Tim Disney's links

At some point in the 2010s, it became clear that Elon was obsessed with an incongruity in digital space. The man was getting richer and richer, and everyone around him was telling him he was cool and successful. But on Twitter, a site that mattered a lot to his social cohort,…

Big day for docs 🙂🙂

Responding to Noah Smith

Thoughts on the Meta CEO’s recent essay on AI, “The Future Is for Everyone”—less a manifesto, and more like nine blog posts in a trenchcoat.

Reflections on AI's writing ability and how AI models get more capable.

Sincere disagreements about AI are usually disagreements about future AI capabilities.

you might find yourself standing in the unremarkable center

If the film’s pre-release critics, enraged over Nolan’s decision to adopt a diverse cast, had waited for the release, they could have instead been angry over its themes, which position the film as speaking for the norms and civility of the Western civilization those critics cl…

of atmospheric threads

I've been iterating on a coding-agent orchestator for atproto for a while now and it's finally in a place I'm happy with. Still early, bugs, rough onboarding, etc. but finally coming together.

First of a new series on sensemaking patterns we're observing on Semble

We now have more details of what happened. Every time we learn more details, it somehow makes things seem worse.

Rob Miles: Some people become staggeringly credulous as long as the story sounds cynical and jaded enough. “Our product sometimes goes out of control and commits…

We now have more details of what happened. Every time we learn more details, it somehow makes things seem worse.

How I shipped 108 PRs in eight days with an agent working my task board: the loop engineering setup, constraints, and lessons that made it work.

Kudos to OpenAI for sharing their recent experiences with a misaligned internal model, where they encountered problems sufficiently severe they were forced to take the model offline to work on new mitigations and defense-to-depth.

people and planet need open models to win

It should have been disowned, not republished

Chatto is built with agentic engineering. Here's my workflow.

We want blogging to feel less like posting to the void, and more like sharing with a community. From pckt notes to better discovery, everything we've built over the last few weeks is designed to help bloggers connect.

by helping a thousand people make a million

A calmer front door and more tools to work through your reading backlog

A self-only space is the smallest shape permissioned data can take: one authority, one member, one repo, and private bookmarks that share a record shape and a write path with public ones.

My keynote at ICML 2026

...

Lab Notes 032: custom pages and nav for all your publications — with a whole bunch of great examples showing sites made with Leaflet

In a world obsessed with moats, the open social web drops drawbridges

Semble API launched, and the community didn't waste a minute. Plus share-outs from ATScience, the MIRA (modular research) workshop and a Frontiers of Open Science panel.

decentralized networking is best realized with decentralized ownership

Internet darkness is turning into deadness and time is running out

The uncomfortable lesson is that tool schemas are not neutral, at least not on
Anthropic models. We like to pretend that a schema is an abstract contract and
the model is a general reasoner that will follow it, but that might no longer be
the case for some of the tools.

I generally disagree with Sayash and Arvind, but this is the prediction of theirs that I’ve thought about the longest, without being able to find any decisive refutation. It’s a great test case! If AI hits top-human level forecasting and then flattens off, maybe there’s someth…

Approaching a shared method of single sign-on for the entire open social web.

the door's open

Lurking under this idea that AI models themselves should be “more democratic” is sometimes the idea that all products in society should reflect the correct opinions of the good majority, and never reflect the opinions of the bad minority. I think a better solution here would b…

Sync your reading list across the Atmosphere. Read magazine-like collections.

Idiocracy’s major mistake wasn’t the future it predicted, which would arrive a lot quicker than 500 years as it turned out, but who the movie blamed for it. President Dwayne Elizondo Mountain Dew Herbert Camacho, a Black former pornographic actor and wrestler, doesn’t look lik…

Comments are about as old and basic as the social web gets, so I assumed ATProto would have them figured out. It doesn’t, not yet.

Yeah comments are gnarly. Want to support something like them for linkblog shares in @skyreader.app but holding off for now due to exactly these…

At last, we have invented the abundance machine from the cautionary tale "the abundance machine is really cool actually, someone should make one of those". The threat to intelligence on demand is not resource scarcity, it is unexpected externalities -- like, say, holding a mo…

Skyreader can now highlight and integrates with Margin. Everything is faster now too.

But the emergence of agentic coding has changed our perspective, and we’ve gone from being skeptical to being excited about the possibilities.

The internet you grew up on isn't dying. A commercial veneer glued on top of it is.

Underneath that layer is another internet: older, slower, less polished, harder to monetize, and much harder to kill.

But, it takes time. And that can be frustrating for those of us who want to live in that better future.

easy to get cynical when the future we can imagine isn't here yet. but we can just build things!

Lab Notes 028: a new suite of features to make Leaflet great for collaborative writing and collaborative publishing!

The mode of production behind the man

Not endorsing the predictions here but worth thinking through the reasoning.

Arguments against simulation: Too weird.

Yeah, I kinda think that the lesswrong and adjacent crowds would be taken more seriously if they just stopped talking about the simulation argument. Too weir…

Coding agents as normal technology

After two days of experience with Claude Fable 5 I think the best way to describe it is relentlessly proactive . It knows a whole lot of tricks and it will deploy pretty much any of them to get to its goal. I'll illustrate this with an example. I was hacking on Datasette Agent…

Very excited for this!

Making it even easier to build with Semble.

Saves in @skyreader.app are off protocol currently so this is a great starting point for thinking through what a save/bookmark integration would require.

Boris makes the point that Skyreader linkblogs should actually point to the link so now it does! The other suggestions are good too and on the roadmap.

Edit: mentions should work too: @bmann.ca

Sharing the sharing post. How meta.

So excited for permissioned data!