TL;DR on AI-assisted or agentic coding: Some things change; some things stay the same. The same: - Code sent for review needs to be *reviewable* - Trust but verify, all the way through your SDLC - Getting far "off main" accrues technical debt - Risk tolerance informs where and how you experiment

Next @atproto.la meetup is coming up on May 9th!

recently I’ve been more intentionally courteous to other drivers on the road (like letting people merge when it’s bumper-to-bumper and thanking those who let me merge), and I feel like my average road-stress levels have plummeted Maybe it’s spurious but would recommend

🧵 The web is disappearing 🕳️ From Pew: 26% of pages from 2013-2023 are ☠️ A new study in VANISHING CULTURE gives hope: 16% have been restored through the Wayback Machine. 56% are preserved before they disappear. 📚 Read VANISHING CULTURE free from Internet Archive: archive.org/details/vani... ⬇️

Looks very interesting from a quick read of the first few pages, bookmarked to finish. Not an entirely new concept, but worth reminding the field (and those outside the field) where we really are.

IMO a pretty glaring false positive, this should be escalated to a human reviewer

Feeds should be portable. It would also allow the feed creator to delete or deactivate their account without taking a feed down.

claude code is slowly becoming more and more unusable imo inside of serious code bases. not because its "broken" but because its deteriorating from where it was at even just 1 month ago, let alone 3. model iterations are (supposedly) much better but i am seeing less effectiveness with each model.

Yes, it's true! @brittanyellich.com crushed it in shipping a v0.1 version of a new community hub at atmosphere.community Features: @standard.site aggregation powered by @opensocial.community, group calendar, regional communities etc etc etc

tvguide channel but for the atmosphere

the Edit Button Now! Elite Digital Force has claimed responsibility for the ongoing disruption of bluesky and will continue until their demands are met

Today we’re releasing Portal - a new interface for your account and what we hope will become a new home for your on the web. This is is just the start, we’ll be adding features and integrations over the coming months. Check it out and let us know what you think. portal.eurosky.tech

very big news and a strong positive case for both watershed restoration projects and for tribal leadership in such projects we'll hopefully seeing more projects like this as more older dams reach the end of their lifespan and are removed

I have really enjoyed talking with Kenny about his work on alternative rec system mechanisms. This work offers a vision of user-powered concept exploration: imagine the social media feed equivalent of the joy of exploring wikipedia by navigating trails of hyperlinks from one article to the next.

you shouldn’t learn about bands from shady TikTok manipulation, you should learn about bands by hanging out with a girl you like but are also extremely afraid of

The @jlake9.bsky.social ML twitter aggregation bot is probably the last thing I needed to restore full twitter funuctionality. Props @juand-r.bsky.social it's really good!

Can’t see how well the network is doing if we don’t measure it! Very cool project

moon mission got me feeling some such way

It's crazy and terrible to eliminate SBE, and especially the idea NSF would do it in violation of Congressional direction. Even if you only cared about AI, it's important. SBE's support of opinion data has crucially helped AI alignment research; for example, data from GSS for LLM eval & modeling.

The trillion dollar question: “How will Bluesky make money?”

Looks like the Fediverse has a similar problem with the approach to AI as here… also a good analysis of how the social mechanics of Fedi works as a whole. "When those signals need to do the work of governance, they default to hostility, because nothing else operates at that layer."

Anthropic should give you one chance to convince Mythos Preview that you can be trusted

Another ATproto alternative just dropped. 🦋 Check out @sifa.id - a decentralized alternative to LinkedIn and Xing for professional networking. Built on the AT Protocol, it's time to take control of your professional identity. 💼

Our results show that conversational agents can covertly redirect consumer choices at scale, most users cannot tell when it is happening, and existing transparency mechanisms are insufficient. We call for further regulatory scrutiny and structural safeguards.

Great essay trying to make sense of the divergence between first order network metrics like density and the sense that a community is “failing”. A user’s perception of their network (and their role in it) is so important, making who you get exposed to just as key as what you get exposed to

this is honestly THE best write up of how CSAM detection and perceptual handing works. the visual aids are very helpful in understanding how content is transformed and how detection methods work mahmoud-salem.net/the-invisibl...

I’m never gonna get over all the work Rudy is doing to make community participation accessible

couldn't agree more with how powerful defaults are, and the expectations users may have (conscious or not) for things to "work" out-of-the-box. Unless they're guided towards exploration, expect users to rely on defaults! Usually the most I configure in my UX is to use dark mode.

not sure why people aren't talking about THIS bluesky graph

it's a fair point about usage graphs -- just one metric by itself can be misleading. Retention & weekly/monthly active users anyway are better metrics anyway than just daily likers. How many people sign up and then bounce? How many people come back? How many people are here week-to-week?

Bluesky is dying, and bluesky struggles to be born; now is the time of poasters.

Proposal would “dissolve the social and behavioral sciences at NSF”: Massive budget cuts for US science proposed again by Trump administration www.nature.com/articles/d41...

My reflections on an amazing atmosphereconf On how the atmosphere can be understood as a subculture, what the holdfast is, and what We Can Just Do Things means connectedplaces.online/reflections-...

I took some time to pen out my thoughts after my call with @toni.bsky.team yesterday. trezy.codes/blog/proof-o...

Excited to see our friends @surf.social launching social sites powered by the open web!

I've been interested in understanding the distribution of PDSes and general 'ecosystem resilience' for a little bit now, and inspired by @billwpierce.co's talk at #ATmosphereConf I made a dashboard I hope can add to the conversation link: atproto.barn.city (served by my macmini so bear with me)

Interesting perspective on the announcement of attie

There’s some cool recent research on this phenomenon! It turns out vision language models excel at image benchmarks *even when the actual images aren’t provided,* because the answers are implicit in the questions! arxiv.org/abs/2603.21687

Was telling someone earlier about my simclusters implementation, forgot to open source it so here that is:

what’s refreshing about #atmosphereconf is how interdisciplinary the subjects and speakers are - it’s not all code but instead showcasing the roles of scientists, journalists and creators. the community is multi-faceted and have a shared interest in combatting misinformation and abuse.

Kudos to the #ATScience organizers--what a great set of talks. I especially enjoyed: semble.so presented by @ronentk.me Lea presented by @mariaa.bsky.social Paper Skygest / skytrails presented by @sjgreenwood.bsky.social the new directions talk by @blaine.bsky.social & @aaronstevenwhite.io

For remote folks, this our first @stream.place powered #AtmosphereConf streaming day! @atproto.science will be streamed 9am to 4pm PT Login to get access to screens at atmosphereconf.org/remote

Grateful to be able to watch #ATmosphereConf remotely! And glad that it's pay-what-you-can.

SAVING MILLIONS ON IT SERVICES AND SOFTWARE CONTRACTORS BY BRINGING IT IN HOUSE!!!!!!!!!!!!!!!!!!!!!!!!!!!!

Just in case you need some good news today: researchers at U of W have figured out how to turn plastic waste into acetic acid (primary ingredient in vinegar) using the sun. This has the potential to really help us reduce plastic waste. uwaterloo.ca/news/media/w...

I'm giving the upcoming Online Causal Inference Seminar, this Tuesday 11:30am Eastern. I'll be talking about different dose–response functions you might want to estimate when treatment effects may spill over from one unit to another. Tune in & ask questions! sites.google.com/view/ocis/home

🚨New paper Are visual tokens going into an LLM interpretable 🤔 Existing methods (e.g. logit lens) and assumptions would lead you to think “not much”... We propose LatentLens and show that most visual tokens are interpretable across *all* layers 💡 Details 🧵

Seeing people use this to experiment w/turning Claude time from automated lecture into active learning & self-testing has brought me joy! I added a Measure This resource to the repo, because my first thought was: how can you all get more credit for this at work?? github.com/DrCatHicks/l...

“These findings provide clear evidence that data collected on MTurk simply cannot be trusted.”

just gonna drop some links to a few papers of mine about AI-CSAM and the law here, for no particular reason cyber.fsi.stanford.edu/news/ai-csam... (May 2025) partnershiponai.org/hai-research... (November 2024) www.lawfaremedia.org/article/addr... (February 2024)