Jonathan Warden

Bluesky now has an on-demand research bot. Mention @getcontext.bot in a reply and ask a specific question. It reads the conversation, searches the web, checks original sources, and gives a brief response using Claude. No signup. Free with rate limits. Give it a try. getcontext.bot

Recent study on political knowledge and social media participation: “people who objectively know moreabout current affairs are actually less likely to express themselves politically on social media” techpolicy.press/who-particip...

The folks on the other platform continue to innovate on Community Notes.

if/when we do community notes, it would be cool if everybody who interacted with the OP got a notification when an annotation goes live. (idk if twitter did that)

Fascinating open letter from Stephen Pinker to Scott Alexander: great minds discussing great topics. quillette.substack.com/p/an-open-le...

Online toxicity silences the majority and fuels power users On social media, extreme or hostile content often gains the most visibility. This creates a spiral of silence among most users But polite and constructive environments encourage silent readers to share www.psypost.org/how-online-t...,

People are more pro-social than we think An experiment in which 5,980 lost letters were dispersed in communities found that return rates for ingroup (Italian) & outgroup (foreign) recipients were the same Places that returned more letters to ingroup members also returned more to outgroup members

X continues to improve community notes to help fight misinformation. Why doesn’t Bluesky do the same?

Increasingly convinced that specs are now the source and that code is an artifact akin to a lockfile.

Bluesky seems broadly unaware of this but this is only true of the very first LLMs. Their outputs were essentially their “intuition” - basically outputs that popped into their heads fully formed. But more recent models string those ideas together to work towards a solution.

Muse's security architecture is good, darnit! research.meta.ai/blog/securit...

From my friend Felix Dietze, who is not on here. felx.me/posts/mental...

I'll be in Madrid in April so this is convenient.

Just thought I'd repost this essay: jonathanwarden.com/the-law-of-a...

the logs of the agent swarm that solved navier-stokes are pretty damning

Bluesky team is very responsive! They already fixed the @getcontext.bot thread-visibility issue.

Here's a direct link to the new mysky.social feed (formerly Green Earth) by @jonathanstray.com and team. First atmosphere feed I know of designed to promote constructive conversation. bsky.app/profile/mysk...

Wish I was in the Bay Area and could attend this event with @jonathanstray.com tonight luma.com/the-113t

Social media discourse is driven by false #polarization We find that: 1) ideologues post opinions more than moderates 2) extreme attitudes are posted more than moderate one 3) This replicates in 40 countries 4) Opinions that users post are more hostile than the beliefs they keep to themselves

Maybe we could use scientific techniques to take a representative sample of people's opinions and lived experiences, and compare/contrast that with empirical data. Either that or see who the yell the loudest and most convincingly on social media 🤷

@getcontext.bot is the lab-leak hypothesis generally rejected by serious scientists?

According to major US medical associations, we should all get COVID shots this year. standard-reader.app/a/did:plc:an...

Trump sucks, and yes he fired McEntarfer, but the people putting these together are the same career bureaucrats that have always been, and there's no evidence these numbers are cooked. Revisions aren't new either, it's how the process has always worked. We have enough real things to be mad about.

it’s pretty wild that we as a society of have just completely normalized (and at times encouraged) people to act insanely rude and inappropriate to complete strangers online, in ways they wouldn’t in real life, to the point that a lot of pro-social behavior has is even stigmatized on social media

Deaths from the glacier landslide in Nepal probably can't be blamed on Trump admin.

We need to understand political polarization through a global lens: Evidence from 53 countries suggests that high levels of affective polarization are correlated with democratic backsliding, but high levels of ideological polarization are not. www.nature.com/articles/s41...,

We've entered that time of cycle for uninformed and/or dishonest election doomerism so, quick reminder: Voter rolls are overseen by professional and non-partisan civil servants. The loudest claims of "purges" are often based on gross misreadings of basic rules. www.lonestarleft.com/p/suspense-s...

all I know about the data center debate is that being dogmatically “pro” or “anti” are equally untenable and intellectually lazy positions to hold. the real work of organizing labor power and local community agency is happening in the big messy middle. only the trolls engage at the polar opposites.

Collaborative reasoning is often superior to individual reasoning Correct reasoning was executed by 9% of individuals but by 75% of groups. Groups were 8X smarter! Groups got the correct answer, even when *every member* of the group came to the wrong conclusion www.tandfonline.com/doi/abs/10.1...

This might be the most beautiful thing I've read in my life. What a privilege to be alive in 2026. ordinaryabundance.com

This is a very important concept

A new study finds that fact-checking often has limited coverage, arrives too late, and reaches too few people to keep up with political misinformation. Corrections can still provide useful information, but they often don't reach enough people in time to have their greatest impact.

Can reading non-fiction books reduce partisan hostility? We tested 25 books using simulations and the winners are: “Lovingkindness: The Revolutionary Art of Happiness” + “Facing the Fracture: How to Navigate the Challenges of Living in a Divided Nation”. www.powerofusnewsletter.com/p/can-ai-hel...

I sometimes hear people express concern that bridging-based ranking will homogenize discussions, flattening minority perspectives. Many answers to this, but the first one is: not all minority perspectives are worth preserving. I'm sure you can think of a few. We need a criterion besides "minority."

It's surely just regurgitating from the many counterexamples to the Jacobian Conjecture that were in the training data

I have high hopes for @mu.social as a proving ground for democratic social media, given ample support from the atmosphere community. With the help of civilized-discourse expert @jonathanwarden.com we can have Community Notes on atproto, starting with mu.

big fan of this model of the human body

“Expect the unexpected.” I just realized that this is a declaration of entropy—the expected value of surprisal.

This is an interesting and well-thought-out proposal for a pro-social feature. Another kind of annotation to add context, like community notes. But not a fact check — a “community check” to help people see the disconnect between the vocal minority and actual community sentiment. @newpublic.org

Age verification laws are based in magical thinking. Full stop. All legal considerations, privacy, and human rights concerns aside, there is simply no safe, reliable or practical way for social media companies to verify the age of every single user. Full stop. www.politico.eu/article/age-...