Vertical jobs—where you end-to-end solve a problem—seem more stable right now than horizontal jobs, where you specialize in doing one slice of a process.
Vertical jobs—where you end-to-end solve a problem—seem more stable right now than horizontal jobs, where you specialize in doing one slice of a process.
Just deleted my ChatGPT account. I'd rather use an LLM from a company that aligns with my values. I don't support mass surveillance or AI weapons without a human in the loop, and OpenAI does.
The decision was made easier by just how good the other LLMs are. We are spoiled for choice!
Four months later, AI agents have crossed a threshold. They're a permanent part of my toolbelt now.
I think soon the specs and tests will BE the software
Nice! I’m surprised Meta was okay giving up control. There must be something I don’t understand about the incentIves at play.
I‘ve been writing a big spec for a library I’m working on and I really enjoy it.
I’ve always liked UI work because I get to spend my time designing an output that feels good. Now API design feels similar.
Underrated! Most things can afford to go offline for a bit. It’s okay.
This is so healthy! What a great way to sustain the community for the long term.
I am building an alternative to the open source library MathQuill for inputting math equations on the web
A classroom setting is certainly one major use case for multiplayer
Stellar article. I’m still excited about building realtime math collaboration even though I mostly agree with the takeaways.
I’ve never seen @jay.bsky.team miss. Bluesky team has their eye on the ball
Those things change often but are also low effort so I don't consider it a waste of time.
Aside from that, agents really like using CLIs, so investing in that makes sense, I don't know if you saw what Hono was doing or not: blog.yusu.ke/hono-cli/
I wrote about the most ambitious form of AI-assisted software development I've seen yet - Strong DM's "Software Factory" approach, where two of the guiding principles are "Code must not be written by humans" and "Code must not be reviewed by humans" simonwillison.net/2026/Feb/7/s...
Thank you!
What are the best @standard.site reader apps? I want to subscribe to everyone's blogs!
You don't have to niche down _that_ far to find something that only 20 other people care about as much as you
What a miracle to live in a time where you can find & befriend everybody who cares about a hyper-specific niche as much as you do.
Now I understand how the punchcard → C folks felt when they saw my generation learning JavaScript and Python without knowing fundamentals
Been mulling over @bnewbold.net's recent post on community spaces and atproto's next frontiers. I wrote about why building for organizers (not just users) might be the key to making it all work.
I feel like it makes sense to let each service define their own custom schema and then let folks define super-lexicons that aggregate results from multiple services into a unified shape (possibly losing a bit of app-specific data along the way)
bsky.app/profile/pull...
This is fun!
First to like zod on @npmx.dev :)
I’m actually… surprised there’s nothing longer?
This just got more interesting
Right. Which is a much smaller group of decision-makers. Good call
I also don’t see how the hash chain could prevent the registry from deleting the newest entries at will, but again, folks would probably notice in practice
But I also don’t really imagine this being a practical problem considering how neutral the purpose of the registry is. The registry isn’t really making any decisions. It just needs to keep the lights on. And if it goes fully offline, then getting folks to move would probably be easy
That is mostly enough to convince me that this is a robust solution. I’m a little worried about #2 because the Twitter saga showed us that moving everyone at once is almost impossible.
This is definitely helpful. I think the two pieces I was missing are
1) Everything the registry does is auditable by the public
2) There’s no technical barrier to copying all the data and creating a new, successor registry except for the social barrier of convincing everyone to use it
@danabra.mov I'm struggling to grasp how a user's ATProto identity can survive changes in domain/hosting/etc without too much centralization on the registry.
As far as I understand, it seems like the registry just... is? Centralized?
Maybe it's fine. But I'm curious if you have more insight.