Running on Apple Silicon: https://x.com/awnihannun/status/2016221496084205965

In 2026, you should be implementing MLS instead of Matrix.

what? that's like saying "you should implement TLS instead of HTTP"!

They do entirely different things: MLS is a key agreement protocol, equivalent to the Double Ratchet that Matrix uses for E2EE today. Matrix can use both.

Getting the AI to work with Rocq is a useful goal, Lean has been useful so far.

I mostly write lean4 now and emit proof-carrying System F Omega via rfl. It's the right level of abstraction when the droids have been pinned to theory laden symbolisms. It's also just pleasant to use.

Whoops, I literally did the same thing as this guy earlier this week, but did the testing using `claude -p` so I can identify when Claude Code would (or would not) load Skills for a particular prompt, so that I could improve the skill definition.

Who knew that using Claude to introspect on itself was against the ToS?

I've had AI write ~100% of my code for the last 7 months, but I acted as the "agent" so the AI had very high levels of direction, and I approved all code changes at every step, including during debugging.

Mostly Gemini Pro 2.5 (and now Gemini Pro 3) and mostly Clojure and/or Java, with some JavaScript/Python. I require Gemini's long context size because my approach leans heavily on in context learning to produce correct code.

I've recently found Claude Code with Opus 4.5 can relieve me of some of the "agent" stuff I've done, allowing the AI to work for 10-20 minutes at a time on its own. But for anything difficult, I still do it the old way, intervening every 1-3 minutes.

Each interaction with the AI costs at least a $1, usually more (except Claude Code, where I use the $200/month plan), so my workflow is not cheap. But it 100% works and I developed more high-quality code in 2025 than in any previous year.

Just share your skill folder (with the SKILL.md file). Your description is unclear.

The reverse is also true: secure code is difficult to write, and LLMs at scale will make it much easier to develop secure code.

Spreadsheets replaced developers for that kind of work, while simultaneously enabling multiple magnitudes more work of that type to be performed.

And I would argue speadsheets still created more developers. Analytics teams need developers to put that data somewhere, to transform it for certain formats, to load that data from a source so they can create spreadsheets from it.

So now instead of one developer lost and one analyst created, you've actually just created an analyst and kept a developer.

I do agree, that’s like my go to thought.

Citizen developers were already there doing Excel. I have seen basically full fledged applications in Excel since I was in high school which was 25 years ago already.

If anything, there were a bunch of low barrier to entry software development options like HyperCard, MS Access, Visual Basic, Delphi, 4GLs etc. around in the 90s, that went away.

It feels like programming then got a lot harder with internet stuff that brought client-server challenges, web frontends, cross platform UI and build challenges, mobile apps, tablets, etc... all bringing in elaborate frameworks and build systems and dependency hell to manage and move complexity around.

With that context, it seems like the AI experience / productivity boost people are having is almost like a regression back to the mean and just cutting through some of the layers of complexity that had built up over the years.

The WIDGET model of "working geniuses" is one possible answer, it does explain a lot of team dynamics in my experience.

Since no one has all six working geniuses, and you're only a genius at two, it takes a collection of people, proportional to the work that needs to be done, of each type.

The opposite note widget is the “film director”.

Eg they are not the best screenwriter, producer or cinematographer but they are the best at getting all of those people to work together.

Seeing as how it's is trivial to buy followers, that metric should be completely abandoned, it's not legitimate.

It's maybe slightly less trivial to do, but still incredibly common to buy awards, recognition, press releases, positive reviews and commentary in publications.

You might be shocked to find out how much the performers being written about in magazines or discussed on TV shows is a direct line to the production company promoting them. Similar for awards.

Shouldn't be shocked, really. This is exactly what people pay PR companies for.

> You might be shocked to find out how much the performers being written about in magazines or discussed on TV shows is a direct line to the production company promoting them. Similar for awards.

I mean Payola as a term literally came from bribing DJs on radio stations to play your / your artist's music.

Money is an option, regardless. EB-5 is about $1M invested into your own business, 10 people hired.