It's the insider's problem. You stay on the inside because you think it is easier to change the system from the inside. But you just end up becoming part of the system. They can't change from the inside, disruption is always external.

Every crypto bro cheering this bill is either on Coinbase’s payroll or can’t read. I read all 278 pages. You’re getting played. I’ve been in crypto since 2012. That’s 14 years of watching governments pretend to be confused while quietly building the cage. Trump promised to make America “the crypto capital of the world.” His party just delivered a surveillance framework that would make the CCP blush. Today I’m launching the Day2026 Bill Tracker. It does one thing: exposes how both parties collaborate to build your digital prison while you cheer. First up: The Senate Digital Asset Market Structure Act. 278 pages of “regulatory clarity” from Senator Tim Scott. Translation: 278 pages of compliance theater that kills everything crypto was built for. Here’s what your favorite influencers won’t tell you because their bags depend on you not knowing: MANDATORY TRADE SURVEILLANCE - Every exchange must implement real-time monitoring. Every. Single. Transaction. The NSA called, they want their playbook back. UNIVERSAL REGISTRATION - Exchanges, brokers, dealers, even “associated persons” must register. Anonymous participation? Dead. Satoshi’s vision? Buried. FULL DISCLOSURE TO THE STATE - Token issuers must hand over source code, transaction history, and tokenomics to regulators. Open source for thee, total transparency for me. MANDATORY GOVERNMENT CUSTODIANS - Your coins must sit with approved custodians. Self-custody for regulated activity? Effectively illegal. Not your keys, not your coins just became federal policy. DEFI IN THE CROSSHAIRS - For the first time ever, DeFi developers face registration requirements. Building permissionless systems now requires permission. Let that sink in. YOUR DATA GOES GLOBAL - Transaction records flow to the SEC, CFTC, and foreign regulators. Your wallet activity shared with central banks worldwide. Bullish, right? WHO ACTUALLY WINS: Coinbase gets a regulatory moat that buries competitors. You think Brian Armstrong is lobbying for YOUR freedom? Chainalysis gets permanent government contracts. Surveillance as a service, funded by your tax dollars. BlackRock and Wall Street get clear on-ramps while DeFi gets strangled in the crib. The SEC and CFTC get expanded empires and fresh revenue streams. You get watched. Tracked. Controlled. But hey, number go up. THE PROCESS: Senators got 48 hours to review 278 pages. Democrats asked for more time. Denied. Because nothing says “deliberative democracy” like speed-running financial surveillance. They call it regulatory clarity. I call it regulatory capture gift-wrapped for the donor class. THE REAL GAME: This is what “bipartisan consensus” means in 2026: both parties racing to build total financial surveillance while fighting about pronouns on cable news. Republicans say they oppose CBDCs. Then they vote for infrastructure that makes CBDCs inevitable. Democrats say they want consumer protection. Then they vote for bills written by the corporations they claim to regulate. Different jerseys. Same owners. THE UNCOMFORTABLE TRUTH: Trump isn’t saving crypto. He’s domesticating it. The goal was never to ban Bitcoin. The goal was to make it legible, trackable, and taxable. Mission accomplished. Every laser-eyed profile pic celebrating this bill is either naive, compromised, or selling you something. WHAT I’M DOING ABOUT IT: Full analysis with threat scores, beneficiary tracking, and talking points: (day2026.com/legislation/sena…) Every major bill gets this treatment. PATRIOT Act. TARP. CARES Act. REAL ID. GENIUS Act. Executive orders. All of it. Exposed. THE ANNOUNCEMENT: Neither party will protect your financial freedom. Neither party actually opposes CBDCs. Neither party will stop the technocratic merger of corporate and state power. That’s why I’m exploring a run for US Senate in New Hampshire. Not to join the club. To burn down the velvet rope. The algorithm buries truth. Make it work for us.

Teaching an experimental class for MBAs on “vibefounding,” the students have four days to come up and launch a company. More on this eventually, but quick observations: 1) I have taught entrepreneurship for over a decade. Everything they are doing in four days would have taken a semester in previous years, if it could have done it at all. Quality is also far better. 2) Give people tools and training and they can do amazing things. We are using a combination of Claude Code, Gemini, and ChatGPT. The non-coders are all building working products. But also everyone is doing weeks of high quality work on financials, research, pricing, positioning, marketing in hours. All the tools are weird to use, even with some training, but they are figuring it out. 3) People with experience in an industry or skill have a huge advantage as they can build solutions that have built-in markets & which solve known hard problems that seemed impossible. (Always been true, but the barriers have fallen to actually doing stuff) 4) The hardest thing to get across is that AI doesn’t just do work for you, it also does new kinds of work. The most successful efforts often take advantage of the fact that the AI itself is very smart. How do you bring its analytical, creative, and empathetic abilities to bear on a problem? What do you do with access to a very smart intelligence on demand? I wish I had more frameworks to clearly teach. So many assumptions about how to launch a business have clearly changed. You don’t need to go through the same discovery process if you build a dozen ideas at the same time & get AI feedback. Many, many new possibilities, and the students really see how big a deal this is.

.@solanamobile is proving that open mobile crypto works, with over $2.6B in volume on Seeker so far Next up: Guardians decentralizing the app store and a path to hardware expansion with MediaTek, the chipmaker powering nearly 2B Androids globally

Follow the Constitution. Read the bills. Stop governing by emergency. End the forever war. Cut taxes & spending. Stop borrowing trillions. No CBDC. Protect free speech. Repeal the Patriot Act & FISA 702. No qualified immunity for government officials. End civil asset forfeiture.

Professional Software Developers Don’t Vibe, They Control Vibe coding isn't how experienced developers actually use AI agents. The term has exploded online. Practitioners describe an experience of flow and joy, trusting the AI fully, forgetting code exists, and never reading diffs. But what do professionals with years of experience actually do? This new research investigates through field observations (N=13) and qualitative surveys (N=99) of experienced developers with 3 to 41 years of professional experience. The key finding: professionals don't vibe. They control. 100% of observed developers controlled software design and implementation, regardless of task familiarity. 50 of the 99 survey respondents mentioned driving architectural requirements themselves. On average, developers modify agent-generated code about half the time. How do they control? Through detailed prompting with clear context and explicit instructions (12x observations, 43x survey). Through plans written to external files with 70 steps that are executed only 5-6 steps at a time. Through user rules that enforce project specifications and correct agent behavior from prior interactions. What works with agents? Small, straightforward tasks (33:1 suitable-to-unsuitable ratio). Tedious, repetitive work (26:0). Scaffolding and boilerplate (25:0). Following well-defined plans (28:2). Writing tests (19:2) and documentation (20:0). What fails? Complex tasks requiring domain knowledge (3:16). Business logic (2:15). One-shotting code without modification (5:23). Integrating with existing or legacy code (3:17). Replacing human decision making (0:12). Developers rated enjoyment at 5.11/6 compared to working without agents. But the enjoyment comes from collaboration, not delegation. As one developer put it: "I do everything with assistance, but never let the agent be completely autonomous. I am always reading the output and steering." The gap between social media claims of autonomous agent swarms and actual professional practice is stark. Experienced developers succeed by treating agents as controllable collaborators, not autonomous workers. Paper: arxiv.org/abs/2512.14012 Learn to build effective AI agents in our academy: dair-ai.thinkific.com/

Big moment for Postgres! AI coding tools have been surprisingly bad at writing Postgres code. Not because the models are dumb, but because of how they learned SQL in the first place. LLMs are trained on the internet, which is full of outdated Stack Overflow answers and quick-fix tutorials. So when you ask an AI to generate a schema, it gives you something that technically runs but misses decades of Postgres evolution, like: - No GENERATED ALWAYS AS IDENTITY (added in PG10) - No expression or partial indexes - No NULLS NOT DISTINCT (PG15) - Missing CHECK constraints and proper foreign keys - Generic naming that tells you nothing But this is actually a solvable problem. You can teach AI tools to write better Postgres by giving them access to the right documentation at inference time. This exact solution is actually implemented in the newly released pg-aiguide by @TigerDatabase, which is an open-source MCP server that provides coding tools access to 35 years of Postgres expertise. In a gist, the MCP server enables: - Semantic search over the official PostgreSQL manual (version-aware, so it knows PG14 vs PG17 differences) - Curated skills with opinionated best practices for schema design, indexing, and constraints. I ran an experiment with Claude Code to see how well this works, and worked with the team to put this together. Prompt: "Generate a schema for an e-commerce site twice, one with the MCP server disabled, one with it enabled. Finally, run an assessment to compare the generated schemas." The run with the MCP server led to: - 420% more indexes (including partial and expression indexes) - 235% more constraints - 60% more tables (proper normalization) - 11 automation functions and triggers - Modern PG17 patterns throughout The MCP-assisted schema had proper data integrity, performance optimizations baked in, and followed naming conventions that actually make sense in production. pg-aiguide works with Claude Code, Cursor, VS Code, and any MCP-compatible tool. It's free and fully open source. I have shared the repo in the replies!

The greatest threat to your constitution, your liberty, your prosperity - is the Republican party. Until the Democrats get back in power. Then they'll be the worst. Until the Republicans get back in power. And so on. The largest government in history is the number 1 threat you face - and whoever is running it is your greatest danger.

His narcissism reaches levels unlike any we have seen by an American president. In 2026, he will decree that Washington, D.C. be renamed to Trump Town, D.T. The announcement will be made in front of the Arc de Trump, next to the 30 ft tall statue of Trump as a wartime general.

I'm super excited to let you guys know I'll be attending @socallinuxexpo this year, March 5th through the 8th. I'm still preparing my speaking plans, but I wanted to pass along coupon code "LINTV" which will give you 50% off registration. See you there! socallinuxexpo.org/scale/23x

At a Claude Code specific meetup in KC. It’s just getting started, but there are a ton of people here. Way more than I expected!

I tried Debug Mode in Cursor 2.2 on a client project today and it was REALLY good. Instead of the agent dumping 200 lines of code, it slowed down. It added logs. We reproduced the issue. Saw what was actually breaking. Then fixed it in like 3 lines. And that’s when it clicked. This is how debugging has always worked. Even before AI. You write code. You test. You add logs. You understand the problem. Then you fix it. Debug Mode just follows that same mental model. Faster. Cleaner. Less prompting. This is genuinely GOOD thinking.

I've never considered myself 'an effective altruist', though I'm obviously 'adjacent'. Don't given enough to the global poor; don't care enough about animals to make the sacrifice of veganism; have donated precisely zero kidneys to strangers. Don't meet the standard I see others set. As with most broad movements, plenty of interpretations of the basic principles I disagree with too, enough so that it wouldn't make sense for me to think of myself as an -ist (true for most things for me). And mistakes have most certainly been made, by EAs and in pursuit of EA goals. But I've not encountered any community as enriched for intellectually and morally virtuous people, or as earnestly trying to do good. I think there will be a lot of things they're doing that older me will look back on with gratitude and admiration. And I'm on the whole ashamed of how academia has treated effective altruists.

MIRI is running our first fundraiser in six years, and it's a very ambitious one; we're trying to raise $4.4M (the first 1.6M of which will be matched by the Survival and Flourishing Fund). Our most recent formal fundraisers raised $600k (in 2019), $950k (in 2018), and $2.5M (in 2017), so we're shooting for a very ambitious target. MIRI currently has enough funds to keep doing the work we've been doing for 15 months. Succeeding in the fundraiser would mean raising that to 24 months, which is a level where we can feel a lot more confident in making plans, hiring, and focusing on the mission. The mission's been going well — If Anyone Builds It, Everyone Dies has made a huge splash, and there's a growing list of opportunities we're jumping on to capitalize on that. We're spending an increasing amount of time in conversation with policymakers in D.C., and just a few days ago our CEO (Malo) testified to the Canadian House of Commons about the danger posed by AI companies' race to build superintelligent AI. This does not seem like the time to pull back and stop trying. If you can help at intelligence.org/donate, I'd be incredibly grateful. And given how ambitious our fundraising target is, if you have any wild ideas for ways to help the fundraiser succeed, I'd love to see people tapping their networks and trying different approaches.

Critics of the economics profession often urge us to be more open-minded about government. I, in contrast, urge my colleagues to be more open-minded about being closed-minded about government. Enough is enough!

Trying to locate the clause in the Constitution which authorizes the @FCC to exist. We're sure it's there, since everyone assures us that this is the kind of the thing the Founders would have wanted a federal government to do, licenses and all.

Seeking students & open-source contributors to join some ML-for-science projects at @huggingface. If you're curious about ML × biology or ML × materials science, this could be a great way to learn contribute. I can offer Pro subs, support, and (for longer-term work) recommendation letters. If you or anyone you know might be interested—reach out (DMs open).

I love it when software pops a dialog that says "$SOFTWARE encountered an error". Like the software was grazing in the savanna when it was ambushed by a prowling error, and with the error's jaws firmly clamped around the software's neck the last thing the software could do before blacking out was to bravely pop up that dialog.

At the beginning of my career, I learned Perl using vi on linux and SunOS. Eventually, I moved on to IDEs and programming text editors like VS Code. Recently decided to try nvim out for programming for old time’s sake. Eh, still like Pycharm better. :)

“However good intentions may be, they can never render unsuitable means any more suitable.” - Ludwig von Mises