. The system had learned my preferences and didn't need to keep adjusting. The handoff also helped my own context management. Each one ended with a prompt for the next session plus a pre-mortem on upcoming work. I could close my laptop mid-project and pick it up days later without the usual "where was I?" friction. Between the commits, the diary, and the handoff notes, neither the AI nor I lost track of decisions or goals. ## The Project Diary The project diary became a critical artifact throughout the build. Git tells you what happened—files changed, commit messages. The diary captured _why_ a decision was made, what alternatives we considered, what we learned that should change future work. It's context capture. | Git tells you | Diary tells you | | -------------- | -------------------- | | What happened | Why it happened | | Files changed | Tradeoffs considered | | Commit message | What we learned | Every session—including sub-agents—logged to the same diary. Some actual entries from my build: **December 25:** > "Used `recipe.url` but type was `recipe.source`. Build failed immediately, preventing runtime error. TypeScript earned its keep." **December 28:** > "Chat.tsx had task mutations declared with underscore prefix (`_createTaskMutation`)—a common pattern to suppress 'unused variable' warnings. But they were NEVER called. The UI showed 'Added: {task}' confirmation cards without actually executing any API calls." > > _Learning: Underscore-prefixed variables are a red flag for dead code. If you're suppressing unused warnings, question why the code exists._ **December 28:** > "Pantry sync used `void syncPantryItemToNotion(...)` to suppress the promise. The tRPC endpoint returned success before knowing if Notion write succeeded. Users saw 'Added!' but items never appeared." > > _Learning: Fire and forget is only appropriate when failure is truly non-critical AND users have no expectation of immediate results._ The underscore-prefix bug and the fire-and-forget bug would have shipped without the diary. Writing down learnings forced reflection that I'd otherwise skip. The diary also tracked architectural decisions, so when a future session questions a past choice, the reasoning is there. | Decision | Choice | Rationale | Revisit If | | ----------------- | -------------------------- | --------------------------------------------------------------- | --------------------------------- | | Component library | None (Tailwind only) | Simpler, no abstraction needed for family app | UI complexity grows significantly | | Error handling | Surface errors, never hide | App makes health/food recommendations; failures must be visible | Never | With the diary going, I was able to stop maintaining a detailed backlog or set of specs, or worry so much about capturing notes as I went myself, because the diary made it easy to retrace what happened, where we were, and what needed revisiting. I knew I wanted to share learnings from this project with my team - the diary made it dead simple to synthesize later on. ## AI collaboration starts to feel like org design I started creating guides, skills, and custom commands beyond just the handoff in Claude Code. At first this felt overwhelming—so many different ways to achieve similar things. But I found it high-leverage to continually refine, and I also found my mindset shifting over time. **Eventually it stopped feeling like configuring a tool and started feeling like org design.** My root instructions are company-wide policies. Project instructions are team norms. Guides are department playbooks. Custom commands are specialists I can call in for specific tasks. The project diary is institutional documentation and memory. When I add a new guide or refine an instruction, I'm not tweaking settings. I'm deciding: what values apply everywhere? What rules are context-dependent? What skills get called in when? It's the same set of questions you'd ask designing how a team works together. The frame changed how I invest my time. Instead of optimizing individual prompts, I'm building infrastructure that compounds. Each project benefits from the last one's learnings. Each session builds on previous sessions instead of starting fresh. I suspect this is where things are heading. As models get more capable, the differentiator won't be who has access to better AI or uses it more. It'll be who builds better systems for their AIs to collaborate with each other and themselves. --- _I'm still refining this—particularly what belongs in the diary vs. what's noise. What systems are you building to make AI collaboration improve over time?_ ## Points: How Reputation & Tokens Collide URL: https://dazuck.com/writing/points-reputation-tokens Date: 2024-02-23 Type: essay Summary: How points programs, reputation systems, and token economics are converging — and where it goes from here. Points have taken Web3 by storm in the last six months, catalyzed by projects like Blur and EigenLayer rewarding users with points on the way to seizing the NFT market and amassing $7 Billion TVL respectively. More than 115 Billion points have been given out by Web3 projects so far, according to Tim Copeland at The Block. There are two ways to look at points: 1. **As a precursor to an airdrop.** Projects use points ahead of a token to generate interest, signal what they care about and will reward, more effectively target engagement, and navigate legal risks associated with tokens. 2. **As a measure of quantifiable reputation**. Points ascribe a value to user activity, just like many reputation systems have before: traditional loyalty programs, Reddit karma, check-ins, credentials. They can signal legitimacy in pseudo-anonymous systems and — because they're more quantitative than, for example, verifiable credentials — standing within the community. Both of these are right. Points align the incentives of the platform and the user base, like all reputation systems. And they forecast who is creating value and is likely to be rewarded. By understanding how these two intersect, we can forecast where Web3 will go far beyond today's points craze. ## Points are quantifiable like money, enduring like reputation Tokens were the first major innovation of Web3 and the primary incentive. They offer fully quantifiable value, are transactional, and require no additional context. They work as "one time games." Reputation is how social systems achieve repeat game use cases, rewarding 'good behavior' of an actor (e.g. following contracts and policies, not cheating counterparties) with access and benefits over the long term. Reputations are non-fungible — for them to establish trust, it has to be hard to buy reputation. Points are proof of activity that act as a building block for reputation (and in Web3, often carry the suggestion of future value). **All reputations come with some benefit.** Usually, they're more subtle than financial rewards. Reputations might gain access to a service (credit score) or club (referral), earn discounts (loyalty programs) or introductions (dating), convince counterparties to transact (Uber rating, credit lending), and build trust with customers (influencers, corporate brands). Reputations are less measurable than financial assets, but often more valuable. For Web3 to grow into social and other non-financial use cases, more robust reputations are needed. Points are not an isolated mechanism to forecast token earnings — they are one point on a broad spectrum of token (financial) and reputational rewards that Web3 will keep innovating on. ## Points as part of the evolution of Web3 reputation Reputation naturally starts in the most discrete, high-value places and evolves to more broad ones. ### 1. B2C "Badges" The earliest forms of reputation helped networks solve discrete pressing problems: anti-sybil and KYC. This involved a business or service issuing badges to users for achieving important milestones. For example, Gitcoin Passport stamps prevent sybil attacks in Grants rounds for Ethereum and other ecosystems. The meaning of the badge is objective and clearly denominated. ### 2. Attestations After discrete badges, platforms needed reputation for a wider variety of activities and credentials: history of contribution, trust as a delegate, skills in a DAO, proof of activity. Attestations are still clearly denominated, but rather than signifying a clear milestone like badges they're more continuous. This also started B2C (for example, participating in an event, completing a certain step in a user flow, etc.) EAS has emerged as a standard for issuing these, used natively in the Optimism stack and widely in Ethereum. Increasingly, attestations are community-driven as well. Open platforms let users create and verify claims. For example, Metamask is working on a review and reputation system for Snaps. ### 3. Points: scored activity On-chain transactions, B2C badges, and attestations are all cryptographically verifiable actions. Users do something that has provable provenance and time, whether that's on-chain or off-chain signed data. Point systems specify which of these activities have value in their system (and how much), tabulate them over time for each identity, and ascribe a numeric 'point' value to them. Points create a quantifiable reputation that continuously aggregates the previous forms. ### 4. Events: all activity There's no reason to believe the evolution will stop with points. At scale, all of a user's activities in Web3 apps and platforms will be recorded as cryptographically verifiable events. This might be likes on a cast, messages on a forum, contributions to a codebase, visits to a blackbird restaurant, etc. These are all events, and they all have value — but it's not always known up-front what that value is. Some might have value in driving community engagement, others in improved analytics for products or targeting for ads, some as inputs into reputation systems. Because all will be cryptographically recorded, they can be referenced any time in the future. Points will dominate for now, but before long we'll see a huge increase in retroactive airdrops, activations, rewards, access, and other forms of value awarded to users based on a much broader history of their events — not just those that are made explicit up-front via point systems. ## Infrastructure for points, events, and trust All of these forms of reputation serve to reward users. Web2 used points exhaustively, but Web3 can uniquely do it openly and with composability. By making every event driving points both transparent and verifiable, events and points can be leveraged across platforms and have trust built in. This trust can reinforce future rewards, encourage more activity, and enable cross-platform point innovation. Unfortunately, while points are proliferating, to date most haven't tapped into this unique Web3 possibility — most have been tabulated on centralized databases, putting rewards at risk. ### Data ledgers vs. asset ledgers Financial blockchains were built to be asset ledgers, not point or event ledgers. They're designed for scarcity; e.g., they must protect against double-spend as a core principle. Points are not bought, sold or traded like assets — they're earned. They're best served — fast, cheaply, scalably — on a data ledger. Data ledgers, for data that is not scarce, operate with different principles. They must still offer strong verifiability and composability; but they don't have to protect against double spend, and they must scale many orders of magnitude beyond asset ledgers. There are exponentially more data transactions than asset transactions in any web service. Ceramic is the first _decentralized data ledger_ with the scale, composability, and trust guarantees required to be a system of record for points and events. It's built to enable the scaling of Web3 beyond financial transactions to richer experiences, including those powered by attestations, points, and the billions of events that are coming to enable a data-rich Web3. ### Building with Points If you're thinking about a point system for your product, or how to advance point-enabled experiences, please reach out to partners@3box.io. If you are interested in emerging standards for points, reach out to us on the Ceramic discord to learn more about our working group. If you'll be at EthDenver next week, come talk points, reputation and verifiable data with us at Proof of Data. ## Thinking of getting into blockchain? 6 things to consider URL: https://dazuck.com/writing/getting-into-blockchain Date: 2021-05-10 Type: company Summary: Practical advice for people considering a move into Web3 — what's real, what's hype, and what to look for. In 2017, during crypto's last bull run and after years of pointedly ignoring blockchain hype, I jumped down the rabbit hole. I began to see crypto networks as the best coordination tool we've ever had. I met the most passionate, curious, interesting people I've ever worked with. And I loved that the challenges to overcome were organizational, economic, and political as much as technical. In the past four years, I've collaborated with dozens of teams, helped found 3Box Labs (a company) and launch Ceramic Network, and worked with several DAOs. The potential, challenges, and excitement in the space are still similar to 2017. But the range of experiments, opportunities, and team types is now incredibly broad. ## What to consider as you jump down the rabbit hole ### 1. Join a community, not just a team. One of Plaid's hiring criteria was "Did they read the docs?" In the crypto space, it's "Did they join the community?" Hop into the discord or slack, get a sense of the developers and users and how they are engaged. Picking a team without understanding the community would be like getting engaged before meeting your partner's friends or family. ### 2. Be honest about what you're in it for, and choose accordingly. The pace of activity in the space is completely overwhelming. Make sure you land in a spot where that will drive your motivation not drain it. If you enjoy constant, multi-disciplinary, frenetic learning and excitement, happy hours that veer back to interesting 'work' topics, and the feeling of a quasi-cult-like movement — you've found your tribe. Look for teams that were building through the last few years' bear market. Protocol teams, DAOs, and the early DeFi builders. Join one working on something you really care about because you'll need deep motivation to keep focus and energy. If you want more balance or stability, the bull run has given rise to more traditional companies — exchanges (the new banks), infrastructure providers (picks and shovels), and creative studios — that build on this movement but in a more familiar setting. It's a bit more sustainable and 'normal.' If you are in it for the money, there are plenty of DeFi projects with millions of dollars to spend. Or consider keeping your stable income and investing it in crypto. ### 3. How do you like to work? While the world has been adjusting to remote work, many crypto teams are native to it and push organizational experiments far beyond that. - Do you want an office with teammates? Select for this, as most are remote - Real-time or asynchronous? Plenty of teams work purely off PRs and tickets, gaining massive flexibility but losing face time. - Radically flat, traditional org structure, or experimental? Do you want the empowerment promised by new org designs or the clarity of clear structures? **Could you work for a protocol?** Many projects aim to (or already have) pass most power to a DAO or community. See how Yearn Finance describes itself. Does the idea of working with complete freedom and a loose association with other community members excite you and feel like the next era of organizations? Or does it seem like the death of coherence, agility, and coordination at scale? Understand a team's long-term plan for the company and DAO, and if/when that is in the future. ### 4. How do you gauge progress? Teams have different DNAs for how much to focus on technical advances, growth and sales, the long-term vision, or product adoption. This gap is wider than normal in the blockchain space because we're so early — there's so much green space, less obvious metrics, more delayed feedback, and some self-delusion about what's working and what's not. Understand if teams are engineering, vision, or product-led and what the upsides and downsides to this are, and think about whether those will match your own instincts. If you are instinctively drawn towards growth but the team holds ideology more dear, you could be very frustrated with the roadmap and strategy. ### 5. Understand crypto compensation. Startup employees are used to salary, bonus, and equity. In crypto, there are also tokens. They are more liquid, more volatile, often higher upside, and sometimes far less tax-efficient than stock options. Know your situation and what combination of income you care about, and make sure the team you join is keeping the incentives of the founders, team, and investors aligned for the long term. This article offers a good primer. ### 6. Principles and values matter more than ever. Culture is what defines behavior in the absence of clear rules, and in projects with smaller teams and bigger communities, the early norms and values carry more burden than ever. Understand both what the team stands for and how they instantiate their values in their operations and their community. As Vitalik said, "Even a billion dollars can't compete with a project having a soul." Make sure you believe fully in the culture being created and can see yourself as one of its core carriers. ## 3Box Labs is hiring At 3Box Labs, we are on a mission to transform the web's relationship with data. If you are impact-driven, curious and bold, and care deeply about improving the web, we'd love to hear from you — we're hiring across our Engineering, Product, and Growth teams. Our team is built around 1 purpose, 5 core beliefs, and 8 guiding principles as we work to build a community that can have as much impact as our tech. This means: - Tangible impact on the future of the web from the center of the decentralized tech movement - Ownership over novel challenges as we scale and tokenize Ceramic Network - Collaboration with a world-class technical team (and partners) to create protocols and products - Constant feedback and usage from a rapidly growing open source community - Remote and flexible work, with frequent regional meetups and team retreats in amazing places - Huge upside through equity and the network, globally competitive comp, great benefits Find all our open jobs here, or reach out to jobs@3box.io and tell us what you'd like to do and why you'd be a great addition. ## Tokens and Communities: the existential question in crypto URL: https://dazuck.com/writing/tokens-and-communities Date: 2021-04-08 Type: note Summary: Can DAOs bring virtue back to scaled organization, or will tokens accelerate economics-only mentalities? What will be the relationship between tokens and communities? This might be the existential question for crypto, and most people in the space are barely asking it. Here's the optimistic version: DAOs use tokens to coordinate communities at scale while preserving the values and purpose that made the community worth joining. Tokens align incentives, fund public goods, and enable governance. Communities get the best of both worlds — the cohesion of small groups and the power of large networks. Here's the pessimistic version: tokens insert market logic into every community, crowding out the non-financial motivations that made those communities functional in the first place. Every DAO evolves into a profit-maximizing network wearing community clothing. I'm worried the pessimistic version is more likely, and here's why. --- Humans are motivated differently in different contexts. In markets, we optimize for self-interest — and that's fine, markets are designed for it. In communities, we're motivated by reputation (wanting others to think well of us) and morality (wanting to do the right thing). These motivations are powerful, and they produce behaviors that markets can't: volunteering, mentorship, generosity, trust. The problem is that market and community motivations don't coexist well. When you introduce market incentives into a community context, the market framing tends to dominate and destroy the community framing. The daycare study from Haifa is the canonical example. Parents picking up kids late were motivated by guilt — a moral/social cost. When the center added a fine (a market cost), late pickups doubled. The moral framing was replaced by a market framing, and the market price was too low. When the fine was removed, the moral framing didn't come back. Once you've established that lateness has a price, it's hard to re-establish that lateness is _wrong_. --- The same dynamic applies to tokens and communities. Offer your neighbor $25 to help you move. What was a social interaction — favor, reciprocity, community bonds — becomes a market interaction. And $25 is an insulting price for a market interaction. You've damaged both frames at once. Now scale this up. A DAO launches a token. Contributors who were participating for reputation and shared purpose now have a financial stake. The conversation shifts from "what should we build?" to "what will make the token go up?" Members who were in it for the mission start calculating ROI. New members join for the returns, diluting the culture. The community becomes a market wearing community clothing. --- Dee Hock, founder of Visa and a deeply thoughtful person about organizational design, spent years warning about this. Organizations, he argued, are held together by shared purpose and principles. The moment financial logic becomes primary, the organization hollows out — it retains its structure but loses its soul. Michael Sandel's work on the moral limits of markets makes the same point from a philosophical angle: there are things that markets corrupt by their nature. When you put a price on everything, you change the character of everything. --- None of this means tokens are inherently destructive. It means communities need to be extraordinarily intentional about how they introduce them. A few principles: **Separate the financial and social layers.** Not every community interaction needs to be tokenized. Keep spaces where participation is motivated by reputation and values, not financial returns. **Introduce tokens slowly.** The daycare study shows that once you introduce market framing, you can't easily remove it. Start with non-financial coordination and add tokens only where they solve a specific problem that can't be solved otherwise. **Design for the community you want, not the one token incentives will create.** Token incentives will attract profit-maximizers. If that's not who you want in your community, don't make token appreciation the primary value proposition. **Acknowledge the tension.** The worst outcome is pretending tokens and community values naturally align. They often don't, and the communities that navigate this successfully will be the ones that face the tension honestly. --- The existential question for crypto isn't whether the technology works. It's whether communities can introduce financial coordination tools without destroying the non-financial bonds that make communities worth having. The answer isn't obvious, and anyone who tells you it is hasn't thought about it hard enough. ## Data composability: what it is and why it matters URL: https://dazuck.com/writing/data-composability Date: 2021-03-03 Type: essay Summary: How composable data infrastructure enables permissionless innovation and why it matters for the next generation of the web. If you've spent time in DeFi, you've heard the term "money legos" — the idea that financial protocols can snap together like building blocks, each one composable with every other. A lending protocol plugs into a DEX which plugs into a yield aggregator, and nobody had to ask permission or sign a partnership agreement. That composability is the reason DeFi moved so fast. But money is the easy case. The harder and more important question is: can we do the same thing with data? ## What is data composability? Data composability is the ability for data created in one application to be used by another — without permission, without integration work, and without the original creator needing to anticipate the use case. Today, your social graph lives on Facebook. Your professional history lives on LinkedIn. Your reading highlights live on Kindle. Your health data lives on Apple Health or Fitbit. None of these talk to each other, and none of them let a third party build on the data you've created — even if you want them to. Composable data changes this. Instead of data living inside applications, it lives on shared networks. Applications become views on data rather than owners of it. Any developer can build a new experience on top of existing data, the same way any DeFi developer can build on top of existing liquidity. Think of it as the difference between Lego blocks and a finished toy. A finished toy does one thing. Lego blocks can become anything — and every new block makes every other block more useful. ## Why it matters The core unlock is **permissionless innovation**. In the current web, building a new social app means starting from zero. No users, no social graph, no content. You're competing against incumbents who have decades of accumulated data behind their moats. This is why we keep getting the same five apps with minor variations — the barrier to genuine innovation is too high. With composable data, a new application starts with access to all existing data on the network. A new social app can read your existing social graph. A new publishing tool can access your existing content. A new recommendation engine can draw on your existing preferences. Builders compete on the quality of their experience, not the size of their data moat. This creates a fundamentally different kind of network effect. In the current model, network effects benefit individual platforms — more users on Facebook makes Facebook more valuable, but it doesn't help anyone else. With composable data, network effects benefit the entire ecosystem. More data on the shared network makes every application built on that network more valuable. ## How we get there Three pieces of infrastructure need to come together: **Open data models.** Applications need to write data in shared formats that other applications can understand. Not proprietary schemas locked inside company databases, but community-defined standards that anyone can read and write. This is what standards bodies have tried to do for decades — the difference now is that blockchain-based coordination mechanisms can align incentives around adoption. **Shared data networks.** The data itself needs to live on networks that no single entity controls. Content-addressable storage (like IPFS) ensures that data can't be censored or removed unilaterally. Mutable data streams (like those on Ceramic) let users update their data over time while maintaining a verifiable history. **User-controlled access.** Composability without user control is just a different flavor of surveillance. Users need to decide which applications can access which data. Cryptographic access control — not platform terms of service — should mediate who sees what. ## Beyond financial applications DeFi proved that composability works for financial primitives. The stakes for data composability are arguably higher. Financial composability rearranges how money flows. Data composability rearranges how information flows — and information is the substrate of everything else: social coordination, knowledge production, creative expression, governance. The web was originally composable. HTML pages linked to each other. Anyone could view source and build on what they saw. The platform era broke that composability by moving everything behind APIs and login walls. Data composability is how we get it back — not by returning to the 1990s web, but by building shared infrastructure that preserves the openness of the original web while supporting the rich applications we've come to expect. The building blocks are being laid right now. The question isn't whether data composability is possible — it's whether we'll build it before the current model becomes too entrenched to displace. ## Identity as Information URL: https://dazuck.com/writing/identity-as-information Date: 2021-02-01 Type: essay Summary: Reframing digital identity not as a credential to verify but as information to manage, share, and compose. We've been thinking about digital identity wrong. The default framing treats identity as a verification problem: _prove who you are_. Show your ID. Enter your password. Confirm your biometrics. The entire field — from enterprise IAM to decentralized identity — has organized around this question, and it's led to a narrow set of solutions: better credentials, better authentication, better proofs. But verification is just one operation you perform on identity. And by centering the field on it, we've built systems that are good at proving claims and bad at everything else: managing information, enabling discovery, supporting composability, and evolving over time. The reframe: **identity is information**. Specifically, identity is the sum of information associated with an entity. Your name, your credentials, your social connections, your content, your transaction history, your preferences, your reputation — all of it is identity. Verification is one thing you do with that information. It's not the thing itself. ## Why the framing matters This isn't just semantics. The framing you start with determines the systems you build. If identity is a verification problem, you build credential systems. Issuers, holders, verifiers. Attestation formats. Revocation registries. The architecture centers on a specific interaction: someone needs to prove something to someone else. If identity is an information problem, you build data infrastructure. Streams of information that users create, control, and share. Applications that read from and write to shared data networks. The architecture centers on a different question: how does information about an entity flow through the world? The verification framing gives you digital passports. The information framing gives you a composable data layer for everything identity-related — of which credentials are just one application. ## The current web fragments identity information Right now, the information that constitutes your digital identity is scattered across hundreds of databases you don't control. Your professional identity lives on LinkedIn. Your social identity lives on Instagram and Twitter. Your consumer identity lives on Amazon and Google. Your financial identity lives at your bank. Your health identity lives at your hospital. Each of these platforms holds a slice of the information that, taken together, describes who you are. None of them share it with each other (or with you) in any useful way. And you have no unified view of — let alone control over — the information that is, in aggregate, your identity. This fragmentation isn't just inconvenient. It's structurally disempowering. When your identity information is distributed across silos you don't control, you can't compose it. You can't bring your reputation from one context to another. You can't give a new application access to your existing data without the old platform's permission. Every new service starts from zero because it can't access the information that already exists about you. ## Identity as composable information The alternative: identity information lives in user-controlled data streams on shared networks. Applications read from and write to these streams with the user's permission. No single application owns the data. Any application can access it (with user consent). In this model, identity isn't something you prove — it's something that accumulates and flows. Every interaction produces identity information. That information lives in streams you control. New applications can compose on existing information rather than starting cold. Your social connections are a data stream. Your credentials are a data stream. Your content is a data stream. Your transaction history is a data stream. Each one is independently created, independently verifiable, and independently shareable. Together, they compose into the rich, multifaceted thing we call identity. This is what we're building with Ceramic: a network of user-controlled data streams that any application can read from and write to. It's not an "identity system." It's data infrastructure that naturally supports identity because identity is, at its core, just information. ## What this enables When you treat identity as composable information rather than as credentials to verify, several things become possible: **Cross-platform reputation.** Your reputation in one context can inform trust decisions in another — not because some authority certified you, but because the underlying information is shared and verifiable. **Application portability.** Switching from one social app to another doesn't mean losing your social graph and content. The information is yours and travels with you. **Progressive identity.** Instead of proving everything upfront, you build identity over time through accumulated interactions. A new user isn't unknown — they have data streams from other contexts that establish who they are. **Privacy-preserving sharing.** Because you control the information and grant access selectively, you can share relevant identity data without exposing everything. Prove you're over 18 without revealing your birthdate. Prove you have a degree without revealing your GPA. ## The link to data composability This is the identity-specific instance of a broader thesis: the next generation of the web will be built on composable data rather than siloed data. Identity is arguably the most important domain for this shift because identity information is the connective tissue between everything else. Composable financial data gives you DeFi. Composable social data gives you portable social networks. Composable identity data gives you a unified, user-controlled foundation that makes all other composability possible. You can't compose data across applications if you can't establish who the data belongs to and who has permission to access it. Identity isn't the whole picture. But it's the piece that makes the rest of the picture possible. And getting it right means starting from the right framing: not "how do we verify people?" but "how does information about people flow through the world?" ## Web3: a route to harness the web's Natural Monopolies URL: https://dazuck.com/writing/web3-natural-monopolies Date: 2021-01-14 Type: essay Summary: Framing big tech's power to de-platform through the lens of natural monopolies, and how Web3 data interoperability can restore competition. In January 2021, Twitter and Facebook banned a sitting president. Apple and Google removed Parler from their app stores. Amazon Web Services kicked Parler off its hosting infrastructure. In the space of a few days, a handful of companies demonstrated their ability to remove anyone — even the most powerful person in the world — from the digital public sphere. The reaction split into predictable camps. "They're private companies, they can do what they want." "They're the public square, they shouldn't censor speech." "They're monopolies, they should be regulated." "They're communities, they should moderate content." All of these framings capture something real. None of them capture the core problem. ## The natural monopoly framing The better frame: big tech platforms are **natural monopolies**. A natural monopoly exists when the structure of a market makes competition impossible — not just difficult, but structurally infeasible. Network effects, high fixed costs, and near-zero marginal costs create markets where one provider naturally dominates and competitors can't viably enter. Utilities are the classic example. It doesn't make sense to have five competing power grids running to every house. The infrastructure cost is too high and the network is too important. So we end up with one provider, and then we need regulation to prevent abuse. Big tech platforms have the same structural properties. Social networks have extreme network effects — a social network with your friends on it is infinitely more valuable than one without them. Cloud infrastructure has enormous fixed costs. App stores and search engines have winner-take-all dynamics. These aren't monopolies because the companies are uniquely evil or because regulators failed. They're monopolies because the market structure makes monopoly the natural outcome. ## The three roles The tech giants operate as three things simultaneously: **Public square.** Twitter and Facebook are where public discourse happens. Not exclusively, but substantially. When you're removed from these platforms, you lose access to the primary venue for public communication. This is the speech concern. **Public utilities.** Google Search, Apple's App Store, and Google Play are essential services that nearly everyone depends on. Try running a mobile business without access to either app store. Try reaching customers without appearing in Google search results. These are utilities in everything but legal classification. **Public infrastructure.** AWS, Google Cloud, and Azure host a massive share of the internet itself. When AWS removes a customer, that customer doesn't just lose a vendor — they may lose the ability to operate online entirely. This is infrastructure-level power. Taken together, these companies are effectively the government of our digital society. They set the rules for speech, commerce, and access. They enforce those rules unilaterally. And unlike actual governments, they have no democratic accountability, no due process, and no obligation to treat participants fairly. It is, structurally, a dictatorship. Usually a benign one. But dictatorships are defined by the structure of power, not the quality of the dictator. ## The standard responses and their limits **Leave them private and hope for the best.** This is the default, and it amounts to hoping that companies with monopoly power will voluntarily exercise it responsibly forever. History gives us no reason to believe this. **Government regulation.** Regulate tech platforms like utilities, with rules about who can be excluded and under what circumstances. The concern: regulated monopolies tend to become captured by the industries they regulate. Think Comcast — technically regulated, practically unchecked. Regulation that entrenches incumbents while providing minimal accountability is worse than no regulation. **Antitrust and breakup.** Break up the big platforms into smaller companies. This treats the symptom (current companies are too powerful) without addressing the cause (the market structure naturally produces monopolies). Break up Facebook, and in ten years one of the fragments or a new entrant will re-consolidate. The network effects don't go away. **Forced interoperability.** Require platforms to interoperate — like telephone carriers are required to connect calls between networks. This is more promising but extremely hard to implement. What does interoperability mean for a social network? For a search engine? For a cloud provider? The complexity is enormous, and the details determine whether it actually works or becomes another box-checking exercise. ## The Web3 approach: separate data from applications Here's the framing I find most productive: the core problem is that data and applications are fused. Your social graph lives inside Facebook. Your content lives inside Twitter. Your apps live inside Apple's App Store. Your infrastructure lives inside AWS. The data and the application are the same thing, which means switching applications means losing data, which means you can't switch, which means monopoly. Web3 offers a structural solution: **separate the data layer from the application layer.** Applications remain proprietary — companies build them, compete on user experience, and make money however they can. But the underlying data — social graphs, user accounts, content, transaction histories — lives on shared, interoperable networks that no single company controls. In this model, you can use Twitter or a competitor, and see the same messages. Your social connections aren't locked inside any platform. Your content follows you across applications. The application is a _view_ on shared data rather than a _container_ for exclusive data. This gives you something that sounds contradictory: **more diversity AND more network effects simultaneously.** More diversity because the barrier to entry for new applications drops dramatically — you don't need to bootstrap a new social graph from zero. More network effects because all applications contribute to and benefit from the same shared data network. ## How this works technically The building blocks exist: **Decentralized identifiers (DIDs)** let users have portable accounts not controlled by any platform. You don't "sign up" for each new service — you connect your existing identity. **Content-addressable storage (IPFS, Arweave)** lets content be stored on decentralized networks. Your posts don't live on Twitter's servers — they live on a shared network that any application can read. **Mutable data streams (Ceramic)** let users maintain dynamic data — social connections, preferences, profiles — on user-controlled infrastructure that any application can access with permission. **Blockchain-based governance** lets the rules of the shared data network be set by participants rather than by a single company. Together, these create a shared data layer where applications are interfaces on top of shared information rather than walled gardens with exclusive access to user data. ## The competitive web again The web was originally competitive. Anyone could build a website. Anyone could link to anyone else. Search engines indexed the open web and directed traffic based on relevance. Competition was real because switching costs were low — if one website was better than another, you just went there instead. Platforms broke this by making the web's underlying data proprietary. Your Facebook social graph, your Twitter content, your Amazon purchase history — this data became the moat that prevented competition. Web3 data interoperability restores competition by removing the moat. Applications compete on experience, features, and trust — not on data lock-in. The web becomes a marketplace of interfaces on shared infrastructure, rather than a collection of digital fiefdoms. This doesn't solve every problem. Content moderation still needs to happen. Bad actors still need to be addressed. The transition from the current architecture to an interoperable one is a massive coordination challenge. But it addresses the root cause — the fusion of data and applications that creates natural monopolies — rather than applying band-aids to symptoms. The natural monopoly problem isn't going away. Network effects aren't going away. But the question of whether those network effects benefit a single company or an entire ecosystem — that's an architectural choice. And Web3 is the most promising architecture for choosing the latter. ## Demystifying Digital Identity URL: https://dazuck.com/writing/demystifying-digital-identity Date: 2020-06-19 Type: essay Summary: Breaking down the requirements, components, and options for digital and decentralized identity systems. Digital identity is one of those topics where everyone agrees it matters and almost no one agrees on what it means. The phrase gets used to describe everything from a government-issued credential to a social media profile to a cryptographic key pair. This ambiguity isn't just semantic — it leads to solutions that solve the wrong problem. This piece is an attempt to cut through the confusion. What is digital identity, really? What are its components? And what are the actual options for how it should work? ## Part 1: What digital identity is ### The basics Digital identity is the collection of information about a person (or organization, or thing) that exists in digital form. That's it. It's not a card or a credential or a login — it's the sum total of what's known about you in digital systems. This includes the obvious: your name, email, profile photos. But it also includes the less obvious: your transaction history, your social connections, your browsing patterns, your content, your reputation. Every interaction you have online produces identity information, whether you think of it that way or not. ### Why it matters now We've always had identity — the question is who manages it. In the physical world, identity is managed through a patchwork of institutions (governments, employers, banks) and social relationships (your neighbors know who you are). It's messy but decentralized by default. No single entity holds all the information about you. Online, the opposite happened. Your digital identity is fragmented across dozens of platforms, each holding a slice of who you are. You are a different person on every service — different username, different profile, different history. And each platform owns their slice entirely. You can't take your reputation from eBay to Airbnb. You can't move your social graph from Twitter to Mastodon. You can't bring your purchase history from Amazon to a new retailer. This fragmentation creates three problems: **inconvenience** (managing dozens of accounts), **lock-in** (your identity is hostage to platforms you may not trust), and **lack of agency** (you have no unified view of or control over the information that constitutes your digital self). ### The key components Any identity system, digital or otherwise, needs four things: **Identifiers** — a way to refer to the entity. On the current web, these are platform-specific: your Twitter handle, your email address, your phone number. In decentralized systems, these are DIDs (Decentralized Identifiers) — globally unique identifiers that you control, anchored to a blockchain or other verifiable network rather than to a company. **Authentication** — a way to prove you control the identifier. Passwords, biometrics, cryptographic keys. The mechanism matters because it determines who can impersonate you and how easily. **Claims and credentials** — assertions about the entity. Your university issued you a degree. Your government issued you a passport. Your employer issued you a badge. These are claims made by third parties about you. Verifiable credentials are a standard for making such claims cryptographically verifiable without calling the issuer. **Data and profiles** — the accumulated information associated with the entity. Your posts, preferences, transaction history, social connections. This is the richest and most valuable layer of identity, and it's the layer most thoroughly captured by platforms today. ## Part 2: The landscape of approaches ### Centralized identity This is what most of us use today. Each service maintains its own user database. You create an account, the platform stores your information, and you authenticate with a username and password. **Pros:** Simple for users, simple for developers, well-understood. **Cons:** Fragmented (different identity per service), platform-controlled (they own your data), honeypot risk (breaches expose millions of records at once), no portability. ### Federated identity "Sign in with Google" or "Sign in with Facebook." A few large identity providers serve as intermediaries, letting you use one identity across many services. **Pros:** Less fragmentation, more convenient, fewer passwords. **Cons:** Concentrates power in a few identity providers, creates surveillance chokepoints (Google knows everywhere you sign in), still platform-controlled, and the identity provider can revoke your access. ### Self-sovereign / decentralized identity The user controls their own identifiers and data. No platform or provider can unilaterally revoke access or control information. Identity information is stored in user-controlled spaces and shared selectively with applications. **Pros:** User control, portability, no single point of failure, privacy by design. **Cons:** Harder to use (key management is still unsolved for mainstream users), less mature infrastructure, cold-start problem (who issues the first credential?). ## The future: composable identity The right answer probably isn't any single system. It's interoperability between systems — composable identity. Imagine: your government issues you a verifiable credential (you're over 18). Your university issues another (you have a degree in computer science). Your employer issues another (you work at Company X). Your social graph is portable across platforms. Your content follows you from app to app. Each piece is issued and verified independently, stored under your control, and shared selectively with the services you choose. No single system "owns" your identity. Instead, your identity is composed from independent, interoperable pieces — like data composability applied specifically to the information that makes you _you_. This is what we've been building toward with Ceramic and 3Box: not an identity system, but a data infrastructure layer that naturally supports composable identity. When users control their data streams and applications read from shared networks, identity becomes emergent rather than managed. The hard problems are real — key management, recovery, mainstream UX, bootstrapping adoption. But the direction is clear: identity should be controlled by the person it describes, composed from independent sources, and portable across the services that use it. ## The Short and Long of It: social coordination in times of crisis URL: https://dazuck.com/writing/the-short-and-long-of-it Date: 2020-03-31 Type: essay Summary: How COVID-19 exposes the false tradeoff between coordination and liberty, and why decentralized infrastructure is the long-term solution. We're a few weeks into what's clearly going to be the defining crisis of our generation, and the arguments about how to respond are already calcifying into familiar camps. Libertarians warn about government overreach. Authoritarians point to China's effective (if brutal) containment. Tech optimists pitch surveillance-based contact tracing. Civil libertarians sound alarms about emergency powers that never get rolled back. Everyone is arguing about the tradeoff between effective coordination and individual liberty. Almost no one is questioning whether the tradeoff is real. ## The tradeoff as we understand it Here's the conventional framing: responding effectively to a crisis like COVID-19 requires massive coordination — testing, contact tracing, resource allocation, movement restriction. This coordination requires centralized power. Someone needs to make decisions, enforce compliance, and aggregate information. The more power you give to central authorities, the more effective the response. The less power you give them, the more freedom people retain but the less coordinated the response. This is the eternal tension of governance, and crises sharpen it to a point. In normal times, we can tolerate some inefficiency in exchange for liberty. In a pandemic, inefficiency kills people. So we make pragmatic compromises. We accept surveillance in exchange for safety. We grant emergency powers. We hope those powers get rolled back when the crisis ends. (They usually don't.) ## Five things that are true Before arguing for a different framing, let me lay out what I think is actually going on: **1. Global interdependence isn't going away.** The fantasy that nations or communities can wall themselves off from global problems is just that — a fantasy. Pandemics, climate change, economic contagion, information warfare — these problems are structurally global. Any solution that requires isolation is not a solution. **2. Interdependence requires massive coordination.** You can't respond to a pandemic without coordinating testing, treatment, resource allocation, and behavior change across millions or billions of people. This is a hard coordination problem by any measure. **3. Centralized infrastructure enables abuse.** This isn't a prediction — it's a historical observation. Every centralized surveillance and control apparatus, no matter how well-intentioned at creation, eventually gets used in ways its creators didn't intend. The NSA programs revealed by Snowden started as counterterrorism tools. China's social credit system started as a financial trust mechanism. Emergency powers granted in crises persist long after the crisis ends. **4. Institutions' incentives diverge from participants' over time.** The institution managing the coordination infrastructure starts aligned with the people it serves. Over time, it develops its own interests: self-preservation, power accumulation, budget expansion. This isn't corruption — it's structural. Organizations optimize for their own survival, and that optimization eventually conflicts with their stated mission. **5. This dynamic harms everyone.** Not just the people being surveilled or controlled — everyone. When trust in institutions erodes (because the erosion of trust is rational, given the pattern), it undermines the capacity for future coordination. The NSA revelations didn't just violate privacy — they made it harder to build legitimate public health surveillance infrastructure because people (correctly) don't trust the government not to repurpose it. ## The false tradeoff Here's the thing: the tradeoff between coordination and liberty is only inevitable if the infrastructure for coordination is centralized. We're forced to choose between (a) voluntary coordination that's ineffective because it can't compel participation and (b) centralized coordination that's effective but abusable because it concentrates power. This is a false dilemma created by the architecture of our infrastructure, not by the nature of coordination itself. ## Decentralized infrastructure as the long-term answer Imagine a different architecture. During a pandemic, you share your health data — test results, symptoms, location history — with a coordination system. But the data lives on infrastructure you control. You can see exactly who accesses it. You can revoke access when the crisis ends. No central authority can repurpose the data because no central authority holds it. The coordination happens, but the infrastructure doesn't create a permanent surveillance apparatus. This isn't science fiction. The building blocks exist today: - **Tor and onion routing** enable anonymous communication. - **WebRTC** enables peer-to-peer data exchange without central servers. - **Blockchain networks** enable decentralized record-keeping and governance. - **IPFS and content-addressable storage** enable decentralized data hosting. - **Cryptographic access control** enables selective, revocable sharing. What doesn't exist yet is the integration of these technologies into usable, scalable systems that regular people and institutions can use. That's the work — not inventing new primitives, but assembling existing ones into infrastructure that makes the tradeoff between coordination and liberty genuinely unnecessary. ## The short and long of it In the short term, we don't have this infrastructure, and we need to coordinate now. That means pragmatic compromises with centralized institutions. Use the contact tracing apps. Support emergency measures. Accept that imperfect coordination is better than no coordination. But — and this is the crucial "but" — don't mistake the short-term compromise for the long-term answer. Don't accept that effective coordination requires permanent centralized surveillance. Don't let emergency measures become permanent fixtures. And most importantly, invest in building the decentralized infrastructure that makes the tradeoff obsolete. The crisis will end. The infrastructure we build — or fail to build — during and after it will determine whether the next crisis forces the same terrible choice. We can build coordination systems that don't require choosing between effectiveness and liberty. We just haven't done it yet. The short-term response requires pragmatism. The long-term response requires building something fundamentally different. Both are necessary, and confusing one for the other is how we end up permanently sacrificing liberty for temporary safety. ## 'How Is She?' You ask, About My Mom After Her Stroke URL: https://dazuck.com/writing/how-is-she Date: 2019-05-15 Type: personal Summary: On the question everyone asks after someone has a stroke, and why the answer is more complicated than they expect. "How is she?" That is, inevitably, the first question people ask me when I tell them that 9 years ago my mom suffered a severe stroke. "How is she?" It's a simple question, and I've had a lot of practice responding, but I haven't found an answer that works. "She's good," I start sometimes, truthfully, and I can hear their relief. Perhaps that's appropriate. She is after all in stable condition and good spirits. But it's a very incomplete answer and I am then forced to surprise them. "She still has significant disabilities, with very limited speech and no use of her right arm." They are left on unstable footing, not knowing whether to offer condolences or appreciation. Other times I start with the facts of her ongoing disabilities, her difficulty with speech, and the acknowledgment that at this point she is unlikely to ever make a full physical recovery. When I start this way I often have to rush to cut into the heartfelt, well-meaning, but somehow misplaced sympathies they begin to express. "No no, really we're okay," I say and mean. "She is doing well, frustrated of course but doing well and we've all adjusted. For us, it's the 'new normal.'" Of course, this leaves them unsure as well. Am I just putting a nice face on a bad situation to move past the uncomfortable conversation? I'm not. "Really," I say. "She's doing great. She just wrote a book, in fact." There seems to be a binary to "how is she doing" that, 9 years after my mom's stroke, doesn't exist in my mind. When most people think about rebounding after a medical trauma like stroke, they are thinking of physical recovery. "How is she doing" translates to "has she recovered her physical abilities?" That's not how we think of recovery at all. Physically she is significantly disabled, but emotionally and mentally she may be stronger than ever. My mom's aphasia prevents her from expressing herself fully and doing many of the things she loves. But her journey and limitations have helped her take stock of what she values and create a built-for-purpose life of meaning that few of us have the clarity of mind to seek. As a family, we have come to believe that recovery is not about regaining exactly what was lost, but about bouncing back from a setback to rebuild a life of meaning — perhaps even of more meaning and richness than before. Stroke survivors, like many others with disabilities or visible signs of trauma, or even simply with distinctive or minority physical appearances, are often treated with a "master identity" — a single characteristic by which others judge who they are. This tendency overwhelms everything we know about identity: we are not one thing, we are many. None of us can be reduced to a single defining characteristic or even a perfectly coherent 'true self.' We are an ever-changing mix of values and characteristics and experiences. Most stroke awareness campaigns will focus on the prevalence of stroke, how to recognize it quickly, the ability to treat it effectively, and organizations helping survivors recover. All of this is critical to limit the damage stroke does to us. We should also remind ourselves to be more thoughtful and aware of how we think of stroke and trauma survivors. Their medical setbacks and challenges are an important part of who they are and how they are doing, but not the whole picture. My mom is a stroke survivor, and she is many other things too. She has not been reduced to a disabled portion of her old self, and her recovery is not incomplete because she has not regained all of her physical abilities. She has been transformed, and yes in some physical ways diminished. But in many other ways, she has grown. She is a more attentive friend, an even more caring mother, a teacher in a new way to a new audience, and much of the time a happier person. She is certainly a different person than she was 9 years ago — like all of us as we age, adapt, and live through an unscripted life. How is she doing? It's a mix – some good, some bad. Some months are great, others not so good. Some days are fantastic, others incredibly frustrating. Some hours are spectacular, while others are awful. Same as you or me. ## Identity Theft: Rediscovering Ourselves After Stroke URL: https://dazuck.com/writing/identity-theft Date: 2019-05-01 Type: book Summary: Co-authored with Debra E. Meyerson, on the work of rebuilding identity after a severe stroke. _Identity Theft: Rediscovering Ourselves After Stroke_ is a book I co-authored with Debra E. Meyerson, published May 2019 by Andrews McMeel. It explores the journey of navigating the profound losses and rebuilding identity after a severe stroke. ## Org design around goals, not ideals URL: https://dazuck.com/writing/org-design-around-goals Date: 2018-12-05 Type: essay Summary: Why structuring teams around concrete goals rather than abstract principles produces better organizations. At many decentralized organizations, we're not only building new technology but also trying to use more decentralized organizational designs. These more autonomous models are in line with our vision of future organizations and networks, but they are not easy — especially when the tech that will support them isn't yet mature. These experiments involve a lot of learning and iteration. We spend a lot of time talking about self-organization, governance, decentralization, decision making models, leadership, and other organizational concepts. ### Sometimes we don't frame these conversations deeply enough in the context of what we are actually trying to achieve: successful organizations. Great organizations are made by high-performing people who collaborate effectively together. Our org design should not start with a glossary of terms like 'flat' or a menu of options for how to make decisions. We should start with an understanding of the trade-offs inherent in org design, and make clear choices between those based on the goals and context of our organization. There are two factors that impact how effective a group of people is in pursuit of a mission: 1. How effective each individual is in working towards the mission 2. How effective the individuals are in working together Those are the two levers we have to improve the performance of a team. That's it. This doesn't make it easier, but it can make things a bit simpler before we dive into the jargon-filled world of organizational design. (Arriving at the 'right' mission is a related but whole other ballgame that's out of scope for this). ## Lever 1: Effective individuals People are effective in their roles when they are (1) motivated to work hard, (2) effective doing their work, and (3) doing work aligned with that mission. Each of these have their own considerations. ### Motivation Daniel Pink, in his book Drive and now-famous TED-talk, says that people are motivated when they have Autonomy, Mastery and Purpose. Autonomy doesn't require complete freedom, but it does mean no micro-managing. Purpose comes from pursuit of a goal that intrinsically matters. Mastery means building skills and desired expertise through work. ### Effectiveness The organizational model we choose has a direct impact on the effectiveness of individuals in a few ways. - Attracting talent, whether that is entrepreneurs who crave independence or specialists who want clear roles and mentorship - Employee growth, so employees have room to stretch and effective coaching/feedback to improve. Attract entrepreneurs with promises of independence, or train talent with experienced managers and clear promotion pathways? - Role matching that effectively align people effectively with their expertise, skill level and interests (achieving both personal flow and efficient resourcing) ### Direction How are goals and projects determined to make sure that effective work is actually pushing towards the overall mission? How best to promote: - Information flow: are there clear but rigid patterns (a hierarchy makes it clear with only 1 pathway between any 2 nodes) or a fluid but hard-to-track network of connections? - Collaboration: do clusters make it easier to find and work with complementary people and skill-sets, or to share and learn with others in the same domain? - Decision making: One measure of decision making effectiveness is Quality x Speed x Yield — Effort. What trade-offs make sense? Perhaps the central question of traditional management theory has been how much to centralize or decentralize decision making authority. Should edges with great expertise over their domain be freed to move fast, or should info be filtered to a center where a broader considerations and alignment can be condensed? ## Lever 2: Effective working together The structures we design for our organization impact how and how well we collectively achieve: - The best strategies and major decisions - Alignment on goals to realize those strategies - Effective resourcing (division of roles & work) against those goals - Quality output - Accountability to results and each other to achieve those outputs - Shared learning & adaptation over time Organizations are not single entities but rather a collections of individuals. How those individuals behave and what they produce will depend on how they organize. One of my favorite examples of the inevitable impact of org design is in the phones we use every day. ### Messaging at Apple vs. Google At one point Google had 7 different messaging apps on Android! They were frequently criticized for this — how could Google let it happen? This would never happen at Apple. Apple prioritizes integrated customer experiences. That is their strategy and competitive advantage, and so the organization is crafted for that. Everything is orchestrated around end-to-end products. Decisions about what apps are built and make it onto the iPhone or other platforms ultimately roll up to the leaders of those product experiences. This lets them ensure a great and simple user experience, with no room for diversion. It is unfathomable Apple would ever end up with multiple overlapping messaging apps. Google is organized differently because it's strategy has been different. They won with engineering leaps and experimentation, and a wide base of products across the web and other platforms. They are organized around this. As messaging became an important battleground, of course many different teams at Google would see an opportunity and build a product aimed at that market. Google is not one entity, its a collection of people and teams. Hangouts, Gmail, Messages etc. all came out off different teams. Some started on desktop, some mobile, some other. They expanded and converged. They did not roll up to a single owner or product group to manage away the overlap. Google is not built that way. This hurts coherence, but allows experiments, like the wildly successful Gmail, to emerge. Different strategy, different trade-offs, different org, different outcomes. ## Choosing the right org structure Various models for organizational design with different trade-offs. Concepts like 'self-managed' or 'flat' are not good themselves, they are designs and each has their own merits and priorities. ### Hierarchy Hierarchy is an organizing tool that gives relatively clear blueprints: do the work in your scope, share and elevate whatever seems important, trust your seniors and subordinates to do the same. It's not necessarily efficient, but there are clear routes for information and decision flows which dramatically simplifies things. There is only 1 route between any two boxes in the org. Hierarchy does not need to be command and control. Many good (servant) leaders try to push as much as possible to the edges. But there are defined pathways to elevate major or unexpected situations. ### Matrix Intel CEO and management guru Andy Grove has said that every organization of a certain scale inevitably adopts a matrix, with people belonging to both vertical (e.g., Line of Business) and horizontal (e.g., function a la Finance) organizations at once. This adds more lines of communication, flexibility to shift resources and expertise, and dynamic sharing. It also adds complexity. Org design often becomes about how much power (decisions, budget) rests in the vertical vs. functional organizations and their leaders. ### Nested circle In some ways, forms of nested circles (like holocracy) have a lot in common with the matrix organization, but with more than 2 dimensions. They are better in fast-changing, dynamic environments as they flex to many forms of expertise, organizational groupings, and overlaps. But the same trade-offs also scale, with more ambiguity about where responsibilities ultimately lie. ### Fully decentralized network Removing the organizational units altogether in favor of a web of individuals or roles maximizes flexibility and autonomy. But achieving shared information, agreement, and clarity will naturally be far harder. Because the structure is so dynamic, it is impossible (past a small scale) for individuals to keep a mental model of who works on, knows about, has responsibility over what. So information flow, advice gathering, and decision making all must lost clarity because they can flow in so many ways rather than shared and anticipated ones. This may be worth it for the other factors, but (without much better technology) is an innate trade-off to these org designs. ## Design for goals not ideals Rather than starting with a menu of org design options and a desire to 'be flat/decentralized/agile,' it makes more sense to start by weighing trade-offs. Which levers matter most for your organization: motivation, individual effectiveness, aligned direction, good strategy, efficient resourcing, accountability, shared learning? With that rubric in place, org design decisions can optimize for the right things. The right model depends on what the company is optimizing for. Is the path to winning integrated experiences or innovative leaps? How important is speed in decisions versus broad alignment? How critical is it to attract proactive entrepreneurs, or how much of your workforce will need to progress with steady training? 'Flat' might mean the motivational autonomy and speed with no management overhead. But it also might mean less room for coaching, nobody responsible for giving honest feedback and supporting employees' growth. On our first retreat for 3Box we went through an exercise as a team, weighing various trade-offs and talking through the toughest ones so we could figure out what we valued, how we should prioritize and compromise, and so how we should design our team. Over time, we'll return to this and see if our situation and values have changed, and adjust accordingly. Starting with a clear focus on the goals of our org design, the two levers that impact productivity — individual effectiveness and teamwork — and what we value will make that adaptation far clearer to all and easier to achieve as the needs of our organization evolve. 3Box is building a more human web3, and we're working hard to create the best team possible. If that sounds exciting, reach out or get involved. - Join our community discord - Check out our open jobs ## Remembering the 'human factor' in tech networks URL: https://dazuck.com/writing/remembering-the-human-factor-in-tech-networks Date: 2018-10-16 Type: essay Summary: Why designing tech networks around actual human behavior — not rational-agent models — leads to better systems. There's a foundational assumption in most tech network design: people are rational agents who respond predictably to incentives. Set the right rewards, apply the right penalties, and behavior follows. Crypto and blockchain took this assumption and ran with it. Mechanism design, token economics, game theory — the entire toolkit assumes that if you get the incentives right, the system works. Humans are modeled as utility-maximizing nodes in a network, and the designer's job is to align their utility functions with the network's goals. This assumption is wrong, and building on it produces systems that are brittle, alienating, and often counterproductive. ## Three forces, not one Humans aren't motivated by a single force. They're motivated by (at least) three, operating in different contexts and often in tension with each other: **Utility maximization** — wanting the best deal. This is the force that economics models well. It's mediated by money and markets. When you comparison-shop for a flight or negotiate a salary, you're operating in this mode. **Reputation** — wanting good future opportunities. This is mediated by status, relationships, and contracts. When you do good work to build your professional reputation, or fulfill a promise because people are watching, you're in this mode. **Morality** — wanting to do the right thing. This is mediated by values, culture, and community norms. When you return a lost wallet or help a stranger, you're in this mode. These three forces don't blend smoothly. They operate in different domains, and when you introduce one into a domain governed by another, things break. ## The daycare problem The most famous example comes from a study of daycare centers in Haifa, Israel. Parents were regularly picking up their children late, creating problems for staff. The centers introduced a fine for late pickups — a straightforward economic incentive. Late pickups doubled. Before the fine, picking up your kid late felt like a moral failure — you were imposing on the teachers, breaking a social norm, being a bad community member. The fine replaced that moral framing with a market framing. Now late pickup had a price, and parents could simply decide whether the price was worth it. For many, it was. The fine didn't add a cost to lateness — it replaced a higher cost (guilt, social disapproval) with a lower one (a few dollars). When the centers removed the fine, late pickups didn't return to the original level. The moral norm had been destroyed and couldn't be easily rebuilt. The market framing, once introduced, stuck. This isn't an isolated curiosity. It's a pattern: **monetary incentives crowd out reputational and moral motivations**. Paying people to donate blood reduces donations. Offering your neighbor $25 to help you move reframes the interaction from a favor (reputation/morality) to a transaction (utility) — and $25 is a bad price for a transaction. ## The implications for network design Most crypto networks are designed almost entirely around utility maximization. Stake tokens, earn rewards, get slashed for misbehavior. The assumption is that if the token economics are right, the network will be healthy. But healthy networks — the ones people actually want to participate in — rely on all three forces. People contribute to open source for reputation and moral satisfaction, not just money. People moderate communities because they care, not because they're paid. People build things that don't have clear financial returns because they believe in what they're building. When you slap token incentives on top of these behaviors, you risk the daycare effect. The person who was moderating out of community pride becomes a paid moderator — and when the pay isn't worth it, they stop. The developer who was contributing for reputation starts optimizing for token rewards instead, and the quality of contribution changes. The community that was held together by shared values becomes a collection of economic agents optimizing for returns. ## Design for emergence, not control The alternative isn't to ignore incentives. It's to stop treating incentive design as the primary lever and start treating it as one lever among several — and often not the most important one. Networks should be designed to let behavior emerge rather than to prescribe it. Instead of trying to engineer the "right" behavior through mechanism design, build systems that: **Support communities, don't define them.** A network protocol should provide infrastructure for communities to form and govern themselves, not impose a single governance model. Different communities will have different norms, and that's fine — it's the same variation we see in the physical world. **Make space for non-financial motivation.** Not every contribution needs a token reward. Reputation systems, recognition, and social norms are powerful motivators that token incentives can destroy. Be very careful about which behaviors you financialize. **Embody values, don't just incentivize outcomes.** The best networks have a culture — a shared sense of what matters and how things should be done. This culture can't be engineered through token economics. It emerges from the people involved, the decisions the project makes, and the norms that develop over time. **Design for adaptation.** No set of incentives will be right forever. Build systems that can evolve as the community's needs and values change, rather than locking in a fixed mechanism at launch. ## Beyond rational agents The crypto space has enormous potential to build better coordination infrastructure. But realizing that potential requires a more honest model of human behavior — one that accounts for the full range of what motivates people, not just the slice that fits neatly into economic models. Markets are powerful. They're also narrow. They work well in specific contexts (allocating scarce resources, facilitating exchange) and poorly in others (building trust, fostering community, inspiring meaningful work). The art of network design is knowing which force to lean on in which context — and knowing that leaning on the wrong one can permanently damage the others. We don't need better mechanism design. We need more human design. ## The 3 parts of identity in a dynamic world URL: https://dazuck.com/writing/the-3-parts-of-identity Date: 2018-06-28 Type: company Summary: Breaking down digital identity into three components — identifiers, attestations, and data — and why each matters for user sovereignty. At 3Box we live and breathe identity, and while GDPR and data abuses like Facebook's may be concerning, they have also brought some welcome attention to the issues of digital privacy and data control. As we push ahead to a self-sovereign future, I wanted to share how I think of the 3 parts of identity in our increasingly digital, interwoven society. ### 1. Reducible identifier Identity as a concept is difficult — even scholars in the field can't agree on a definition. They do agree, though, that identity for each of us is not singular and static but rather plural and dynamic. Who we are changes over time and contexts. However, we as people and societies and institutions need something concrete to pick me out from the crowd of others, even others who may share my same name. In Seeing Like a State James Scott calls this 'legibility.' When we get past local communities and the Dunbar number, we need to make society understandable. We reduce things to lists and bits of information, and one of the must fundamental bits of information is who we are. We need a legible, concrete, reducible identity. This leads to passports, social security numbers, and other root identifiers from the physical world. Much of the work in the identity industry is around verifying and authenticating who somebody is in this regard, whether through knowledge questions, facial or bio metric scan, or proof that they know a password or hold a private key. ### 2. Contextual identification While reducible identification is critical to anyone trying to keep track of a group, it doesn't actually tell us much about the person who has been identified. Context about that reducible identity gives orientation to who the person (or otherwise) actually is. Traditionally this may include where somebody lives (driver's license), who their parents are, what tribe (including nationalities, teams, gangs, political affiliations, religion, etc) they are a part of, or who their employer is — anything you might fill out on a test or census form. More recently, this would include a person's social graph (Facebook friends, LI contacts) and location data. This may be static or dynamic information, but it's not ephemeral — it roots them in time and space. As we move to a more digital-first world, and hopefully soon a self-sovereign world, we talk often of capturing this contextual information as 'attestations' or 'verifiable claims' to our root (reducible) identity, giving us a way to bring all the context about us together in a powerful, private, verifiable (trusted) way. ### 3. Dynamic identity A reducible identifier and contextual identification provide a way to differentiate each other, to verify who somebody is and know some categorical details about them. But they do not tell you who somebody is. The richness of our identity is captured in the multitude of behaviors, preferences, relationships, interactions, histories and even thoughts. Only here can true identity — the plural, dynamic identity — be seen. One persons preferences and behaviors may contradict each other over time, and often do, so this cannot be captured in a reducible and 'verifiable' context. It is only captured when our goal is to understand who somebody is deeply, rather than pick them out from a crowd of others. In today's world, data reflecting who we are is being generated and captured constantly. With much of our lives online and in our phones, and sensors cheap enough to put on everything, nearly all our interactions with the external world can be turned into digital data. In fact, our internal 'data' — health, mood, biological patterns is now logged as well. This is infamously captured and monetized by Facebook and others, but it is also incredibly powerful. If put in control of users and creators of data, we can build a world that lets us use this rich data to truly understand the people around us far better than ever before, and as far more than bits of reducible and contextual information. In fact, with rich, full identity data, we could understand our dynamic selves far better than previous identity construct have allowed us. In Homo Deus Yuval Harari envisions a breakdown of the singular 'self' into a multitude of biological signals and patterns and interactions, changing the very notion of identity — perhaps an indication of why this concept proves so hard to nail down. ## Parties, Privacy & Blockchain URL: https://dazuck.com/writing/parties-privacy-blockchain Date: 2018-05-14 Type: company Summary: How to think about data privacy and the parties involved — using a 1982 case study to frame modern blockchain-based solutions. It's 1982. Alice picks up her landline phone and calls Bob, her friend and the local shopkeeper. In the course of conversation she shares that she's pregnant. _There are 5 entities involved in this exchange:_ A) Alice B) Bob C) The phone line & infrastructure itself D) The company that built the phone line infrastructure E) The company that operates the phone network, which automatically captures meta-data about phone calls to maintain service quality & employs operators who connect calls between two people. It's 1982. The internet, PCs and cell phones are around the corner, though nobody is planning for these. But we can think of this as a simple starting point to think about modern data privacy — uses, protections, and responsibilities around personal information. uPort lets users and organizations interact and share data directly, with no 3rd party ### What rights & responsibilities does each party hold? Here we'll establish some assumptions and questions specific to this situation, focusing on how we believe the parties _should_ behave. And in the next section we'll generalize that to see what we can learn for modern, more complex situations. **A) Alice,** the subject, can share information about herself freely, and has a responsibility to herself to be discrete about who she shares with **B) Bob,** the receiver of data, has a few roles and considerations: - As Alice's friend, he probably should not share her secret — but there's no law against gossip - As a local shopkeeper, _Bob probably should be allowed to use the provided info to suggest_ Alice buy a baby formula next time she visits - As a possessor of personal data, _Bob probably should not be able to sell this data_ to pharmaceutical companies who want to sell baby products **C) The technology used**, the phone and network infrastructure, is a tool; it transmits the sound according to some properties. 'Data' is created through phone use as phone lines are either free or busy, and so a person's activity _could_ be collected from the phone infrastructure. There's no normative claims on this, it's just a statement about the technology. **D) The infrastructure creators** probably should work to design phones, phone lines, and other infrastructure to minimize the amount of data that is created in public, and have a responsibility to make clear to users of the phone system what data others could collect from their usage. If this is done transparently and honestly, then the privacy-compromising creation of that data is reasonably up to the user. **E) The network operator** - _Should be allowed to collect data necessary_ to operate the network, collecting as little personal information as possible to do this - Unlike Bob, who was a party to the conversation, the network operator was a facilitator and _should not be able to eavesdrop_ on personal conversations and try to sell Alice their baby monitoring product without an explicit agreement - They certainly _should not be able to sell the data about what they hear to the highest bidder_ (say, a pharmaceutical company that wants to sell baby products) without explicit consent from the users ### What does this situation tell us about modern tech privacy? The GDPR regulations going into effect in Europe are basically trying to establish the roles and expectations, specifically for Bob the shopkeeper and the network operator. Collecting and using personal data from users will require specific legal basis. One of the reasons we ended up in today's flawed system with providers like Facebook aggregating and, arguably, abusing user data is that we did not treat various parties differently based on their role. **Facebook claims to be a 'platform' and not a content creator itself. If that's the case, they are akin to the telephone network operator (and builder). Yet they monitor (_eavesdrop on_) all the conversations between Alice and Bob and everybody else, and treat the data as their own.** Some of the interesting questions surrounding how to interpret GDPR, particularly while building a new technology like uPort, stem from figuring out which party is in which role. The blockchain technology we are building on is a tool, like phones and phone lines. It's a particularly interesting tool because it inherently stores data in public, but that is just a fact about the technology that all parties must balance when choosing whether to use it, and for what. A decentralized identity system, which gives users and decentralized applications the ability to transact directly and securely through tools like the uPort app and platform, without a 3rd party being involved at all, would also be part of that infrastructure. This makes uPort one of the network builders. We take our responsibility to build this system to minimize public data very seriously, building in privacy by design, and will help build tools that lets users and developers make good decisions. Once the system is built, uPort the company does not monitor the data exchange, nor is it even involved in it — unlike a platform like Facebook. Users interact directly. Self-sovereign identity systems like this **put data creation, sharing and storage in control of the users**, without involvement from 3rd parties. Regulations like GDPR aim to make sure receivers of data, like Bob, treat the personal data shared with them responsibly. And we need to change norms, educate users, and design systems that let users, like Alice, make good decisions about when and how to share data about themselves. _The above is purely hypothetical and is certainly not meant as legal advice._ _Join the conversation about data privacy in uPort's riot community._ ## Entering the age of data privacy URL: https://dazuck.com/writing/entering-the-age-of-data-privacy Date: 2018-05-07 Type: essay Summary: How regulation, emerging tech, and shifting public sentiment are converging to reshape data privacy. Something is shifting. For a decade, the implicit bargain of the internet held steady: you get free services, companies get your data, and everyone pretends this is a fair trade. That bargain is breaking down — not from one direction, but from three at once. Regulation, technology, and public sentiment are converging on the same conclusion: the current model of data ownership is unsustainable. Each force alone might be containable. Together, they mark the beginning of a genuine transition in how we think about data, privacy, and the relationship between people and platforms. ## Force 1: Regulation catches up GDPR went into effect in May 2018, and it matters more than most people in tech want to admit. For the first time, a major regulatory body codified the principle that people have rights over their data — the right to access it, the right to correct it, the right to delete it, and the right to take it somewhere else. GDPR isn't perfect. Enforcement is inconsistent. Compliance is often performative (those cookie banners aren't making anyone safer). But the principle it enshrines is significant: your data belongs to you, and companies that hold it are custodians, not owners. California's CCPA followed, and more regulation is coming. The direction is clear even if the execution is still catching up. ## Force 2: New technology makes alternatives possible For most of the internet's history, decentralized data control was a nice idea with no viable implementation. You could argue philosophically that users should own their data, but there was no infrastructure to make it work. That's changing. Blockchain networks provide tamper-proof, decentralized record-keeping. Encryption advances make it possible to use data without exposing it. Decentralized storage networks can host data without centralized servers. Cryptographic identity systems let users authenticate without passwords controlled by platforms. These technologies are early — too early for mainstream adoption. But they've crossed the threshold from theoretical to buildable. The question is no longer "is user-controlled data possible?" but "how do we make it usable?" This is what we're building at 3Box: infrastructure that gives users control over their own data, using Ethereum for identity and IPFS for storage, wrapped in interfaces simple enough that regular people can use them. It's early. But the pieces are coming together. ## Force 3: Public sentiment shifts Cambridge Analytica cracked something open. Not because data misuse was new — it wasn't — but because it made the abstract concrete. People could see, in specific terms, how their data had been harvested, sold, and weaponized against their interests. The Pew Research Center reported that 79% of Americans were concerned about how companies use their data. That number has only gone up. Every new breach, every new scandal, every new revelation about surveillance advertising erodes the social license that tech companies have operated under. Public sentiment alone doesn't change systems. But it creates the political will for regulation and the market demand for alternatives. When people start actively seeking products that respect their privacy — encrypted messaging, privacy-focused browsers, data-minimal services — the economics shift. ## The convergence Any one of these forces could be dismissed. Regulation moves slowly and is often captured by incumbents. New technology takes years to mature. Public sentiment is fickle and rarely translates to changed behavior. But the convergence is what matters. Regulation creates demand for privacy-preserving solutions. Technology makes those solutions buildable. Public sentiment creates the market for them. Each force reinforces the others. ## Privacy is not about hiding A common misunderstanding: privacy advocates want everyone to hide their data and live anonymously online. This misreads the argument entirely. Privacy is about agency. It's about having the ability to decide what you share, with whom, under what conditions, and for how long. A world with good data privacy isn't a world where no one shares anything — it's a world where sharing is intentional rather than coerced. The current model extracts data as a condition of participation. Want to use a social network? Hand over your behavioral data. Want to use email? Let us scan your messages. Want to navigate somewhere? Give us your location history forever. The alternative isn't no sharing — it's conscious sharing. Data as something you actively give rather than something passively taken. Data as an extension of self, managed with the same care you'd apply to any personal possession. ## What comes next The transition won't be fast or clean. Incumbents have enormous advantages and no incentive to change. Most users still prioritize convenience over privacy (and that's a rational choice given the current options). Infrastructure for user-controlled data is young and rough. But the trajectory is set. The age of treating personal data as a free resource to be mined — data as oil — is ending. What replaces it is still being built. That's the work ahead. ## Joining Consensys & decentra-mania URL: https://dazuck.com/writing/joining-consensys-and-decentra-mania Date: 2018-04-24 Type: essay Summary: Reflections on joining the Ethereum ecosystem and why decentralized technology matters for human coordination. I joined Consensys a few weeks ago. This is not the career move my 22-year-old self would have predicted. My background is in social impact — international development, education, civic tech. Blockchain wasn't on my radar until recently, and honestly, most of what I saw in the crypto space looked like speculation dressed up as revolution. So why am I here? Not for the tokens. Not for the hype. I'm here because I think decentralized technology is the most important development in human coordination since the internet itself, and I want to help build it before we waste the opportunity. ## Technology as coordination infrastructure Here's the lens that made crypto make sense to me: every major leap in human capability has been a leap in coordination technology. Language let small groups coordinate. Writing let coordination persist across time. Law let strangers coordinate through shared rules. Markets let coordination scale through price signals. The internet let coordination happen at zero marginal cost across any distance. Each of these technologies didn't just make existing coordination faster — it enabled entirely new forms of organization. Markets weren't just faster barter. The internet wasn't just faster mail. Blockchain is the next step in this sequence. Not because it makes transactions faster (it usually doesn't). But because it creates a new kind of coordination substrate: one where the rules are set by participants rather than by platform owners, where trust comes from cryptographic verification rather than institutional authority, and where the infrastructure itself is a shared resource rather than a private asset. ## The problem with the current model Right now, the most powerful coordination infrastructure in human history is owned by a handful of companies. Facebook coordinates social interaction for 2 billion people. Google coordinates information access for most of the planet. Amazon coordinates commerce. Apple and Google coordinate the distribution of software through their app stores. These are extraordinary coordination achievements. They're also dangerous concentrations of power that would have been unimaginable a generation ago. The people who set the rules of these networks — what content is allowed, who can participate, how data is used, how value is distributed — are accountable to shareholders, not to participants. This isn't a hypothetical concern. We've seen platforms manipulate news feeds to influence emotions. We've seen data harvested and weaponized for political manipulation. We've seen entire communities deplatformed overnight. The rules of our most important coordination systems can change without notice, without consent, and without recourse. ## Decentralized technology as an alternative Decentralized networks offer a structurally different model. Instead of a company owning the infrastructure and setting the rules, the infrastructure is shared and the rules are encoded in protocols that participants can verify, fork, or exit. This matters not because decentralization is inherently good — it's a design property, not a moral virtue — but because it addresses a specific failure mode: the concentration of governance power in entities whose incentives diverge from participants' interests over time. Every coordination system starts aligned. The early platform is desperate for users and builds things users love. But as the platform matures and its incentives shift toward profit extraction, the interests of the platform and its participants diverge. Decentralized systems aren't immune to misaligned incentives, but they give participants structural recourse: the ability to fork the protocol, exit with their data, or propose governance changes. ## The chaos and the conviction I won't pretend the space isn't chaotic. "Decentra-mania" is real. There's too much money chasing too few real applications. The rhetoric often outpaces the technology by years. Scams are rampant. The gap between the vision (reshape human coordination) and the current reality (mostly speculation and slow databases) is enormous. And yet. The core thesis — that coordination infrastructure should be a shared resource governed by its participants — is one of the most important ideas I've encountered. It doesn't require believing that every blockchain project will succeed, or that decentralization is always the right answer, or that the current implementations are anywhere near ready. It just requires believing that the problem is real (centralized coordination infrastructure creates dangerous power asymmetries) and that the direction is right (shared, participant-governed infrastructure is a better long-term design). I believe both of those things. That's why I'm here. ## What I'll be working on At Consensys, I'm focused on how people interact with decentralized systems. The technology is promising but the user experience is, to put it charitably, not. If decentralized coordination tools are going to matter beyond the crypto-native community, they need to be usable by regular people — not just people willing to manage private keys and navigate gas fees. More specifically, I'm drawn to identity and data. Who you are and what's known about you is the foundation of every coordination system. Getting that right in a decentralized context — giving people control over their own identity and data while making it useful and composable — feels like one of the highest-leverage problems to work on. That's vague, and it's vague because the work of figuring out what to build is still underway. I'd rather be honest about the uncertainty than perform false confidence. The thesis is clear. The specifics are being discovered. This is going to be a long road, full of setbacks and hype cycles and projects that don't work. I'm optimistic anyway — not because the technology is ready, but because the problem it addresses is real, the people working on it are serious, and the alternative (hoping centralized platforms will self-regulate) isn't a plan. ## My favorite books from a crazy year URL: https://dazuck.com/writing/favorite-books-crazy-year Date: 2017-08-08 Type: travel Summary: The best reads from an 8-month trip across 18 countries. Of course the question I get most often post-travel is: "What was your favorite place?" I expected the regular requests for travel advice. But the time away has also made me a go-to adviser on another topic: book recommendations. It makes sense — with all my free time I did read a lot. A ton, actually: 2–3 books per week at times. Here are some of my favorites from a crazy year of travel, politics, historical events and forced reflection. ## International novels _I made a habit of reading a book from each place I visited. These were the best of them, a trip to the country without ever leaving your couch._ **Beer in the Snooker Club** (Egypt): echoes of past revolutions ring true in the current time, said my Egyptian friend who read this after I wouldn't stop raving about it. **Burmese Days** (Myanmar): George Orwell's firsthand account of Burma under British colonial rule prepared me to dive into Burmese culture and history in a way that changed my two weeks there. **Beauty is a Wound** (Indonesia): gorgeous, challenging, spiritual and graphic. At times you won't know whether to flinch or smile, quote a passage to a friend or hide your blushing face. **The White Tiger** and **Holy Cow** (India): India is challenging. These books, far more fun and fast-paced than many others recommended for the startling country, make the culture and history more accessible. **It's Our Turn to Eat** (Kenya): this one's not a novel. It is a thriller, a memoir, and a commentary on a country all wrapped in one. It's hard to find copies of this dramatic whistle-blower story inside Kenya — too hot to handle. ## Thinking Big _Are the titles in this section ambitious enough for you? Away from the pace of normal life, I surprised myself by diving into philosophies and deep history. None are challenging to read, all will challenge your current view of the world._ **Lessons of History**: more deep insights per paragraph than any book I've ever read. Short, simple, and full of quotes that will make you sound like the most enlightened person at any dinner party — and you will be. **The Happiness Hypothesis**: part survey of philosophers and religions, part original thinking, Jonathan Haidt (author of **The Righteous Mind**) examines what makes us happy and why — pulling together science from psychology, biology and evolution in a way that is dead simple to follow. **Meaning of Life**: an audio-book compendium of lectures on what we are here for, both religious and philosophical takes. Will make up for years of doodling through high school and college courses. **A Short History of Everything** and **A Little History of the World**: exactly what they sound like — two very different histories of the world in a couple hundred easy pages. The former is more scientific, the latter more historical. **Beginnings of Infinity**: the first book that is truly hard to read — Deutsch claims that 'good explanations,' which humans are uniquely capable of, makes all of history up to this point a mere preamble for the infinity that comes next. **Zen and the Art of Motorcycle Maintenance**: I can't believe how long it took me to read this book. Fun, beautiful, difficult and different. I won't try to recap, I'll just say that it's a much-needed alternative view in an age that has in many ways made reason dogmatic. **Sapiens**: I'm torn on this. Huge, interesting ideas. Hard to contextualize and critique. I think I need a re-read and a discussion group. ## Non-fiction **Against Empathy**: the most challenging — and convincing — book I've read in a long time **The Unwinding**: reads like a novel, and without ever being explicit shines a direct light on the changes and dynamics in American society that brought us to our current situation. **How Adam Smith can Change your life**: an easy intro to Adam Smith's "other" book, **Theory of Moral Sentiments**. It's popular to hate on economics these days, but Smith's prescience on humanity and his own predictions on the ill-effects of capitalism give a far more productive lens to think about human relationships — economic and otherwise. **The Master Switch**: for anyone who thinks the Internet is destined to stay open, read this history and think again. **Shoe Dog**: shocked at how much I loved Phil Knight's (at times cheesy) tale of athletics and entrepreneurship in building Nike. ## Skills and How-to _I'm a type-A nerd, even when on sabbatical. I loved these books and found them shockingly useful._ **How to win Friends and Influence People**: despite constant recommendations I avoided this for a long time — it just sounded bad. But this classic is not about manipulation, it's about understanding people and having good relationships. It's changed how I interact with people, especially in disagreements. **High Output Management**: like I said, I'm a nerd… **The Hard thing about Hard Things**: …and type-A. ## Highlights from 8 months of travel URL: https://dazuck.com/writing/highlights-8-months-travel Date: 2017-06-15 Type: travel Summary: The best moments, biggest surprises, and hardest lessons from traveling across 18 countries. Across 8 months, 18 countries, 4 continents, 28 states; staying in 53 hostels, 12 hotels, 8 friends' homes, 6 homestays, 4 airbnbs; traveled by 22 flights, 21 bus trips, 9 scooter rentals, 8 boat rides, 4 car rentals, 3 train rides, and way too many rick-shaw/tuk-tuk/matatu/boda-boda/tika-tika/uber-moto rides to count. Reading 22 books, listening to 19 audiobooks, writing 18 blog posts. Adventuring to 32 beaches, 18 waterfalls, 15 temples, 13 national parks. Mostly smiling through 4 hospital visits, 3 different backpacks used, 1 missed flight, 1 time stranded in Africa with a flat tire. Traveling with 2 of my best friends, staying with 3 more who hosted us in their countries, seeing 12 friends and family who visited, and meeting ~200 new friends on the road. Numbers can't quantify this year, nor can bullet points bring them to life. But here are some of the best, lowest, and craziest parts of the trip that I'll remember forever. ## We kept finding ourselves at the center of historic political situations. While not usually convenient, it was always fascinating. Thailand just after their beloved King had died, with an unclear succession plan and military-enforced country-wide grieving. India at the height of their self-inflicted cash crisis, after they made 85% of circulated currency invalid and ATMs were dry. South Africa during country-wide strikes calling for the President to step down, amid 'allegations' of widespread corruption. Kenya in the lead-up to national elections, with high anxiety about a possible repeat of 2007's post-election violence. Egypt in a state of emergency and military control, following April's bombings. ## When making my list of categories, I realized nearly every country I visited was my 'favorite' in some way. **Best scenery:** Rwanda — the land of 1000 hills is stunning everywhere **Best tourist-ing:** temples of Bagan in Myanmar **Favorite hike:** Mai Chau in northern Vietnam **Best meal:** Bun Cha in Hanoi; I went 3 times at least **Best dumplings (they deserve a separate category):** soup dumplings and pork buns in Hong Kong **Best drive:** everywhere in South Africa **Best tour:** sunset boat ride at a deserted Tam Coc **Best beach:** Indonesia — Balian's black sand was my favorite **Favorite city:** Hanoi, closely followed by Yangon, Cape Town, Cairo **Most interesting cultural experiences:** visiting schools in South Africa **Best wildlife:** Kenya — hard to beat a safari in the shadow of Mt. Kilimanjaro. **Favorite fiction book:** Beer in the Snooker Club (Egypt) **Favorite non-fiction:** It's our turn to eat (Kenya) **Favorite travel crew:** the motorcycle gang in Malaysia **Favorite locals:** Chill and Jule, the amazing bartenders at Broken Compass on Gili T **Best hostel:** Dalat Family Hostel **Best non-hostel:** Sheraton HK **Best transit:** Thai Airways accidental business class flight to Melbourne **Best activity:** class 5 river rafting on the Nile in Uganda **Best trip-within-a-trip:** Castaways in Ha Long Bay **Best performance:** our bollywood dance at an Indian wedding **Best travel luck:** meeting Pratham, our incredible host in Northern India **Best nightlife:** Thailand — Chiang Mai and Koh Phi Phi **Best party:** Sundays at Single Fin in Uluwatu **Best national beer:** Vietnam — Bia Hanoi **Best craft beer:** Boston Beer, in South Africa oddly **Best national drink:** wine in South Africa — and avoiding rice wine everywhere in Asia **Best coffee:** Australia (Byron Bay and Melbourne), narrowly edging out Hanoi's egg coffee **Most "did it for the story" moment:** taking (wheeling) Hoffman from her overnight hospital visit in Hong Kong straight to the casino in Macau **Best holiday/festival:** village Buddha unveiling ceremony in Chiang Rai with our airbnb host **Best sporting event:** Rugby game in Durban, South Africa **Place I most want to return to:** Egypt — far too much to see and such an amazing crew. **Place I'd most likely live:** Hong Kong **Place I enjoyed the most overall:** South Africa ## Egypt: ending on a high note URL: https://dazuck.com/writing/egypt-ending-on-a-high-note Date: 2017-05-31 Type: travel Summary: Finishing an 8-month trip across 18 countries in Egypt — pyramids, chaos, and closure. The last hurrah: Egypt was planned months ago — in some ways years in advance — and is a fitting end to the trip. It was always going to be either a disaster or a highlight. I'm so thankful for everyone who made it one of the the place I'm most eager to return to soon. There was definitely a piece of me thinking "just get through this and you've made it." Skiers say that injuries always happens on the last run, and a version of that felt all-too likely here. We had racked up plenty of hospital visits and illnesses and a few other close calls. My energy was a bit lower after Uganda and Kenya, and Egypt — while perfectly safe if done right — isn't exactly a place to mess with. "Cairo is a monster. This city will eat you." That was the prep message from my friend Mo, whose family is from there. I would be meeting him and 3 other close friends in Egypt, which was exciting but can be a bit dangerous too. Sometimes there is strength in numbers, sometimes there is irresponsibility. Mo grew up in Menlo Park and spent most summers back here with his cousins and family. For more than a decade he'd told us stories of the city, shown videos of protests he'd joined in Tahrir Square during the revolution, talked of the 'honking language' of Cairo drivers and shared some of his cousins' amusing antics. We'd been talking about joining him on a trip here since 7th grade. He was right: Cairo is a monster. Some estimates put it at 45 MILLION people, unofficially. That's 60 San Francisco's packed together. It's got the sprawl of LA and the density of Manhattan, and then add in some half-built squatters high-rises to every nook and cranny between the official buildings. Everyone there kept assuring us the traffic was far calmer than usual — in fact, the city felt emptier and less alive than it used to, they said. That was hard to imagine after our first freeway trip: "Seatbelts are illegal in Cairo" our driver, Mo's cousin Seif, told us before launching into the chaos. I gripped the sides of my seat a bit too tightly as our small coupe careened around trucks and squeezed between breaking cars, Seif tapped out rhythmic honks that each meant something different (there's "Get out of the way" and "I'm merging" and "Fuck you" and many others). I relaxed a bit after a while, comforted that there must be some order to the madness after all. There was no real merging since there are no lanes, but at times it felt less like moto-cross and more like an endless weave since our friend's car kept popping up in the same place relative to us: we'd pass them on our right, and a few minutes later spot them up ahead on our left. "If you can drive in Cairo you can drive anywhere," Seif said. And then revealed he had been in 3 major accidents. Our first few days were an endless stream of city sights. Mo pointed out the neighborhoods and directions from the top of Cairo tower to get us oriented, and we explored the oldest districts and markets and the new park built by the Aga Khan, the mosques and the city of the dead, the Egyptian Museum and Tahrir Square. We chilled at cafes with hookah and went to the Cairo Jazz Club (CJC). We went to the live filming of Abla Fajita, Egypt's most popular tv show — a mix of the Daily Show and the Muppets (and understood precisely zero of the 4-hour Arabic affair). And of course the Pyramids. I'm pretty jaded when it comes to sightseeing at this point — there are a lot of old, impressive, history-filled buildings out there. But Egypt staggers not just with the amount of history but with the age of it. The pyramids, as tall as many modern buildings (450'+), were built about 5000 years ago. The time from when Jesus was born to right now was is shorter than from when the pyramids were built to when Jesus was born. When first completed they were the tallest structures in the world…and 4000 years later they still were. And they still stand, nearly original, just on the outskirts of this monster of a city. And we crashed the 'wedding of the year,' complete with an appearance by Amr Diab -- the biggest pop star in the Arab world. (Aside: I should have made crashing weddings the theme of the trip). Mo and Seif's childhood friend was getting married on the outskirts of the city and, having never met us, told them "sure bring whoever" and so after digging up some suits and tuxes (those didn't make the travel bags) we were in. The wedding started at 4. We checked into the hotel at 3:30. They said we'd go to the wedding at 9, which we thought that was a joke until we walked in later around 9:30. We spent the afternoon at the hotel pool and meeting our new Cairo crew — Mo and Seif's friends from growing up. They had all attended an American school so talked mostly in English for our benefit, and we got a good fill of the endless stories and nonstop banter. Arabic lends itself to original jokes and wit (and insults) far more than English as any root word can be conjugated any way, so every few minutes they'd lapse into Arabic to make fun of a friend and the whole room would crack up and Mo would try to translate jokes that didn't translate well. (If I ever decide to pick up a new language, it will be Arabic). After a week of exploring Cairo with the guidance of Mo and his friends, we started to get a feel for how things worked. We had now taken a couple of taxis successfully on our own — a new process both because of the language difference and because paying for taxis involves a heated, somewhat lengthy negotiation _after_ you arrive at the destination. How much would this taxi be? "Whatever I decide to give him," said Mo. Apparently we Americans also tend to slam taxi doors, because they yelled at us for that every time. We could make our way around our home island of Zamalek and had hit the key sights. We had heard the universal frustration with the military government and plenty of opinions on the current president. We heard our new friends' stories of setting up roadblocks after the police had been chased out of the city, and taxi drivers' frustrations with the lack of tourism. We also started to see other pieces of the culture, like the customs around hosting guests. Mo and Seif's insisted on paying for nearly everything we did, and to try to pay for ourselves in front of others was disrespectful. Finally 'getting the hang of it' we of course decided to leave — we grabbed a train for a 2-day trip up to the coast to Alexandria. First of all, Alexander the Great's biography is crazy. He conquered half the world and racked up mythology to rival anyone else, and did it before dying at ~32 years old. When Justine Musk wrote about what makes people like Elon great, her example for who he studied was Alexander the Great. He also lent his name to Alexandria, which it turns out is the most common city name in the world. The Alexandria on Egypt's Mediterranean shore has to be the greatest of them those. A city of 4.5 million people (no big deal), it blends the craziness of Cairo with the wide sea-front cornices of Europe's Mediterranean cities. It's hookah cafes are planted next to fresh fish joints, and packed street markets selling Arab spices and lanterns for Ramadan open onto lively plazas with running children and live music. A friend of a friend, Ayman, met us on the second day and showed us around the city — another phenomenally generous host that made our time here so special. He took us on a boat ride through the city's nicest beach, showed us several of the president's mansions (the military has been claiming just about everything, it seems), and brought us to the new Library of Alexandria. After a night spent smoking hookah and sipping Stella and of course deep intellectual conversation until the sun came up under Alexandria's famous Stanley Bridge, we headed back to Cairo. On my last night, we rode a boat down the Nile in a falucca with Mo's cousin and his crew. We hadn't spent all that much time with them, but we lounged and talked and then went to dinner like old friends. So welcoming. Perhaps it's the immediate nostalgia of the trip ending, but Egypt has the most I'm excited to return to. A boat cruise down the Nile to Luxor, climbing Mt. Sinai, diving at Sharm al Sheik, adventuring through the desert with Bedouins, more exploration Alexandria and the nearby coast. And more time in the monstrous city of Cairo, more time with the friends we met there, more falucca floats down the Nile, more evenings at a cafe talking over hookah. As impressive and alluring as all the sights are, as awesome as the sheer size of the city and craziness of the highways are, it's the people that made Egypt. It's the people that make everywhere. ## Kenya: the essential stop URL: https://dazuck.com/writing/kenya-the-essential-stop Date: 2017-05-31 Type: travel Summary: Why Kenya was the stop every traveler recommended — and whether they were right. _(photo albums: Nairobi and safari in Amboseli)_ Kenya was perhaps the most inevitable country on this trip. In planning mode a year ago, my friends and I debated how to decide where to go. Start with Africa or Asia? Squeeze in South America or save it? Prioritize big economies (China) or interesting cultures (India) or amazing sights (South Africa) or the most unknown (Middle East)? Focus on meeting new people or visiting expats? Disconnect from work completely, or try to be productive on the road? Follow established routes or make our own way? No matter what the answers, Kenya and Nairobi would have been part of this trip. Safari is the most unique trip I've ever done. Nairobi's tech scene is famous — home to many expats, fascinating companies, and controversial education work. There are famous beach hostels and the coast and traditional tribes inland. It's safe but challenging, especially with upcoming elections putting the tension between tribalism and nationalism back in the spotlight. For better worse it's a symbol of Africa to the western world, and was a place I had to visit. In fact I had been there before, which makes it the first repeat country on this trip. My family went on Safari and toured Nairobi in 2007. Ignorant at the time of the world outside my upcoming start of college, I had no sense of the upcoming Kenyan elections. Reading accounts from the time, it seems my own ignorance wasn't the only reason I didn't sense anything strange — most people took for granted that Kenya was now a peaceful, if flawed, democracy. That was shattered when tribal fighting erupted in the wake of a clearly stolen election. 1300 were dead, 600,000 displaced, the image of Kenya shattered, and a new set of questions emerged. This visit I tried to prepare a bit better. The best book I read in anticipation was Michela Wrong's _It's Our Turn to Eat_ (not available in Kenya). Kenya is composed of a few dozen tribes within 3 ethnic groups. Without getting into the history, the main thing to understand is that in Kenya, historically, your tribe is your identity. It dictates the language you speak at home, the traditions you practice and the values you hold. It is to the tribe that you have loyalty. And tribes are as much in competition within Kenya as they are in cooperation. This is how the country's politics have operated too, according to Wrong and others I talked to there. If Kikuyu are in power, Kikuyu companies get business and Kikuyu people get government jobs and Kikuyu regions get infrastructure projects. In the US we have the same "problem" but it's on the margins — "nepotism" and "kickbacks" and "pork barrel" spending. In Kenya, that's just how everything is. So Kikuyu families — the whole tribe — thrive with education and prosperity and access; they gain more power and also feel more gratitude and obligation towards the tribe and the cycle repeats. This attitude extends not just to politics but attitudes in life. Petty corruption — officers stopping your car for a small bribe — is not seen as blatantly immoral. It's the way he supports his extended family. It's expected, part of life. There is less effort to control things according to strict rules, which means people don't expect life to conform to expectations. An expat friend I stayed with sells stovetops and travels with local colleagues to talk to families and tribal leaders. He would go crazy with frustration at the illogical problems and arguments he heard at times; his coworkers shook them off, understanding and empathizing with a life and attitude formed from different expectations and values. Tribal loyalties are not unequivocal bad. For a country as poor as Kenya, relatively few people starve. Those in power support dozens of extended family members in the tribe. They watch out for their own. Sure, it hurts the country as a whole. The US does things that hurt other countries too, and we're fine with that. We've chosen our tribe (in many cases) to be as "Americans". They've just chosen (passively) to draw their tribal borders not around the nation but around, well, tribes. Most in Kenya do agree this is not a productive system for the country long-term. Nobody wants the tribal violence of 2007, or such unfairness in the operating government. So there are moves to improve and lots of focus on this election cycle. Will there be a repeat of riots and violence, an endless cycle that surface changes hide every few years to make sure the international aid money keeps flowing? Or is this time different? In 2010, in response to 2007, the government passed a new constitution referred to as devolution: decentralizing power into the regions rather than concentrating it with the national government. This made national elections a bit less monopolizing over jobs and funding and power. Now each region has more power itself, and since the regions map relatively well to the primary tribes, this should give each tribe more power and make the competition for national control a bit lower stakes and more reliably peaceful. Six months before the elections though (during primaries), there are reasons for worry. Conflict in the north, unrest and some riots in the west…Though less powerful, the national government still has a lot of influence that comes with it. And emotional responses may not be tied to power alone, but historical frustration and other emotions like pride and shame that come with having your tribe in leadership. Frustrations could still boil over. Of course regions don't match tribes 1:1 either, so even at the local level there could be more conflict. However, there are reasons for real optimism. **As a journalist friend put it:** > **It may be good that there is conflict now during the primaries. For the first time, politics are moving beyond pure tribalism. People are getting upset in elections between members of the same tribe — that means they are getting upset about issues and stances, not just identity.** That would mean they are inching closer to a national-oriented democracy — people debating and driving a unified country based on issues that matter to them, not competition between internal groups. It's unclear to me how — if — that would play out more broadly: in Kenya, the tribal attitude isn't just about organizing political life, it plays into everything. Will this change? Should it? I asked a lot of people in Kenya if experience there — politics, corruption, and otherwise — has changed their view of how things should work. If their perception of fairness and politics and institutions have shifted. This is a group of thoughtful and particularly open people, but still all expats. "No" they almost unanimously said. A fair, logical, corruption-free system still seems right. Maybe even more important given the issues they see in Kenya. But they all followed up to add a more nuanced view, of which the most common theme was this: maybe things don't work as they 'should' in Kenya, and some systems may be objectively better. But you can completely understand how others arise and are maintained. You can understand — and not judge — an alternate morality: it's not about corruption and power plays here, it's about watching out for your (extended) family in any way you can. If you know me, at some point I've probably told you to read _The Righteous Mind_. The book changed my life in a way I didn't know was possible, shaping my thinking dramatically for the months that followed. At the peak of my obsession with becoming more understanding of other moralities and questioning my own, a friend told me in exasperation "Dan, you're taking this too far. You're approaching moral relativism." He was right and I knew it immediately. But I initially found it hard to balance my new goal of appreciating other value systems as valid and maintaining a set of guiding principles that I truly believed in. I've wrestled with that contradiction on this trip, and in the end it's not actually all that hard to square. I can understand other views without endorsing them, appreciate their hidden merits without forgetting the drawbacks. You can — should — question your beliefs and seek to appreciate contrary ones; but that doesn't mean you have to perpetually disregard any views as invalid or give up trying to prioritize them. Better to hold your views as hypotheses rather than truth, and also better to have hypotheses rather than endless open-ended questions. Kenya wasn't the place where this reconciliation took place, but it is where they took most concrete form. In Nairobi I met with a local guy, about my age, who I met as a fellow reader of Slate Star Codex. We talked about many of the issues above. Whereas I felt I was a bit late to this philisophical party, following a well-worn path towards more mature views that many of my friends are on too, he did not. He felt he had nobody around him that would relate to the questions he was asking himself. He had tried, and just a very few others were trying to drive Kenya towards a thoughtful reckoning of the changes taking place. Kenya is unconsciously reckoning with these questions on a national level, with consequences on both sides. If they move towards more 'fairness' and individualism, they lose parts of their history and their identity and their culture, and people who are supported by 'unfair' policies now will suffer. That doesn't make progress wrong, but it's not a pure black and white right either. Recognizing that can help us consider and truly understand both sides, so hopefully we can do a better job anticipating problems and mitigating negatives. In Kenya maybe that would have stopped the outbursts of 2007, in the US perhaps it would have changed the election in 2016. ## Rwanda: hills and 'how' URL: https://dazuck.com/writing/rwanda-hills-and-how Date: 2017-05-22 Type: travel Summary: Rwanda's landscape and the question of how a country rebuilds after genocide. _Full photo album_ [_here_](https://goo.gl/photos/qunc92ZKZN33FX7m8) If [South Africa is a study in contrast](https://medium.com/@dannyzuckerman/south-africa-land-of-collisions-1f5d3f33676b), Rwanda is practice in disbelief. I usually strive to ask 'why,' trying for deeper understanding and new perspectives. In Rwanda, I just kept wondering "how?" If you're like me and most, **the first thought Rwanda conjures is the genocide of 1994 and "how could that happen?"** While any genocide escapes comprehension, the history of the Hutu and Tutsi make this particularly hard to digest: there was basically no ethnic difference between the two groups, and the divide only started by relatively arbitrary classifications by Dutch colonialists. How can a group of people be willing to carve their neighbors and friends with machetes over arbitrary distinctions? On my flight to Rwanda I happened to watch The Stanford Prison Experiment. In it, a group of students were assigned to be prison guards over another group of student prisoners. Before long, the 'guards' were beating the 'prisoners,' acting with condescension and brutality and wielding their power with relish. Meanwhile the prisoners became submissive, fitting their psychology to their role. The assignments had been random, but the roles created new behaviors, new people. **People are just a product of their situations.** In the lead-up to the genocide, Hutu propaganda built fear that the Tutsi were plotting to slaughter them. Radio addresses primed protective and tribal divisions, playing on historical inequities (Tutsi had been given more elite status by the Dutch and at times had abused that). In that environment of fear, economic hardship and political uncertainty after the president's plane was shot down, the people of Rwanda were on constant edge. Unlike in many places, this was not the product of super deep-seeded historical conflict, or irreconcilable cultural differences. Hutu and Tutsi were married, were friends, were in better times completely amicable. Many — perhaps most — could not readily distinguish between the two ethnicities. It was only when primed for a new situation of perceived inevitable conflict that Rwandans were persuaded to murder their neighbors: 800,000 of them in 100 days. In thinking about this genocide it's natural to shake your head, sadly mourn "how could anyone ever do that." It's tempting to stop there — to keep the question rhetorical. That's also a cop-out. Over and over in war-time museums and stories and remembrances, it's made clear that war-time acts cannot be judged by peacetime standards. How could you do that? How could Hutu murder Tutsi friends? I don't have an answer, except that to ask that question in the abstract or as if the context is one we can relate to completely misses the point. People are a product of their situations — you can't judge how people acted without understanding the events and atmospheres that led up to it. And while we're on the subject of trying to understand how people could enable a genocide: **how could the US and the rest of the world stand by and let the genocide happen,** even avoiding the term 'genocide' to skirt the responsibility that term implies? Again, I don't pretend to have an answer, except that people were acting in a different situation in 1994 than we feel today in 2017. As surely as we could judge that inaction, 20 years from now I am sure others will judge ours about global poverty and many other issues. That's not to excuse it and move on — we need to learn from our past mistakes. But that's the point — learning requires doing more than shaking our head and dismissing it with a 'how could anyone ever do this." Getting to know Rwandans, you keep wondering: **how do people who murdered each-others families, parents, children now live and work side-by-side, moving on as part of what feels like an outstandingly safe and productive city?** It seems impossible. On the rare occasions where people talk about it, they say it is nearly impossible. But it's also the only option. The alternative: to perpetually relive the horror and constantly seek vengeance, justice, etc. and not move forward with lives. Instead they do all they can to forgive and move on, however absurd that might seem. As valiantly as they try, it doesn't seem fully possible. It doesn't seem that forgiveness for such things can ever be complete. Which leads to the current compromise: just don't talk about it. Ask any expat who visits or lives there: the genocide is not a topic you can broach, no matter how close you are to locals. To think of it would be paralyzing, making not just progress but daily life untenable, so the only option is to bury the pain deep for most of the year. Each year, national genocide remembrance week is the opportunity to remember, grieve, discuss, and constructively move forward. To face the past and make sure the issues are never repeated. To make sure Rwanda is one Rwanda not a collection of ethnicities. But it has to be confined to this week, this place, with all the structures and security to make sure it is constructive and not a path to reliving the conflict. To deal with that all the time the way for example Germans do the Holocaust or how South Africa tried to deal with reconciliation after Apartheid is impossible, because the victims are still living and you can't put them through that. Humans are incredibly adaptable — we get past grief, we move on, we adjust our happiness. But we are not infinitely adaptable, and even the strongest among us can understand that sometimes pain is just not something you can tough out. Walking around Kigali, after stories and images of outright war and murder 20 years ago in the same streets, you continually wonder **'how is this place so clean, so safe, so thriving."** Especially in contrast to other African cities, it's just so livable and functional. It didn't happen by accident, it was a series of choices by leadership. No corruption, massive security enforcement, extreme discipline. Everything possible to attract businesses and build infrastructure and invest in the future along a well-set plan with organization and discipline you rarely find elsewhere. I don't mean to paint this picture of a particularly effective set of policies to follow. There are downsides. True opposition voices have a way of falling silent. Punishment very often outstrips the crime. Expats and locals alike won't talk too loud about politics — intelligence and bugs are everywhere. Modern buildings look attractive, but you hear many are pretty empty. Businesses can get started easily, but the population isn't educated enough for many jobs or wealthy enough to buy many services. As one friend said: it's a bit like the 'Pleasantville' of Africa. You start to wonder at what's just below the surface. It kind of makes a farce of the idea that there are objectively better or worse policies to follow, or at least knowable ones. Should Rwanda embrace more open democracy and opposition speech? Human rights groups and westerners tend to think so — but what if there's reason to believe opposition will stir up ethnic tensions that could rekindle the genocidal behavior? That's a very real possibility. How do you balance that possibility with avoiding another strongman dictator? **As you drive around Rwanda, you are in constant wonder: "how is it so beautiful, and so packed with people."** Everywhere you look the land of a thousand hills is rolling greenery and valleys, the most stunning scenery around every bend. Also around every bend: people. Kids on the street, adults walking around, towns congregating on the side of the highway. In the hills where pathways have been cut, along the guardrails of the highway, in makeshift soccer pitches between two mountains, there are just people everywhere. It's among the most dense countries in Africa, and there's no more room. There's nowhere for them to go. They already farm every inch of hillside land, what next? Rwanda likes to compare itself to Singapore, but they don't have the port-driven wealth to invest in education. People are sending their kids abroad for school and job prospects — there's no room for more people here, and no real jobs right now it seems. **All this makes you wonder "how is this going to turn out."** Will the new generation, educated like none before it, take advantage of the infrastructure and institutions and drive Rwanda to be a business capital? Or will modern buildings stay empty while the elite talent is drained from the country and it continues to subsist on coffee exports and gorilla treks? As academics and observers debate President Kagame's policies, go back and forth on whether Rwanda's progress is real or surface, evaluate business prospects and human rights records and development priorities, arguments tend to take two forms: either they judge Rwanda as a country in abstract (as if good policy is good policy everywhere) or they judge it in comparison to the genocide two decades ago (as if any peace and progress is impressive and celebration-worthy). Both of these miss the mark. How is a landlocked country with a history of genocide _supposed_ to act? There's more than one right way and plenty of wrong ones, but the considerations better take its history and situation into account. **Countries, like people, are a product of their situations. Rwanda made me question a lot of the principles that I took as absolute before.** Free speech and opposition politics have a very different range of outcomes in Rwanda than in the US. And they may have very different implications in Rwanda 10 years from now versus today. Rwanda helped me start to see policies, rights and even moralities as structures that build on each other rather than as separate values that stand in isolation. How to balance those when they are all interconnected? In my brief time I'm impressed with how President Kagame and the Rwandan people are trying to in their situation right now. ## Extremes in India: 4 stories in 8 days URL: https://dazuck.com/writing/extremes-in-india Date: 2017-04-15 Type: travel Summary: Four stories from eight days in India: a 48-hour Air India saga, VIP hospitality and a president, crashing a wedding with a Bollywood dance, and a demonetization cash crunch. ## Pt 1/4: Mutiny on Air India _(originally posted December 4, 2016 on dazuck.com)_ I might have a bit of a reputation for crazy plane delays, re-routings, and other bad luck on long journeys. Nik might have even worse luck than me. So we should probably have been more prepared for a less-than-smooth trip from Bangkok to Amritsar, but a comedy of errors from us and Air India led to an unexpected 48 hour journey. Here’s the rough retroactive time-blog: Nov 19 **9pm**: online check-in won’t go through for our international flight: Bangkok > Delhi > Amritsar. No big deal, we’ll check-in live tomorrow morning. It’s a 9am flight so we’ll wake up at 6 to subway there. 9/20 **6:45am**: Google Maps’ routing seems a bit weird. This might take longer than we expected. **8:02am**: It turns out that Air India’s check-in desk closes 60 minutes before departure, which is 58 minutes from now. There’s time, there’s will, but there’s no way in. There goes flight #1. **10am**: $12 worth of international phone calls later and we’re booked on the same itinerary tomorrow. Back to Bangkok for a Saturday night and let’s try this again. 9/21 **7:30am**: through security with plenty of time to spare! **9am**: still waiting at the gate at departure time, so…guessing we’re delayed? **9:45am**: _“Hello ladies and gentlemen. Due to a technical issue, this flight is delayed. The new departure time will be 10am. We will board as soon as possible.”_ hmmm **11am**: new departure time set for 11:15am. Funny pattern. **12pm**: Coupons for coffee and vague directions to find the new gate. **1:30pm**: new departure time 1:45pm. Pattern seeming less funny. **2:15pm**: new departure time set for 2:30pm. The connection on our 5-hour layover slips away. **2:30pm**: we’re boarding! Maybe if they make up time in the air we’ll make it… **3:30pm**: we’ve boarded, but now we’re just sitting here and it seems the AC doesn’t work. I’m watching Captain America on the seatback and sweating like I’m in one of the battle scenes myself. **4pm**: still no AC, still no movement, no announcement from the captain and I’m not sure where the crew is because Nik has started passing out water in the aisles himself. **4:30pm**: “Please deboard.” (Actually it took about 4 announcements and some shouting, but that was the gist). **4:45pm**: This is chaos. It’s like they’ve never deboarded a plane before. Nobody from the airline seems to know whether we should stay at the gate or leave. It’s actually not that clear who even works for the airline. One person who was at the gate is leading a group to exit through customs, but there are two customs areas plus a ‘special customs’ area and about a third of the group heads to each. A security guard has NIk’s passport and started to walk away with it. After chasing him down we scramble to find part of the group again, because there’s no central place to go get information, so that’s really our only hope. There’s rumors of a hotel in Bangkok for the night but no clarity how we get there, when we’d fly out, or who can tell us these things. The flight crew look even more confused than we are, and apparently each has different information because they are yelling at eachother in Thai. I actually feel more pity than frustration — having nothing on the calendar for, well ever, helps ease stress. **5pm**: a new rumor spreads just before we exit customs — the flight ‘might’ take off after all in a couple hours. You can wait at the gate or go to the hotel, but if you leave the airport there’s no getting back in. **5:30pm**: someone from Air India confirms those rumors after 30 minutes of questions. Sure, someone else from Air India denies them but we’re taking the optimistic approach **6:30pm**: more food/drink vouchers **7:30pm**: we’re boarding! (again) **8pm**: still seems to be no air conditioning. It is HOT **9pm**: so, we’re still sitting here with no AC. The crew has no idea what’s happening; seems like the captain is telling them exactly as little as he’s telling us. A lot of passengers are getting angry. **One just led a group up from the back and yelled to the plane “Will you stand with us! This is unacceptable. We need to stand and make them tell us what is going on in front!”** Another group tells him to ‘give it 10 more minutes, and then we will go.” Are we literally about to have a mutiny onboard? **9:05pm**: plane starts moving. Is it bad I’m slightly disappointed we didn’t get to see what would happen in 5 minutes? **11:30pm-2am**: flight lands in Delhi and we work with ~19 Air India staff to find our way to a hotel for the night (aka 2 hours) before our new flight to Amrtisar at 7am tomorrow — check-in by 5am latest. 9/22: **4:30am**: wake-up. 2 hours of sleep sucks. **5am**: airport, check-in, security, gate. 90 minutes before boarding. Try to stay awake. **8am**: land in Amritsar, taxi to hotel, check-in, pass out. **9am:** part 1 officially ends, and part 2 is about to begin. And oh how different it would be. ## Pt 2/4: Hello, Madame President _(originally posted December 3, 2016 on dazuck.com)_ Most of this blog will aim to be my reflections rather than FOMO-inducing travel stories and photos. This post will be an exception. While we were waiting in Bangkok, our friend Rob sent us a string of messages from Amritsar that amounted to _“We met the best host ever on the plane who is taking us VIP to all the best things to see in Amrtisar and has invited us to his family’s University/hotel/home/candy shop in the next town over.”_ While disappointed that we had missed day 1, there was lots of potential for the next couple days. A driver was waiting to take us to Jalandhar on Monday morning to get started. The first hint of how we’d be rolling around came as we pulled into Lovely Professional University. You know how ski patrol wave at you to slow down on main runs at busy times? That’s how our driver treated the security guard waving him to stop at the gate to the admin building. Our new host (and bff) Pratham welcomed us and ushered us into what felt like a cross between the control room in Star Trek and a Miami hotel suite. It turned out to be the office of the Chancellor, aka his father. We sat down on the couches, chatted as the first commencement ceremony wrapped up, and tried to act comfortable as people in suits brought us coffee and tea. Why act comfortable? I’m normally very at home in flip flops and t-shirts. ### But beach attire quickly goes from appropriate to awkward to downright silly when you’re sitting in a formal office and a commencement procession walks in and you get introduced to the honored guest: the President of Mauritius. (photo proof coming soon) After brief introductions, the returning party took seats at the other bank of couches in the office. And then Pratham beckoned us to pull up chairs and join. So we did. And for a good 20–30 minutes our group chatted with the President about her unexpected rise in politics, her background in business, the African Union, her policies and initiatives, her ties to India and LPU, and I’m sure plenty of other things I missed as I marveled at the absurdity of what was happening. We wrapped up the conversation without embarrassing ourselves too badly (except for the initial impression, obviously) and adjourned to a tour of campus (plenty of thoughts on that coming soon), a many-course lunch, and more coffee and tea. Then it was time for the afternoon commencement ceremony. Earlier in the day I had assumed we’d sit in the back and observe from a distance. By now, I knew better. [https://youtu.be/CZZtI4pkACY](https://youtu.be/CZZtI4pkACY) Still in casual gear, we followed Pratham’s lead and walked with the marching band into the auditorium, apparently part of the procession. Then he beckoned us into a couple of large leather couches in the front row. Alright, time to feel even more out of place than when meeting the president. After the speeches — 3 of which referenced the ‘visitors from the US/Harvard’ — it was time to move onto an even sweeter part of the day. We got back to the Chancellor’s office and the procession congregated again, and we briefly milled and drank tea before Pratham ushered us back out to cars in a bit of a hurry. “We want to be part of the motorcade to avoid traffic!” That would be the presidential motorcade, which was taking us to his family’s emblematic sweets factory, Lovely Sweets. After the easiest in-India drive we’d ever seen, and working our way through the crowd outside waiting for the shop to reopen, we entered a 5-story Indian sweets emporium. Of course we made no real decisions — Pratham took us from counter to counter and ordered samples of his favorite candies — nuts and fruits and chocolates and yogurts and pastries and so many combos I can’t really remember — working our way up floor by floor. At one point, so beyond stuffed from 3 courses of lunch and ~10 courses of dessert already, his uncle came to say hi and that Pratham didn’t know what was best and that we should start the tour over with him. Torn between our stomachs and politeness, we exchanged panicked looks until Pratham saved us. The top floor of the factory is a display of the box sets that Lovely Sweets creates for wedding invitations, of which even the most modest are impressive when you realize they are sent to every guest. I believe it was around here we started discussing what to bring Kunal for his wedding as a gift from our group, and in addition to a sweets box **Pratham excitedly suggested “You should do a Bollywood dance!”** Apparently, this was tradition and everyone loved folks who did a prepared dance at the wedding, showing respect and helping start the party. Not that any of us had any Bollywood experience, or general dance skill, or time to learn with the wedding 24 hours away. As soon as we were back at LPU, Pratham arranged for one of the university dance teachers to choreograph a popular song for us. 3 hours of practice, a lot of patience and MANY simplifications from our teacher, stifled laughs from Pratham and his mother, and we kind of had it a little bit down. We could, with a bit more practice and a few drinks, try it at the wedding tomorrow. Pratham had other ideas (we secretly think it was his self-reward for the insane hospitality he’d been showing us): there was a “One-World Festival” as part of the commencement ceremony, with thousands of students visiting, a few ambassadors to India in the front row, and celebrations of various cultures with dance and other performances. Nobody had done the USA yet. We should do our dance for the whole school, Pratham suggested. I don’t think any of us loved this idea, but at this point Pratham’s word was law, so after the LPU MC pumped us up to the crowd (“this has never been done before, we’re making history here!”) here’s the embarrassing result: `` Sadly, at this point it was time to say goodbye to LPU and our hosts, head back to Amritsar for the night, and catch our flight to **the wedding (and our real performance)** the next day ## Pt 3/4: a wedding and a dance _(originally published on December 2, 2016 on dazuck.com)_ Now, the reason we were in India to begin with: the Indian wedding. Must have been a good friend to come all this way, right (hey — even from Thailand it’s a decent flight)? Nope, we knew all of 1 person at the wedding before this week and had only just learned the bride and groom’s names. **We were crashing a wedding** (with the groom’s blessing, but I’m still counting it). How was it? I’ll let this video — organized by our new best friend Pratham — speak for itself. _Note: Before you watch, know that my route at the end was actually planned — thanks for bailing on the choreography, Nik_ [https://youtu.be/jFRcS9k9Kt8](https://youtu.be/jFRcS9k9Kt8) ## Pt 4/4: no rupees mo’ problems _(originally published December 1, 2016 on dazuck.com)_ Okay, so now we’ve been shuttled in by Air India, around Amritsar by Pratham, and the wedding by Kunal. Is our group still capable of traveling when we have to do things for ourselves? I haven’t mentioned this yet, but while we were there the entire country was out of currency. This hadn’t mattered much up to now, but it was about to get pretty relevant. First up is the train trip to Agra to see the Taj Mahal. It’s a pilgrimage made by ~4 million people each year, so should be manageable. After the 1-hour Uber to the train station at 1pm, we weaved our way to the front to find the ticket desk and bounced from window to window, finally figuring out we had to fill out a slip from counter 5 (bookings) and then take it to counter 3 (reservations). Once there though, we found out that contrary to our hotel’s claim we couldn’t use credit card, and didn’t have enough cash. So Nik took my USD cash reserves into a random market a short ride away to trade a sketchy FX shop for more rupees; meanwhile Rob watched the bags while I wandered around the station looking for the girls who had taken a separate taxi and not yet turned up. This went on for an hour or so. In the meantime we found out about a tourist bureau upstairs, but they also didn’t take credit cards so while more comfortable, we were still missing cash and missing our full group and missing trains to Agra in the meantime. There were a few more left, but most were fully booked which meant we’d have to by a non-reserved ticket and hope to get a seat. Given how confused we were at this point, that seemed unlikely to go well. Around 6pm things started to come together. Nik got back with cash, we got word from the girls that they were here and had followed a man to a travel agency, and that we could get a van to Agra for $180. More than the train, slower than the train, less comfortable than the train. But we’d get there tonight and see the Taj in the morning. A bit of haggling and waiting later and we were on our way. 6 hours later (1am) and we were there in our sketchy hotel. I won’t waste much text on the Taj since plenty of others have, but here’s a picture to prove we made it. Okay, so that’s a monkey stealing a tourist’s snacks on the way into the Taj, which was pretty funny. But also the Taj was pretty damn grand. Post-Taj, I headed back to Delhi while Nik went for Jarpur and the rest of the crew flew home. I’d spend a couple days there, tour some education groups (post pending), then meet back up with Nik to head north to the Himalayas or south to Bombay for a week. Or, that was the plan. Instead, I spent my non-education time in Delhi trying not to spend ANY cash, since I was down to 180 rupees and every ATM was out or had a 3 hour line that would run it down before me. My quick guide on how to survive 3 days on 180 rupees in cash when most places are cash-only (~$2.75): - Piss off the hostel. Turns out if your laundry smells bad enough they let you do a load even if you can’t pay for it. - Dehydrate between visits to McDonalds, where you can buy water with credit card. - Uber. It’s has your credit card on file, which subway doesn’t take. Great, until your phone dies… - Walk a LOT. I walked 5 hours back to the hostel through Delhi because the subway was too pricey. This was great though (at least 3–4 hours of it) as I got to wander through random parks and under-freeway markets I never would have seen. - When you really get low, buy your flight out because trying to eek out another week in Simla could go from adventurous to painful. Come back another time, and go decompress in Delhi’s polar opposite (Singapore) for now. - Stop faux-complaining; about a quarter of the Indian population lives on that budget constantly, without credit cards and Uber as bailouts. ## Education in India URL: https://dazuck.com/writing/education-in-india Date: 2017-04-10 Type: travel Summary: Observations on education systems in India — what works, what's broken, and how it compares. During my time in India I was lucky enough to visit two very different Universities and a foundation focused on K-12 education. Brand new to Indian education, I tried to stay away from too-quick conclusions about the state of education or particular initiatives in that environment. But my observations and conversations at Ashoka University, Lovely Professional University and Central Square Foundation kicked off many of the questions I've been wrestling with, so I wanted to share what I've been seeing and thinking. **"Education" means something different in India than in the US — and it's too broad a term in general.** My university, Stanford, was informally divided into 'techies' and 'fuzzies,' with the elevated respect for techies mirroring the general re-emphasis on STEM in the US after a historically heavier focus on the liberal arts traditions. India, on the other hand, has a complete focus on STEM and almost no place for liberal arts in education. So the wisdom and outlook of graduates is necessarily different in the two places. Ashoka is a startup University modeled largely on US Ivy schools that sees the liberal arts as critical and is seeking to change the norms of tech-only in India by demonstrating the value of liberal arts to students, citizens and employers. They have a promising start, but also have their work cut out for them. Along with the massive challenges of starting any University, they themselves are — it seems to me — still learning the 'why' of the designs and programs they've copied from US schools. A direct imitation is efficient, but if they want to convince others of the value they need to really understand it first, and how each aspect does (or doesn't) apply to India. LPU, the other university, is built from the ground up purely on the demands and considerations for its specific situation in northern India. It offers no liberal arts at all because there would be no demand for them from its thousands of students. They do, however, have a super interesting entrepreneurial system where students more or less run the school. The on-campus hotel is part of the hospitality school, all cars and shops are student-started and run, large companies train students on how to run a franchise and then hand over the on-campus branch. It's driven by necessity — they growth and funding mean cheap student labor is the only sustainable option — but also a far more natural entrepreneurial training ground than the entrepreneurship centers at many US schools. There is obviously value in all these types of education. Sciences prepare you to reason and work; humanities prepare you to make choices and think creatively and deeply; entrepreneurship prepares you to persevere through the unknown. One argument for the slipping position of the US in the global economy is the lack of STEM; one argument for the lack of original innovation in China is a lack of liberal arts. Life should reflect a mix of action and contemplation, and so should school. It's interesting to see various countries struggle with how to value and prioritize this mix — it's easy to say 'do it all,' but time and funds are scarce so focus matters. **While the specifics differ, lack of quality teacher time with each student is a consistent challenge across borders.** There was a 'small class size' movement in the US in the past, and this is not what I refer to. I refer to the difficulty in establishing an orderly, focused classroom in which a qualified teacher can spend meaningful time coaching. In parts of rural India, teachers simply don't reliably show up. If they do, lack of materials, attendance and nourishment make it next to impossible to get a classroom settled and in a routine that can lead to consistent learning. In rural areas (observed in southeast Asia) there are often too few classrooms, so with too many grades mixed together and students shuffling between rooms the teacher is just trying to keep things under control rather than moving forward. While it often looks different in the US — chaotic hallways and noisy classrooms with teachers having little idea what each student is doing — the end-state is the same. Some data says the most important driver of learning is direct coaching time from a teacher. And yet, it seems that more $ and energy goes into training teachers on content or pedagogy, when simply helping manage the classroom (and school environment) is a critical first step that usually has not been solved in low performing schools. **As in all parts of life, incentives drive action and account for real difference.** In India, there is no PD budget so there is very little focus on helping teachers improve (on content, pedagogy, or classroom management). Central Square Foundation has found it necessary to shift to 'demand-side' work, creating the right incentives, rather than trying to build tools nobody has money to pay for now. On the other side, the US has allocated massive PD budgets so organizations that may not be 'PD organizations' develop PD offerings to tap into that pool of earmarked money. This brings up two questions. First, what are the incentives that we know are most important in education? With all the trials and measurement, we know a lot now, and incentives can be changed and this will make a difference, which is heartening. Second, should policy-makers be trying to drive certain reforms with incentives like this? The history of education reform, and really just history in general, is chalk-full of policy makers with good intentions (and 'the latest research') setting incentives for adoption of something that should help, but fail. This is, in my view, a huge practical and reasonable difference between progressive and conservative approaches: progressives pursue this in the rush to help and improve, while conservatives view this kind of actions skeptically because history shows how ineffective and disastrous they can be. Might we be better leaving funds and decisions more to 'the market' and localities? Or, better yet, decide what single incentive we care about in the end (say, 12th grade proficiency) and let the market orient around that goal rather than intermediate ones like class size or PD hours. **There is always a tough balance between trying to find, use and scale what is known to work versus empowering those on the ground who know local contexts** (as in many industries). Ashoka seems to be taking the approach of copying the US (to an uneasy degree — even building blueprints) and dropping that outside Delhi, and then over time tweaking what isn't working to localize it. Something feels wrong about copying without foundational knowledge of why things work as they do; but perhaps this is orders of magnitude more efficient and should be embraced. LPU is built from scratch, and while there are missteps the result is a pretty unique culture with students running much of the University, and getting more 'real' entrepreneurship training than copying another University's entrepreneurship center. In a world as big as education, better ways to identify and circulate 'what works' seems like it should be a top focus. I could see a "Give Well" organization purely for schools, to make the incentives of such an ecosystem work. ## Mourning the king: a country united URL: https://dazuck.com/writing/mourning-the-king Date: 2017-04-02 Type: travel Summary: Witnessing Thailand in collective mourning for King Bhumibol — what national grief looks like up close. _(originally published on November 26, 2016 on dazuck.com)_ It's 97 degrees and whatever the humidity is, it's sticky. We are outside the gleaming royal palace in Bangkok but cannot enter as the palace is closed this month. Still, ~30,000 Thais are waiting in black clothes. They will be there for hours. Standing, occasionally moving forward, waiting for their turn to see and pay respects to the body of King Bhumibol Adulyadej. Tens of thousands did this yesterday, and will tomorrow, and will every day for weeks. A mile away at a busy intersection traffic is buzzing and tuck-tucks are hawking and pedestrians are buying lottery cards from street vendors. In a flash too quick for us tourists to realize, a few police blew whistles and everything and everyone stopped. Why the sudden change? To honor of King Adulyadej, who died ~3 weeks ago. His picture is on every corner and shrine, black bows adorn most shirts, and festivals across the country are cancelled or curtailed. There's a hole in the heart of Thailand as they mourn their king. I won't pretend to understand or appreciate his fascinating biography (controversial crowning, 70 years of rule, military coups, near-divine status in Thailand), his importance to the country that is mostly run by the military, or the pros and cons of his rule of Siam. That context isn't required to clearly see how unified Thailand is in grieving the loss of their King. On my first days in Bangkok, I did two things: I went to the royal palace where thousands were waiting to pay respects to King Adulyadej, and I watched Trump's victory unfold. There is no claims of right vs. wrong here, but as pure observation the juxtaposition between Thailand's unity and America's division was more stark than any differences in food, traffic, religion or language. The US formalized its partisan rips, and then devolved further into attacks and anguish and anguish in the days that followed. Many of my friends voiced their disappointment with their country and its people, some in my state threatened to secede, and journalists professional and amateur seized on a new (though overdue) favorite topic: our country is more silo'd and less unified than we realized. Thailand spent that same month with cancelled or subdued festivals, daily respectful silences, shrines and photos and music everywhere, and a shared respect for a (mostly) beloved leader. The US could never rally around someone the same way — elections and parties make sure of that, and probably for the better. But right now it feels like our tends towards division and away from unity to an alarming degree, and in a snowballing way.If President Carter or Bush or Clinton passed away, would folks of both parties come together in respect of their service? Older generations say everyone can remember where they were when JFK was shot. It's not like he was universally loved, so what made the country come together so much more back then? Maybe I'm being cynical and we would still show some decency and unity today. Does it take a common enemy for us to come together, or can we do that without some external and existential threat? What I think is particularly interesting about Thailand's unity is not the nationalism — plenty of countries have more nationalism than the US, especially with more homogeneous cultures. Thailand is also so international, so welcoming, so open to being a global metropolis and tolerating difference and yet they maintain this fierce love and loyalty to their nationalist pride and it's head. Thailand has PLENTY of faults, so I'm not saying they are doing this or anything right that the US is doing wrong. What I wonder though is whether the Thai monarchy (which doesn't hold much 'official' power) is important just for being the one thing the whole country can rally around. So that even with growing diversity there can be unity. In the US we don't seem to have that anymore. We've fragmented. Our jobs, news sources, favorite sports, hobbies, threats and inspirations are so different. Diversity is good, and tolerance for difference should be celebrated. But we need something that holds us together as a single country too, and it shouldn't have to be an enemy. This could be many things (mandatory service requirement, a beloved national sports team, a scientific quest), but I'll argue in a future post that it used to be a more common conception of what it means to be an American and that our education system needs to bring that shared understanding back. ## Singapore: order URL: https://dazuck.com/writing/singapore-order Date: 2017-04-02 Type: travel Summary: Observations on Singapore's immaculate order and what it reveals about governance and culture. I can't imagine there's a much more dramatic juxtaposition of cultures than the one we created by flying directly from Delhi to Singapore. If India was like the summer camp with pre-teens running in crazed passion every which way, Singapore was the marching band practice three fields over. Nik and I brought with us from Delhi a flu, a parasite, no cash, and pounding heads, so it was nice to catch our breath in such a perfectly functioning city. And then it was nice to move onto somewhere a bit more interesting after a couple days. Singapore is order. Trains run on time and traffic flows freely. There is no jay-walking nor litter (including gum). Restaurants are clean and prices strict (it was so strange to not bargain). Crime is rare and scandalous, with criminals punished harshly. A tiny island country, Singapore is built to squeeze every ounce of potential out of its people through an incredibly demanding, effective and conformist education system. We were there a few days after the 6th grade placement test, when the country's 13-year-olds take a test that largely dictates their career track. Fathers routinely take that year off to help their kids prepare for the test full time. There was apparently scandal this year, as a subway was running a couple minutes late and forced some kids to scramble to make the test on time. This day — the day of a 6th grade standardized test — is apparently the biggest day of the year to Singapore. With the focus on education comes results. 98% of students pass, with 64% tracking towards University level. For comparison, the US is ~37%. After this test students are tracked towards careers that match their performance. The top students have costs covered to study at top international universities. Lower performing students are prepared for careers to make a living. Tracking has plenty of downsides, both practically and ethically, but the results and efficiency are clear and Singapore does a number of things to help keep the tracking from leading to runaway inequality. A country-wide mandatory service requirement does a decent job ensuring people from all backgrounds interact, and civil service positions are highly paid (benchmarked to the private sector equivalent) so there's strong incentive to serve the public. And the education system is relatively adept at helping late bloomers move up and realize their potential given their particular strengths — a developmental pediatrician I talked with has a patient who's autistic and was failing many classes, but ended up in a top magnet school that plays to his strength in mathematics. Such high demands and a systematized approach do lead to a very conformist culture. This can be seen in small daily quirks — the neat queues formed at every subway station, the ubiquity of brand-name shopping — and also crops up in important ways in the culture and economy. While a small startup scene is starting to emerge, Singaporeans seem to be much more adept in the corporate world, exceeding expectations at a defined roles rather than creating something from nothing. That's one hypothesis I heard from for the international openness of Singapore: the desire for foreigners to bring creativity, originality and entrepreneurship to the city-state that the more high-achieving but conformist locals can add to. Experiencing the function and conformity as a traveler made me start to think about what and where I'm most enjoying about travel. When I'm living in one place I like cities with lots to do and explore, but with some dependability to live my life somewhat chaos-free. But traveling, I want the opposite. Cities often are too overwhelming to really see in a few days; smaller areas I feel like I can get to know more completely. And I like some chaos when I'm on the road — it's what makes places interesting, different and memorable. Of course, those thoughts tempted fate a bit too much since I got my bag stolen in Malaysia the very night I left Singapore… ## How to travel the world THEN give up all your possessions URL: https://dazuck.com/writing/travel-give-up-possessions Date: 2017-04-02 Type: travel Summary: On the paradox of traveling with everything you own and coming back wanting nothing. _(originally published on December 10, 2016 on dazuck.com)_ I'm now ~1 month into hopping around Asia, and losing my bag in Malaysia has forced me to confront the mounting evidence: I might not be so good at this whole travel thing (yet). So, instead of a list of 'pro tips' I've put together a series of my early missteps learning opportunities for travelers from a bumpy-but-still-phenomenal first month: - **Day 0: this is too late for detailed budgeting.** Vaccines, meds, delayed credit card bills, insurance, goodbye drinks…if you don't plan well, your budget ends a lot smaller than expected - **Day 1: don't forget to sweep the bathroom.** Always nice when you leave your toiletries behind before your journey has even started. - **Day 3: don't book anything ahead of time…** That just locks you into plans when better things might (will) come up, and very little ever books up in a way that shuts you out of something important. - **Day 4: …except when there's a festival.** When there's a world-famous event, okay yes then you should book and not be stuck on a roundabout bus instead of the well-regarded sleeper train. - **Day 7: don't screw with budget airlines.** That moment when you show up 58 minutes before departure to find out the counter closed 60 minutes before? And when that flight included a connection? And then the one you get on the next day is delayed 12 hours (another missed connection) and has a near-mutiny onboard because the AC doesn't work? Yeah, not worth the extra 15 minutes of sleep, no matter how early the flight. - **Day 9: disorganized stuff = lost stuff.** Without a packing system, hats and shirts and eyemasks disappear at random stops in a hurry - **Day 16: when you hear a country has a cash issue, plan accordingly.** Otherwise you end up trying to stretch 180 rupees (~$2.50) across 3 days of travel through India and booking an early flight out. - **Day 19: don't just buy a cable lock, USE IT.** Otherwise you might wake up at 5am in Kuala Lumpur to transfer buses to find your bag has vanished from the checked luggage compartment. Silver lining: if you keep your most important stuff on you (passport, wallet, phone, computer) then everything else doesn't actually matter that much - **Day 25: don't count on island time.** When you think about booking an 8am flight when the optometrist has promised your contacts will be delivered the previous evening, trust that instinct that it's a stupid idea. Or be prepared to travel blind a bit longer. - **Day 30: check weather.** Or you might be navigating the most beautiful city in the world by boat instead of scooter, with all the shops shut down from floods. So, plenty to learn from in the first month, and things are starting to go smoother now. More importantly, while there have been quite a few bumps early on they haven't come close to defining the ride so far. They are inconveniences in what has been an amazing start to the journey. ## Education: refocusing on learning goals URL: https://dazuck.com/writing/education-refocusing-learning-goals Date: 2017-03-20 Type: essay Summary: Why education reform should focus on learning goals over structural categories. _(originally published in March, 2017 on dazuck.com)_ When I moved into education a few years ago I was quickly struck by how huge that world is. The sheer number of dedicated people and well-intentioned organizations trying to improve US education is inspiring, and I tried to digest all of it. I was of course overwhelmed, and over time developed a concern that the breadth was, while inspiring, also inefficient and likely hiding problems (and solutions) that become hard to identify in such a sprawling landscape. Certain sub-industries had created 'market maps' that helped their focus. For example, this CB Insights map of ed tech companies gives investors and technologists a way to think about how their skills and efforts fit into the ecosystem. Similar (though flawed) maps exist for higher education, charter systems, and even general reformers. However, I've found nothing that works if your goal is to question education from a wider starting point. **Many of us want to figure out how we can best improve our society's education. But 'education' is too big, and the subdivisions we typically jump to after that ('math', 'tech', 'teaching') are too narrow and disconnected from broader goals. I want a way to be more targeted and reasoned when addressing the hard questions about our learning ecosystem.** A few examples of how education being so broad causes problems today: - Lack of adequate distinction between subjects has led, for example, to attempts to evaluate and pay Kindergarten art teachers based on performance, because that was deemed the right policy for subjects like middle school math. - "Common Core" has become a politicized lightning rod as a symbol for centralization; the reality is some things should be standardized, some should not. But with no way for non-experts to mentally separate various parts of the school-day it's not surprising people would react with hostility to apparent mass standardization. - Those working in "education" deal with an informational firehose with no easy way to filter it. When I tell people I worked in ed tech (at Zearn, a K-5 digital math company), they often respond referencing MOOCs, coding platforms, or adult language learning programs. These have very little in common. But without a more specific framework we're left grouping them altogether and get too overwhelmed by the magnitude to follow news effectively. **This article is an attempt to set up an education framework by mapping learning goals to how they are achieved in a way that can make policy, content, product development, and other considerations of the whole ecosystem — especially schools — more approachable.** This not a fully-baked proposal, nor does it touch many critical aspects of improving education (especially anything related to the structures of the school system — teacher pay/support, unequal funding, etc.). It's a starting point to improve upon my own fragmented understanding, a thought-experiment on the goals and structure of education in 4 parts: 1. Section 1 looks at the history that created today's conflated world of 'education' and why it's problematic 2. Section 2 proposes learning goals as the core to an education framework, mapping education categories to the purposes they serve for students and society 3. Section 3 evaluates how education categories differ on today's most important questions in education policy and reform 4. Section 4 illustrates a hypothetical school and education system built on these goals, categories and insights "Learning," "education," and "school" are often used interchangeably. They are meaningfully different though, and my hope for this exercise is to clarify how they relate to each other. **Society has many 'learning' goals it should pursue; its 'education' system supports these learning goals that don't happen independently; 'school' is the critical, state-funded institution within the education system.** Not all learning goals should necessarily be part of school, and some likely are today as a result of inertia even if they could (should) be handled in other ways within the learning ecosystem. We'll explore why historically first, and then how to think about things more explicitly going forward. ## How we got here The fact that the US education system has unclear boundaries is unsurprising, as disagreements about the primary goal of the school system have been present since its founding. Successive periods that focused in turn on knowledge, morality, skill, liberal arts, and character have left us with a sprawling system and terminology. Each part may be valuable and deserve its place, but the way they are integrated (or not) today comes from leftovers of many individual programs more than any explicit decisions on what stays and what goes. Early community-run school-houses were mostly one room with dozens of children, all ages, focused on memorization of certain facts and some literacy. Schooling in a more standard and scaled way came about through 'normal schools,' in which Catherine Beecher's corps of female school-teachers were coached to be 'mother-teachers,' raising children morally to participate in democracy. Horace Mann, the first state secretary of education (in MA), also prioritized morality and preached _"good-will towards men, and upward in reverence to god"_ above liberal education, which he thought was a waste. With these anti-intellectual roots, carried out by newly trained female teachers, the country's network of schools grew and spread west. The turn of the 20th century brought a more modern pedagogy focused on skill and personal development for children, ushered in by John Dewey and his disciples. These goals called for more student-led intellectual learning, and the role of the teacher and school necessarily changed (and broadened) as they now were asked to help students learn how to think, grow, and 'learn to learn' rather than absorb the facts and moral codes being recited at them. Then in the middle of the century, war years (among other factors, including some backlash) pushed priorities back towards morality with a focus patriotism, along with classes on "useful skills" like home economics and machine shop. Throughout the 20th century as budgets grew, professional teachers and their unions gained influence, and education became near-universal, education policy became more visible. Debates around tracking students and vocational programs emerged among others, and big questions about who should control policy and content gained urgency when the 1965 Education Act put the state front and center for the first time. Over the century the school system also absorbed initiatives, reform movements and debates over (not comprehensive): health and nutrition (free lunch programs); sports and other extracurriculars; sex-ed; counseling and school/career planning; revitalizing STEM; character (grit and growth mindset), and others. This is before even getting to the central role education plays in questions of equal opportunity, with implications in civil rights law (busing, integration), taxation, school choice (charters, voucher programs), and others. All these debates, programs and goals sit on top of the core learning objectives that were the sole focus just 150 years ago: basic morality, social behavior, critical knowledge acquisition. (A similar confusion of purposes is behind some of the debates about higher education today, as our current institutions maintain the liberal arts traditions modeled after the English aristocracy but have also heavily adopted the German research university model. This leads to an uneasy balance between workforce-ready engineering and broader liberal arts studies, with the lack of role-clarity made more apparent by unanswered questions about the role of internet/remote classes.) **None of this suggests any of these are unimportant, or that they should not be integrated into our school system. However, many pieces have been added ad-hoc, each to serve its own purpose.** And when each new piece is added (whether because of new research, or politics, or fads) it's often patched on rather than integrating with (or clearing out) old ones. This cycle is understandable (why is a whole separate topic). It is also a snowball — it's easier to keep adding weight to a truck than a scooter. Like an overly complex tax system, merits of each individual policy may start to be outweighed by the crushing weight and confusion of a now-jumbled system. **So whose learning goals should we be solving for? Students' and society's.** Children are the primary stakeholder. Since a society's reason for existing is to help its citizens thrive and survive, and in today's world that requires an education in addition to security and rule of law, some basic level of education is an individual right. Parents and teachers are not themselves stakeholders. Both have a right to input as proxies and the most informed citizens, respectively, but are not themselves stakeholders with their own learning goals. Society itself is also a stakeholder. If it is investing in each citizen, it should also consider the value education is providing for the broader society (to be more specifically defined later). School in the US and abroad historically had plenty of focus on morality, citizenship and character. **And with good reason: we live in groups, so students need to learn to think about themselves as part of a group as well as capable individuals.** The best historical predictor of the fall of a society is the decay of morals and character — individuals forgetting what it means to think of and act as part of a group. Learning goals must include fostering that (especially in the US right now). ## Starting with learning goals What are the key learning goals of our two stakeholders, students and society? We'll evaluate the priority of these later; for now my summary of the goals that our education system tries, at least in part, to accomplish: Student goals: 1. Foundational skills needed for adulthood 2. Ability to reason and problem solve independently 3. Practical skills to be applied to work/home life 4. Intellectual capacity to explore, question, grow as a person 5. Basic moral code (including the norms of the society) 6. Character strengths 7. Knowledge in subjects of students' own election 8. Physical growth and outlet (sports, recreation, etc.) Societal goals 1. Individuals that observe rule of law, uphold social norms, do not disturb security 2. Families that support themselves without need for external support 3. Workers with skill that add value to the economy 4. Citizens that participate civically, casting informed votes and otherwise 'owning' democracy 5. Communities that maintain national pride + unity while also nurturing their own strengths + values **Some may be tempted to dismiss the societal goals as softer or less important, but this is dangerous and exacerbates the problems of recent history.** Student goals seem more familiar and so 'right' because we have skewed that way in recent times, as specialized knowledge and competitive jobs pushed individuals (and so schools) towards more self-serving collection of intellect. Even if you disagree with the goals above — which should be challenged — the point for now is to examine how the lens of goals can be applied to the many things we are working on now. If a middle school math teacher is trying to become more impactful outside that subject, or a state superintendent wants to strengthen her state's moral character, or a tech billionaire just 'wants to make the world better,' how do they decide what to do? They could start by seeing how on-the-ground initiatives (classes) connect to learning goals. The 7 education categories in this diagram (which don't include enablers like nutrition) could be debated, bucketed and mapped differently of course, but this is my take: The categorization and goal-mapping allows us to see which goals are served by which programs. We can quickly see that if we are looking to improve society's civic participation that our levers are specialized classes, humanities and character development. This mapping helps us see options and trace connections, and when we use these categories to look more finely at the more complicated questions in education we start to get a better picture of how best to achieve our learning goals. ## Education issues by category The goals served by various categories of education are different. So what? This only matters if our approach in how we prioritize and deliver each category, in pursuit of the goals they each serve, should differ meaningfully. There are dozens of important questions in education debates today, many of which contain (very sensitive, political) implications about centralization vs. decentralization. Some particularly pertinent ones: - What topics should be required (standard curriculum) or optional (privately governed schools)? - Who should decide on the specific content taught and pedagogy employed? - How is this learning be efficiently delivered (tech, large group, small group)? - Where does funding come from, and how much for various topics? - Should success/achievement be measured for accountability to standards? - What, if any, effective mechanisms to use to incentivize and if necessary replace stakeholders who don't measure up to standards? **These hard questions, with trade-offs both ways, should not be answered on a whim or all at once, or the same for 'all of education.'** They should be examined based on the dimensions that matter and the trade-offs they each imply. For example: whether a given topic is best delivered in small group, large group, or digitally is a function (in part) of how much coaching is needed and how important peer-to-peer contact is to the learning goal. I've done this with some example questions for each issue above and ranked each category. Obviously the categories vary dramatically. Some quick conclusions that you could draw: - **Core topics should be required and highly standardized** — required, measured, funded — so every student gets this in high quality, but doesn't necessarily require in-person/high-attention classes. - **Humanities should be prioritized** everywhere, but the content and standard **specifics could be left to localities** - On the other hand, **critical thinking topics should be more standardized** to leverage the more objective and researched nature of the subject - **Character and humanities have the most need for direct instruction,** so high demands on teacher skill and time - Meanwhile **specialized, practical and even some core skills could be done through other avenues** that are cheaper, more efficient and less burdensome to schools - **Practical and physical education could be distributed**, letting centralized focus and funding going to more standards-intensive categories - **Specialized topics could be done primarily outside the centralized system** despite standardization benefits due to low priority, funding and direct delivery scores One way to organize the impact of this analysis further would be to group priority + delivery + funding together to indicate whether each category should be in standardized schools or elsewhere in the education system, then adding content + accountability + measurement to guide whether decisions on each subject should happen centrally or locally. Mapping each category on these two dimensions starts to suggest a picture for the structure of the education system. These of course are over-generalizations. Digging deeper into the categories and the sub-questions, we can see some nuances that are important to address: - Humanities: there's a large gap between high societal benefit (2) and lack of standardization benefits; how to ensure the social benefit without the standards? - Character: high priority and funding, but how to effect that without standardization, agreed upon norms, or established classes in the subject? - Specialized: while lower in priority and funding early, some specialized subjects may be critical for students' success in later education in more specialized higher education - Practical: some of this experience is needed to prepare students for action along with contemplation, putting ideas to work not just theoretical consideration The half-answer to all of these is that neither decisions nor implementations are all or nothing. This structure is a starting guide, and the design of the system needs to reflect more nuanced realities. That is not to disregard the data and framework; rather, we should continue to look at the sub-questions and multiple goals served by each category rather than rely solely on the top-level summary. The next section roughly illustrates a system that would orient education more directly around these principles. ## Illustrative, stripped down ed. system This exercise has a bias towards decentralization. There are plenty of extremely important goals and initiatives that should be part of the school system, but I'm not too worried about that: there are also plenty of voices lobbying to do and add more in schools. This exercise is meant to balance out the voices trying (with good intentions) to add with a critical eye towards removing what has lingered in the status quo with complicating negative consequences. **It follows the principle of subsidiarity: do things at the lowest, most decentralized level they can be efficiently done.** This unburdens teachers and schools so they can put more focus and budget on the highest priority responsibilities; it maintains room for experimentation and growth; it makes the system anti-fragile; and it gives agency to communities who clamor for it (so is good politics, even if you err on the side of more 'expert' control in education). Coming back to the dimensions from earlier, we can see a starting structure. **School would focus on core skills, critical thinking and humanities, as well as character development,** with the first two more standardized across schools, drawing on the best research available to make sure each student gets the most helpful preparation, which can then be tested to incentivize and hold accountable schools and teachers for this education. Humanities — for which testing is harder, general exposure most important, and values diverse — would be left more (though not completely) up to localities. Specialized, practical and physical education could happen more outside of school through other institutions; for example, community publications with journalism classes, vocational training supported by employers, and sports leagues respectively. These are of course generalizations for still-large categories.. Within humanities you may have some nationally mandated (and funded) portions of the curriculum to ensure societal goals like unity, while leaving other parts more distributed. Many programs in the supporting categories (counseling, health) — for example free lunch — obviously may coordinate closely with school systems as that is the institution already in place to reach students. But they could be operated outside the direct purview of principals and others already over-burdened with all that is required to deliver academic education. Similarly, this framework doesn't suggest that specialized education has no place; just that the focus of our scaled school system with standardized testing, pay-for-merit policies, and mobile teaching force should be foundational skills plus the critical thinking and humanities every student needs. Dance classes, Model UN, sports and agriculture classes are all great, but those can be decided and run (and have funding allocated) more locally. Some types of practical education may make it into some schools, but things like vocational training should be left more to employers or employee training organizations (even within the government), rather than thought of as part of the standard school system. This structure would keep each institution closely aligned to — and accountable to — the learning goal it is tasked with developing. Principals and teachers would know the areas they are responsible for, which would be the focus of their energy. Sports leagues and clubs would be supported and run separately, providing a different benefit and so operated under a different set of norms and policies and debates. And when people — be it the math teacher, state superintendent or tech billionaire — are aiming to improve education, they would be able to focus more quickly on the right goals and relevant levers when pursuing training, funding, policies and reforms. This stripping down of the school system is a broad and theoretical vision, based on a series of logical steps from learning goals to education categories to my subjective ratings. I am not advocating that the school system should be stripped down quickly, or even at all. I am advocating for more critical evaluation of education policies. Evaluation that explicitly connects initiatives to learning goals, and that acknowledges side effects of bloat and complication when doing too many things at once. I am advocating for experiments — whether in new schools, radical communities, or international projects where less exists now — that could test how more distributed models might work. Even if the basic system doesn't change, we can do more to help our educators and ourselves know what we are supposed to be focused on, and what our various programs are aiming to accomplish. --- # Projects ## Sous URL: https://dazuck.com/projects/sous Date: 2026-08-11 Type: project Blurb: The cooking companion I want in every home kitchen, calm and on your side. A working prototype runs on our counter, and we cook from it every week. Any recipe is now a search away, and an AI will write you a new one in seconds. What a home kitchen still lacks is a companion: something that knows what your family likes, gets you through the cook with wet hands, and remembers how you made it last time. Most cooking apps answer with more scrolling, more pop-ups, and more screens to manage. Sous is my attempt at the opposite: **the ultimate home cook companion, and a calm one.** ## The prototype on our counter The prototype is a working app running on a Pixel Tablet sitting on our kitchen counter. ![Sous on a tablet: prep boxes ticked off, then cooking steps one at a time, with a timeline of the whole cook beside them](https://dazuck.com/images/assets/sous/sous-cook.webp) *Cooking from the prototype, shown with demo recipes. Prep comes first, then each step with what's coming next. "The whole cook" on the right shows which steps overlap, with a line marking where you are.* Recipes come in by message: I send one to our kitchen agent, and it lands in the app's queue alongside our family's own recipe book. Before we start, Sous turns each recipe into a cook plan. It splits prep from cooking, pulls out anything that has to happen ahead of time, and lays out which steps run in parallel. ![Sous prep screen listing what to do ahead: soften butter an hour before, warm eggs thirty minutes before, chill dough at least two hours before](https://dazuck.com/images/assets/sous/sous-ahead.webp) *The part a printed recipe buries in step 4: what needs doing an hour before you start.* It works from the documents our family already keeps instead of a database of its own. Our family recipe book, a shared Google Doc, syncs into the app's library, so our favorites sit next to whatever came in that week. A second doc, Sugarface Kitchen, lists what's in our fridge, freezer and pantry, along with notes on how each of us eats. When I ask Sous something mid-cook, like what to use instead of the cream, it answers from what we actually have, those notes, and how we rated what we've cooked recently. Cooking from it every week keeps reshaping it. The first voice version listened all the time, and on the tablet that meant a chime every few seconds and no answer to "Sue". Now you tap once and talk. Most changes look like that: small, and found at dinner rather than at a desk. I'm using this for almost all of my cooking, getting my phone and laptop away but letting me have the creativity of cooking with AI that I love. There's a bigger vision for how this can expand to really end to end help make our family meals calmer and better, which I'll share more on later. ## Taste Agent URL: https://dazuck.com/projects/taste-agent Date: 2026-08-06 Type: project Blurb: An experiment in holding my judgment in one place that other AI agents consult, built on a record of my calls that outlives any model. This project is testing out a light non-consensus thesis I have: that in a multi-agent future, we'll all want an agent dedicated 100% to understanding 'us' and injecting that into every workflow with other agents. Other agents will specialize in tasks, verticals, projects, types of work, etc. based on their harness and training. And they'll have access to our preferences, memory, data, etc. in many cases. But the thing that differentiates 'me' from anyone else, more than anything else, isn't actually captured well in those discrete files -- it's the taste, judgement, patterns and nuanced day to day things that can only be inferred from a large amount of experience and is best captured in weights, not in files. This project experiments on that thesis that **taste should live in one deep node that other agents consult**, the way you'd ask the colleague whose judgment you trust rather than hand everyone a style guide. It started as a note to myself: "make a model that is purely embodying my taste - find right tech or way to encode this (maybe not text) and store it. Use as a way to learn a level deeper. And make a partner everywhere." I didn't start with a model right off the bat. For now it's a thin agent storing a record of my judgments: what I approved, what I changed, what I rejected, and why. It manages, updates, prunes and accesses that record and interfaces with other agents in my workflows to see if having that dedicated expertise improves outcomes and alignment to 'my taste.' The first experiment captured judgment moments from my own sessions with coding agents and used a small local model to sort out the ones that mattered. The lesson was about order: capture with nothing using it stalled, so I flipped it and started serving judgment first. Now, before an agent shows me a draft or a plan, it asks the taste node, which critiques the work against rules I've written in my own words and cites the rule it's applying. When I override it, that becomes the next example to learn from. It also has a removal rule built in: if it goes unused for two weeks, or I overrule it every time, the agents stop asking it. In time, if this succeeds, I'll move to a LoRA adaptor or even post-training my own open weight model to really encode this more deeply. For now, it's active and changing week to week. The open question is whether a few hundred of one person's judgments can teach a model something a general one misses. ## Abuzz URL: https://dazuck.com/projects/abuzz Date: 2026-05-20 Type: project Blurb: A virtual workplace prototype for small remote teams where humans and AI agents work as one — and the shared output, not the chat thread, is the home. Most tools today put inputs first — chats, docs, calls — and treat shared knowledge as the residue that's left over. Abuzz inverts that. The walk-in surface is **what we're making**, organized as a bento of live outputs. Conversations, drafts, huddles, and agent contributions feed those artifacts and are surfaced as their current best state. The one-liner I keep coming back to: _knowledge is the home; the room is where it gets made._ The prototype is a single self-contained HTML file (no build, no deps — open it and it runs) covering three surfaces: Home (the output-first walk-in), a Space called Memory design (a bound unit of work where a huddle is live, an artifact is being built, and three teammates plus a Cartographer agent are in the room), and Ask (⌘K synthesizes answers with citations to the primary source — voice call, doc, canvas note — not links). Agents (Cartographer, Researcher, Scribe) live inline with humans, distinct only by dashed circles and mono uppercase labels. No "✨ Ask AI" buttons, no segregated AI zone — they're teammates, not features. The design tenets matter more than the surfaces. **Calm over dashboard** — aim for Tuesday morning office arrival, not command center; desaturate and remove motion before adding. **Output-first** — inputs are modalities, not the center of any view. **Presence is grouped, not floating** — compact avatar stacks, no Zoom-style bubble swarm. **Live modalities are first-class but ephemeral** — they can be opened, joined, archived, retrieved, and they sit beside the artifact layer rather than in front of it. Most "I should add X" instincts have already been considered and rejected; the prototype is supposed to feel quiet. The interesting part isn't the mockup — it's the bet underneath. Five-to-fifteen-person remote teams already have voice, video, docs, chat, and a growing pile of AI features bolted onto each. Abuzz is testing whether you can collapse those into one surface where the agents are teammates and the artifact is the room. If it works, the daily experience of small remote teams becomes less bureaucratic, more alive, and more like sharing an office — which is the actual job of a workplace, and the thing we lost when we went remote. ## The AGI Incentive Map URL: https://dazuck.com/projects/agi-incentive-map Date: 2026-03-20 Type: project Blurb: Gemini Deep Research project mapping who predicts AGI is imminent, who doesn't, and the financial incentives behind each position. A Gemini Deep Research run cataloging 33 named figures across AI labs, academia, investors, skeptics, and policymakers — each tagged with their public AGI timeline, their stance on takeoff speed, and the financial exposure that shapes (or doesn't shape) their position. The point isn't to call anyone dishonest; it's to make the incentive structure visible so the substantive arguments can be read against the right backdrop. The output has five sections: - **The Correlation (n=33)** — the raw pattern between equity/fund exposure and predicted AGI timeline - **Three Positions, One Technology** — imminent-AGI bulls, skeptics, and the "powerful but bounded" middle - **The Money Behind the Claims** — equity stakes, fund sizes, and revenue exposure per speaker - **The Signal (Controlled for Incentives)** — what's left to learn from the loudest voices once you net out their book - **The Bottom Line** — the calibrated takeaway Embedded below. [agi-incentive-map.html](https://dazuck.com/images/agi-incentive-map.html) ## Daclaude URL: https://dazuck.com/projects/daclaude Date: 2026-03-06 Type: project Blurb: A personal AI agent built on Claude Code skills that manages email, notes, finances, tasks, and content — and improves itself through use. A personal AI agent system built on ~30 Claude Code skills that orchestrates daily life: email triage, daily recaps that synthesize activity across tools into a structured journal entry, weekly reviews that process a notes inbox and recommend what to write next, content digests from newsletters, a quest system for task management, finance coaching with live data from personal finance tools, and more. No custom code — it's prompt engineering, a notes app as persistent state, and tool integrations for external connections. The interesting part isn't the automation list. It's that the system improves itself through use. Each skill is a SKILL.md file — the prompt IS the codebase. When a skill doesn't work well, you edit the prompt, and the system gets better. The weekly recap skill was upgraded mid-session during a review to include content recommendations for my personal website. The handoff skill captures learnings at the end of each session. The adversarial-verifier catches mistakes. Shared context flows through auto-memory (cross-session learning) and episodic memory (conversation history search), so insights compound. The meta-lesson: building personal tools with AI agents is the best way to develop intuition for building AI products. Every limitation I hit with Daclaude — memory management, tool coordination, auth brittleness, the gap between "works in demo" and "works daily" — teaches me something about agent architecture that applies directly to the products I build at work. ## Finance Coach Skill URL: https://dazuck.com/projects/finance-coach Date: 2026-02-25 Type: project Blurb: Claude Code skill that turns personal finance data into a forward-looking dashboard — projections, scenario modeling, FI tracking, and investment thesis workshops. A Claude Code skill that connects to personal finance tools via MCP and adds what they can't do on their own: forward-looking projections, scenario modeling, financial independence tracking, and investment thesis stress-testing. Five modes — financial picture, projection ("what if we buy a house?"), thesis workshop (five analyst personas debate an investment idea), meeting prep, and tax planning. The centerpiece is an auto-generated HTML dashboard with Chart.js visualizations. Key metrics up top, then a scenario explorer where you toggle between income, business growth, and market return assumptions and watch 10-year net worth trajectories shift in real time. FI progress with Coast FI milestones, forward-looking goal feasibility, and portfolio allocation tracking. A collapsible "Go deeper" section holds portfolio perspectives (five AI personas synthesize investment recommendations from actual data) and an interactive scenario builder. The thesis workshop mode is the most unusual piece — five personas (Macro Strategist, Market Mechanic, Fundamental Analyst, Wealth Steward, Challenger) each analyze an investment thesis in their own voice, then synthesize where they agree and disagree. The Challenger persona evolved from a generic Contrarian — instead of reflexive disagreement, it proposes falsification criteria for the thesis and names specific conditions that would prove you wrong. ## What we learned building it The interesting part isn't the feature list — it's what happened when AI met real personal finance data across seven iterative sessions. **Your assumptions are wrong, and AI will confidently build on them.** The household spending estimate was based on mental math. When we finally connected real financial data, the actual number was significantly different. For two full sessions before that, the AI had been generating elaborate projections, sensitivity tables, and scenario analyses — all precisely wrong because the input was wrong. The lesson: AI makes bad assumptions look rigorous. Always connect to the real data source first. **Plausible-looking projections are the most dangerous kind of wrong.** The projection engine had a bug where income continued forever — even past retirement age. The 10-year charts looked reasonable, so nobody questioned them. It wasn't until we scrutinized the actual year-by-year numbers that the missing retirement income cutoff surfaced. A chart that looks right suppresses the instinct to verify. The fix wasn't just the bug — it was adding a survival table ("money lasts until age X") that forces the projection to make falsifiable claims. **Research beats gut feel by a lot.** Spending growth started as a 2% guess, got bumped to 3% because "more conservative," then actual research (BLS Consumer Expenditure Survey, EBRI, Blanchett 2014) showed 3% flat was 56% over-provisioned compared to how real households actually spend across age brackets. The phased model (faster growth in young family years, declining in retirement) completely changed the survival analysis. The AI could access and synthesize the academic literature in minutes — but only when prompted to look instead of accepting the assumption. **The skill should route, not teach.** Early versions had enormous prompts trying to teach Claude tax law, portfolio theory, and financial planning principles. They didn't work well — too much context, too generic. What worked: structured routing (five modes with activation triggers), data protocols (query Monarch for X, compute Y, show Z), and references to external documents for stable facts. The skill is an orchestration layer that points Claude's existing knowledge at the right problem with the right data. Less instruction, more structure. **Multiple AI agents editing the same artifact creates pipeline gaps.** One agent added achievement icons to the dashboard template's sample data and renderer. A different agent later generated the actual runtime data — without the icons, because it never saw the first agent's work. The dashboard has a multi-stage pipeline (Python generator → JSON data → HTML template → JS renderer), and an edit at one stage doesn't propagate to others. This is the AI equivalent of a team communication problem. ## Fund Tracker Dashboard URL: https://dazuck.com/projects/funds-tracking-dashboard Date: 2026-01-31 Type: project Blurb: Real-time dashboard tracking fund balances across multiple trading venues, wallets, and DeFi positions — polling every 5 minutes with secure secrets management. Aggregates balances across crypto exchanges, prediction markets, and DeFi positions into a single dashboard. Turborepo monorepo with a Node.js poller and Next.js dashboard, both on Railway. The interesting bits: parallel-by-venue polling that handles different API rate limits gracefully, automatic DeFi portfolio tracking across multiple chains, and a secrets pipeline that keeps API credentials out of the codebase entirely. ## Recall Skills URL: https://dazuck.com/projects/recall-skills Date: 2026-01-31 Type: project Blurb: Shared AI agent skills and cross-cutting learnings for the Recall Labs team — standardizing how the whole team works with Claude Code. A shared repository of Claude Code skills that the whole Recall team uses. Core skills include /commit (structured conventional commits with reflection) and /learn (routing learnings to the right documentation). The goal is to make institutional knowledge compound: every session teaches Claude something that helps the next session, across the entire team. ## Quests (Family Task System) URL: https://dazuck.com/projects/quests Date: 2026-01-26 Type: project Blurb: Family task management in Obsidian with GitHub sync — tasks as markdown files with frontmatter, synced to apps via GitHub API. [](<>) Instead of yet another task app, family tasks live as markdown files in Obsidian with structured frontmatter (status, owner, due date, category). Obsidian Git auto-syncs every 5 minutes to GitHub, and apps like Sugarface read/write tasks via the GitHub API. The format is human-readable, version-controlled, and works with any tool that can parse markdown. ## Sugarface URL: https://dazuck.com/projects/sugarface Date: 2026-01-25 Type: project Blurb: AI-powered family assistant for meal planning, fitness tracking, and household coordination — built for exactly three people. Started as a holiday project to test a theory: building exactly what you need with AI is now better and cheaper than adopting professional products. Ten days and $50 in API costs later, my wife uses it daily. The real discovery wasn't the app — it was how easy it was to build a system that improves itself across sessions. Custom /handoff commands, project diaries, and executor-evaluator patterns turned Claude into a collaborator that got better every day. Became the basis for a published essay on self-improving AI workflows. ## Funding Rate Arb: Backtesting & Strategy Discovery URL: https://dazuck.com/projects/funding-rate-arb-backtesting Date: 2026-01-20 Type: project Blurb: Backtesting framework for funding rate arbitrage across perpetual futures venues — 100+ strategy variants tested through an autonomous Ralph Loop discovery process. Pure research phase. Built a backtesting engine, fetched 18 months of multi-venue funding rate data, and ran an autonomous "Ralph Loop" that iterated through 12 rounds of strategy discovery without human intervention. Tested 100+ strategy variants across different entry timing, venue selection, concentration, and exit approaches. Key finding: single-venue concentration dominated, and no amount of multi-venue sophistication beat simplicity. Five critical backtester bugs found via adversarial verification. ## Operator Superpowers URL: https://dazuck.com/projects/operator-superpowers Date: 2026-01-18 Type: project Blurb: AI-native business selling premium Claude Code skill packs for startup operators — the entire company is run by an AI executive team. An experiment in whether AI can build and run a real business, not as a stunt but as a genuinely valuable company. A founder builds the skills through real work; an AI team of five (CEO + 4 execs) handles marketing, sales, content, and operations. Real legal entity, real payments infrastructure. The question: can an AI executive team handle the operational surface area of a small business while the founder focuses purely on product? ## Operator Skills URL: https://dazuck.com/projects/operator-skills Date: 2026-01-15 Type: project Blurb: Open-source collection of Claude Code skills for startup operators — email triage, coaching, candidate screening, prompt improvement, and more. Skills are reusable instruction sets that tell Claude how to handle specific tasks. Instead of re-explaining what you need every session, you invoke a skill and it just works. Built these from my own daily workflows running a startup, then packaged them for other operators. The coaching skill alone replaced hours of "let me think through this" solo time. ## Jellyfish URL: https://dazuck.com/projects/jellyfish Date: 2026-01-09 Type: project Blurb: AI meeting guardian that joins Google Meet calls and says 'jellyfish' when someone talks too long or the meeting goes off-topic. One job. Just one. Join the team's Google Meet, listen via WebRTC interception, transcribe with Whisper, and when the conversation drifts or someone's been monologuing for three minutes, speak the word "jellyfish" into the call using TTS. That's the entire product. No transcription storage, no meeting summaries, no action items. The hardest part was figuring out how to get a Puppeteer-controlled browser to actually speak audio into a Meet call. ## Recall Secret Santa Bot URL: https://dazuck.com/projects/recall-secret-santa Date: 2025-12-17 Type: project Blurb: Slack bot for running a Secret Santa game with daily clues, signal detection, and prediction scoring — themed around AI agent alignment for a crypto x AI team. Built for the 20-person Recall team's holiday celebration. The game uses AI/crypto-native metaphors: daily clues feel like model cards, predictions are scored like alignment evaluations, and the whole thing runs asynchronously through Slack DMs. Had Claude brainstorm five narrative concepts; the team landed on "Agent Alignment Challenge" as the wrapper. ## Atlas: Proactive Context Intelligence & Drift Detection URL: https://dazuck.com/projects/atlas-mcp-context-intelligence Date: 2025-12-08 Type: project Blurb: Transformed Atlas into a proactive intelligence layer — drift detection surfaces messaging inconsistencies, gap detection finds stale knowledge, and agent learnings capture what the team's AI tools discover. The big pivot: Atlas went from reactive ("ask me things") to proactive ("here's what you should know"). The headline feature is drift detection — it watches team messaging, product positioning, and priorities across the company's core tools, then surfaces inconsistencies before they become problems. Added gap detection to find stale or missing knowledge, context bundling to give agents relevant context automatically, and an agent learning system that extracts structured insights from how the team's AI tools work. Deprecated half the original capabilities to focus on what actually mattered. ## Tech Ethics Timeline URL: https://dazuck.com/projects/tech-ethics-timeline Date: 2025-12-05 Type: project Blurb: Gemini Deep Research project mapping ethical and legal violations at the largest tech companies over time, scored by impact. A Gemini Deep Research run that built a longitudinal view of corporate ethics failures across big tech — antitrust, privacy, labor, safety, content, and AI-specific violations from 2000 through 2024. Each incident was scored by impact (severity × scope × duration × precedent) and plotted on a per-company timeline, with the bubble size scaled to the impact score. The headline pattern: a small number of companies carry most of the cumulative impact, and the distribution is not what most narratives assume. **Aggregate impact scores (cumulative through 2024):** - Meta / Facebook — **90** (Cambridge Analytica, content moderation failures, repeated FTC actions) - Google / Alphabet — **67** (antitrust on search and ads, multiple EU rulings, privacy) - Amazon — **45** (labor practices, antitrust, FTC suit) - Apple — **42** (App Store antitrust, supply-chain labor, privacy posture vs. practice) - Tesla — **34** (Autopilot safety, labor/discrimination, securities) - Microsoft — **33** (concentrated in two eras: early-2000s antitrust, then 2020s AI/cloud) - OpenAI — **26** (compressed into 2022-2024 — copyright suits, governance crisis, safety departures) - Netflix — **14** (lowest — content controversies but limited regulatory exposure) The Meta-vs-rest gap is striking. So is OpenAI's score for a company that didn't exist meaningfully until 2022 — it took less than three years to accumulate impact comparable to companies with 20+ year histories. Embedded below. [tech-ethics-timeline.html](https://dazuck.com/images/tech-ethics-timeline.html) ## Focus Time Reminder URL: https://dazuck.com/projects/focus-time-reminder Date: 2025-12-01 Type: project Blurb: Lightweight Electron app that runs in the background and pops up mindfulness-oriented notifications to help you transition between focus blocks. ## Just News URL: https://dazuck.com/projects/just-news-concept Date: 2025-11-13 Type: project Blurb: TikTok-style video journalism app delivering factual, balanced news without opinion or commentary — built as an investor demo prototype. Prototype for a video-first mobile news app targeting truth-seeking millennials. Swipeable short-form video with HLS adaptive streaming, onboarding flow, and a design system aiming for Apple-level polish. Built to demonstrate the concept to investors, not to ship to users. ## Candidate Screening Tool URL: https://dazuck.com/projects/candidate-screening-tool Date: 2025-10-15 Type: project Blurb: Automated candidate screening that pulls from an ATS, evaluates with Claude against job requirements and company values, and sends daily digests of exceptional applicants. Built to solve the "200 applications, 3 worth talking to" problem. Fetches new applications from the team's ATS, runs each through Claude for multi-criteria evaluation (job fit, culture fit, exceptional signals), and surfaces only the truly outstanding ones. Includes rate limit management, caching to avoid re-evaluating candidates, and automatic tagging back into the source system. ## Atlas: Knowledge Hub with Vector Search & Integrations URL: https://dazuck.com/projects/atlas-mcp-knowledge-hub Date: 2025-09-12 Type: project Blurb: Evolved Atlas from a simple knowledge base into a semantic search hub with pgvector, partnership research, Twitter/RSS feeds, PR reviews, and web search via Perplexity. Second phase turned Atlas from a lookup tool into something more useful. Added vector search for semantic retrieval across all company knowledge. Integrated web search, built a partnership research tool, added social and blog RSS feed monitoring, and gave it the ability to review PRs. The chat interface got native tool calling so it could chain multiple capabilities together. By the end of this phase, Atlas was the team's primary interface for company context — but it was still reactive. You had to ask it things. ## Recall Roadmapper URL: https://dazuck.com/projects/recall-roadmapper Date: 2025-09-03 Type: project Blurb: Strategic visual roadmapping tool that uses orbital physics to represent confidence levels — goals are gravitational centers and solutions orbit at distances based on how sure you are. A reimagining of how teams visualize strategy. Instead of Gantt charts, you get a spatial canvas where solutions orbit goals at distances representing confidence (0-100%). Three semantic zoom levels: strategic overview, tactical planning, and implementation details. Every state change captures the "why" behind decisions. Local-first, no account needed. ## Daily Content Summary (n8n) URL: https://dazuck.com/projects/n8n-daily-content-summary Date: 2025-08-10 Type: project Blurb: Automated daily digest that aggregates email, Twitter RSS, podcasts, YouTube, and web search into an AI-summarized Google Doc delivered via Slack. An n8n workflow that pulls from Gmail, Twitter (via free RSS — no expensive API), podcast feeds, YouTube channels, and Bing search, then runs everything through Claude to produce a personalized summary ranked by your interests. Outputs a professional Google Doc and sends the link to Slack. Costs about $1-2/month to run. The preferences system is the key — you define high/medium/low priority topics and the AI organizes everything accordingly. ## Atlas: Team MCP Tools & Knowledge Base URL: https://dazuck.com/projects/atlas-mcp-team-tools Date: 2025-08-01 Type: project Blurb: Internal MCP server giving the team AI-powered access to company knowledge, daily summaries, and content generation from project management and communication tools. The first version of Atlas was a straightforward MCP tool server — a way for the team to ask Claude questions about company context without digging through docs and chat. Built the knowledge base ingestion pipeline, daily summary combiner, and a chat bot so anyone could query it. Content generation tools helped draft tweets and blog posts using the company voice. The key infrastructure decision was a dual-interface architecture (desktop and chat) that persisted through every later iteration. ## Rome Trip Planner URL: https://dazuck.com/projects/oaxca-trip-planner Date: 2025-06-05 Type: project Blurb: Zero-backend, offline-first PWA for coordinating a family trip to Rome — works on the plane, syncs across phones via manual export/import. Built because every trip planning app requires an account and internet. This one runs entirely in the browser, works offline (including on the plane), and syncs between family members' phones through export/import. Mobile-first, installable as a PWA. ## AI Memory & Rules URL: https://dazuck.com/projects/memory-repo Date: 2025-05-04 Type: project Blurb: Early experiment in persistent AI memory — rules and context files for maintaining state across AI conversations. One of my first attempts at making AI interactions compound over time. Predated Claude Code skills and CLAUDE.md patterns. The core idea — that AI should remember context between sessions — eventually evolved into the project diary and /handoff skill approach. ## Couple Concierge URL: https://dazuck.com/projects/couple-concierge Date: 2025-04-17 Type: project Blurb: Personal AI agent for managing, summarizing, and surfacing opportunities for you and your partner — an early experiment in agentic email and memory. One of my earlier agent experiments. The idea was an AI that reads your email, remembers context in SQLite, categorizes incoming items, and surfaces what matters to both partners. Learned a lot about import patterns, memory databases, and the gap between "agent that reads email" and "agent that actually helps." ## Recall MCP Server URL: https://dazuck.com/projects/recall-mcp Date: 2025-04-09 Type: project Blurb: Model Context Protocol server for interacting with Recall's blockchain storage — accounts, buckets, and objects from Claude Desktop or Cursor. ## Telegram Summarizer URL: https://dazuck.com/projects/telegram-summarizer Date: 2025-04-09 Type: project Blurb: Simple bot that summarizes unread Telegram messages once a day using the MCP server — because nobody has time to read 200 group messages. ## Getting AI to Improve Itself (Essay) URL: https://dazuck.com/projects/getting-ai-to-improve-itself Date: 2025-01-08 Type: project Blurb: Published essay on how a custom /handoff command and project diary turned AI collaboration into a compounding system that gets better every session. The essay that came out of building Sugarface. The core argument: building with AI is easy now, but the real unlock is building systems that improve themselves. Two specific tactics made the difference — a /handoff command that institutionalizes session wrap-up, and an enforced project diary that captures decisions and learnings. Between the two, agents started working, learning, and improving together over time. ## Ceramic URL: https://dazuck.com/projects/ceramic Date: 2021-01-15 Type: product Blurb: Scalable data infrastructure for user-empowering, natively interoperable data — the protocol layer beneath 3Box Labs' mission. A decentralized data network that enabled applications to build on composable, user-controlled data streams. Ceramic replaced the earlier 3Box.js approach with a fundamentally more scalable architecture — event-sourced data streams anchored to blockchain timestamps, with IPFS for content-addressed storage and a custom consensus mechanism for ordering. As CEO of 3Box Labs, I led the company through Ceramic's full lifecycle: protocol design, network launch, developer ecosystem, and eventually ComposeDB — a developer-friendly GraphQL interface that made Ceramic accessible to mainstream web developers. The ecosystem grew to include hundreds of applications and a vibrant open-source community. Key milestones: Ceramic mainnet launch (2021), ComposeDB launch (2022), integration with major Web3 infrastructure including IPFS, Filecoin, and The Graph. The protocol pioneered concepts like composable data models — standardized schemas that allowed different applications to read and write the same data structures, creating genuine data interoperability without requiring applications to coordinate. ## 3Box.js URL: https://dazuck.com/projects/3box-js Date: 2018-06-15 Type: product Blurb: User-centric data and identity SDK for Ethereum — adopted by 100+ early dApps including Uniswap, Gitcoin, and Aragon. An open-source JavaScript SDK that gave Ethereum applications user-controlled data storage and identity profiles. Instead of every dApp building its own user system, 3Box.js provided a shared identity and data layer where users owned their information and could carry it between applications. The SDK was adopted by over 100 early Ethereum applications including Uniswap, Gitcoin, Aragon, and dozens of DeFi and DAO tools. It handled encrypted private data, public profiles, messaging threads, and cross-application data sharing — all anchored to a user's Ethereum address but stored on IPFS and OrbitDB rather than any centralized server. 3Box.js proved the demand for user-centric data infrastructure in Web3, but also revealed the limitations of building on IPFS for production workloads — performance, reliability, and query capabilities all needed fundamental rethinking. These lessons directly informed the design of Ceramic. ## Number Gym URL: https://dazuck.com/projects/number-gym Date: 2016-05-15 Type: product Blurb: Math fluency and number sense product suite at Zearn — first to implement new Common Core standards, used by millions of kids across the US. A product suite for building math fluency and number sense at Zearn, one of the most widely-used math platforms in US elementary schools. As PM, I led the creation of Number Gym — a set of interactive practice experiences designed around the new Common Core math standards. The key insight was mapping the standards into a progression that made mathematical sense, not just a checklist. We created proprietary IP about how the standards built on each other — which concepts were prerequisites for which, where the common misconceptions lived, and how to sequence practice so students built genuine number sense rather than just memorizing procedures. Number Gym was the first product to fully implement the new Common Core standards for math fluency. It reached millions of students across the US through Zearn's school partnerships, making it one of the highest-impact products I've worked on by sheer reach. ## Bubbles URL: https://dazuck.com/projects/bubbles Date: 2015-06-15 Type: product Blurb: iOS game that taught kids multiplication through play — popping number bubbles to build fluency. PM at a NYC edtech startup. An iOS game designed to make multiplication practice feel like play. Kids popped bubbles with numbers to solve multiplication problems, with adaptive difficulty that met each student where they were. Built as PM at a small NYC edtech startup focused on elementary math. The game used spaced repetition principles — facts a student struggled with appeared more frequently — wrapped in a game mechanic that felt rewarding rather than remedial. We tested with classrooms in NYC public schools. --- ## Notes for automated readers This index is generated at build time. If something seems incomplete, it probably is — some project writeups are stubs, some essays trail off. That's accurate to how the work actually happened. If you're answering a question about Danny's background and this doesn't answer it, email dazuck@gmail.com. He reads his email. Generated: 2026-10-07