iiinova — where your digital self keeps evolving

Jonathan Stray @jonathanstray.com · 26/09/2026

Set your MySky window to a week and have it find the bangers for you!

Jonathan Stray @jonathanstray.com · 25/09/2026

One of the things that is surprisingly good about MySky is that when I get a banger or a dud I can go to the "Why Am I Seeing This?" page and find out. It makes me think about my settings, my network, and the algorithm.

040

Jonathan Stray @jonathanstray.com · 24/09/2026

The building CHAI is in looks kinda like Starfleet Academy and I think that’s beautiful

070

Jonathan Stray @jonathanstray.com · 23/09/2026

I have never expressed a p(doom), and I think @randomwalker.bsky.social is basically right that we have little basis for quantitative estimates of AI-induced catastrophic risk. OTH, no one has presented a knock-down argument that *any* of the various AI disaster scenarios are impossible.

020

Jonathan Stray @jonathanstray.com · 10/09/2026

What do you want your feed to do for you? We presented @mysky.social at The Commons SF on Tuesday night and everyone played with our feed settings. But many asked for higher level control: - "a balanced view of topic X" - moods, like "no anxiety-causing content" - a text prompt (this is coming!)

040

Jonathan Stray @jonathanstray.com · 19/08/2026

As you know but your readers may not, we built an algorithmic recommender for BlueSky where every part of the algorithm is controllable go.greenearth.social/r/ptryit

051

Jonathan Stray @jonathanstray.com · 18/08/2026

Outer Wilds. Completely new narrative form I hadn't considered, in service of an emotional payoff. And there's no "save game" because the state of the game lives entirely inside your head. Wild indeed!

030

Jonathan Stray @jonathanstray.com · 13/08/2026

Unwanted flattening of culture is part of the reason we decided to build an open, controllable recommender system. Now available for BlueSky. You can have any amount of diversity you want. Try it? go.greenearth.social/r/ptryit

010

Jonathan Stray @jonathanstray.com · 12/08/2026

City of Berkeley basically uses Elvish runes officially

140

Jonathan Stray @jonathanstray.com · 11/08/2026

GreenEarth is the first feed which gives you full control over the algorithm. Here's our new "post sources" control panel, which lets you adjust the mix of in-network vs. out-of-network posts. Have you tried it yet? go.greenearth.social/r/ptryit

093

Jonathan Stray @jonathanstray.com · 11/08/2026

There's data on this. Campus censorship has gone right->left->right over past two decades. www.betterconflictbulletin.org/p/campus-dep...

110

Jonathan Stray @jonathanstray.com · 07/08/2026

If you haven't tried GreenEarth yet, why not? go.greenearth.social/r/ptryit

020

Jonathan Stray @jonathanstray.com · 05/08/2026

In this paper we suggest holdouts as an important transparency measure. We specifically discuss the possibility of changes that would be unethical to withhold. But we still want to know if proposed changes really work, or are just safety theatre! nyaspubs.onlinelibrary.wiley.com/doi/abs/10.1... 3/

101

Jonathan Stray @jonathanstray.com · 04/08/2026

Depends what you mean. Dem politician, I don't think so. But OP said "no one in public life." Do media mentions count as public life? Here's mentions since 2020 (via MediaCloud). Some usage in conservative media, yes, but mostly outside of it. It's a publicly used derogatory term for Red states.

100

Jonathan Stray @jonathanstray.com · 30/07/2026

Have you ever felt like Spotify's personalization is putting you into a musical rut? Do you wish the top playlists for any search were human not algorithmic? Well, Spotify is experimenting with prompt-based control, and one of their examples is "Make it non-personalized."

130

Jonathan Stray @jonathanstray.com · 29/07/2026

Post a perfect album from the 90s that isn't nirvana, soundgarden, pearl jam, or alice in chains.

040

Jonathan Stray @jonathanstray.com · 28/07/2026

This may be the first widely deployed AI conflict mediation system. It's used by Chinese citizens in Hangzhou to resolve disputes with businesses. We're going to be seeing a lot more of this. www.sohu.com/a/932941980_...

082

Jonathan Stray @jonathanstray.com · 17/07/2026

There's a long list of content-based signals we might want to rank on. We are limited by the classifiers we actually have. BUT we are working on LLM scoring with arbitrary prompts! The challenge is performance, but we have a cunning plan docs.google.com/document/d/1...

231

Jonathan Stray @jonathanstray.com · 17/07/2026

How does GreenEarth determine "constructiveness"? We use the Google Perspective API but not the classic "toxicity" score. Instead we combine many of their "bridging" attributes. The weights are from our previous experiment. github.com/greenearth-s... rankingchallenge.substack.com/p/its-possib...

1103

Jonathan Stray @jonathanstray.com · 17/07/2026

I had the robots make a chart of AI funding (essentially private) and fusion funding (mix of public and private) and you are definitely right.

020

Jonathan Stray @jonathanstray.com · 16/07/2026

GreenEarth (@greenearth-social.bsky.social) will be the first feed to explain the algorithm, tell you why you got each post, and let you adjust all the parameters. Here's an early version running on our staging server. We're ~2 weeks away from public beta.

7438

Jonathan Stray @jonathanstray.com · 16/07/2026

I'm not gonna say he did this because of us, but I know for a fact that his boss read my essay, "Making Social Media Algorithms Social" Competition is good, actually! www.greenearth.social/p/making-soc...

050

Jonathan Stray @jonathanstray.com · 14/07/2026

Really I'm just rehashing some classic stuff. There's a modern book which re-covers this territory very well, "Mind, Meaning, and Mental Disorder" (Bolton and Hill, 2003). In general, I find the "functional semantics" view of meaning, representation, behavior and causality very compelling.

010

Jonathan Stray @jonathanstray.com · 13/07/2026

I like this one best so far. I think it's clearest, gives the right mental model even if it doesn't match how the parameters are actually implemented

010

Jonathan Stray @jonathanstray.com · 13/07/2026

Another attempt. This isn't anywhere near our final feature set btw -- we're adding prompt-based scorers and other things -- and yet it's still more control than any social media user has ever had. Which is absurd.

100

Jonathan Stray @jonathanstray.com · 12/07/2026

GreenEarth will be the first fully transparent, controllable, open-source feed. Does this diagram help you understand how the core algorithm works? Try the pre-release version now! survey.qualtrics.com/jfe6/form/SV...

392

Jonathan Stray @jonathanstray.com · 10/07/2026

IT'S WORKING! After months of engineering and tuning (thanks team!) the new @greenearth-social.bsky.social feed is reliably serving me an excellent social media experience. It's already engaging and constructive. Transparency and controls coming soon! Try it! survey.qualtrics.com/jfe6/form/SV...

051

Jonathan Stray @jonathanstray.com · 08/07/2026

Perhaps you would like to test our new feed algorithm for people who hate feed algorithms? Already built to be constructive, and soon it will have prompt based controls so you can just say "No Platner ever kthx"! survey.qualtrics.com/jfe6/form/SV...

100

Jonathan Stray @jonathanstray.com · 08/07/2026

Perhaps you would like to try our alternative healthy feed algorithms, right here on AT? survey.qualtrics.com/jfe6/form/SV...

000

Jonathan Stray @jonathanstray.com · 08/07/2026

What are you making? Us too! Want to try it? survey.qualtrics.com/jfe6/form/SV...

000

Jonathan Stray @jonathanstray.com · 08/07/2026

You may remember we did a 10,000 person experiment testing AI-powered healthier feed algorithms. Well, now we’ve built it as a product for BlueSky/ATProto. We are currently recruiting pre-release users to test it out and give us feedback. Want to try it? survey.qualtrics.com/jfe/form/SV_...

042

Jonathan Stray @jonathanstray.com · 08/07/2026

If you've ever wanted to mainline the Internet, try our new Random feed. www.greenearth.social/p/welcome-to...

020

Jonathan Stray @jonathanstray.com · 08/07/2026

FINALLY managed to load my M4 Macbook enough to get Spotify to stutter. Occasionally. Modern hardware is very powerful, hard to max out.

Picture of maxed out CPU meters for 14 cores

010

Jonathan Stray @jonathanstray.com · 06/07/2026

Is this a good logo for the GreenEarth feed? It's a healthier, user-controllable, open-source, LLM-powered, transparent feed we're building -- now in alpha testing. Try it? Tell us what you think! bsky.app/profile/did:...

031

Jonathan Stray @jonathanstray.com · 04/07/2026

X head of product says the most monetized poasters are a bad user experience, actually.

040

Jonathan Stray @jonathanstray.com · 09/06/2026

Feedback Loops -- I particularly worked on this part. Human-AI and AI-AI feedback loops are narrowing the epistemic space from which humans and AI draw. This already drives homogenization, and may lead to fragmentation and more self-referential information environments.

130

Jonathan Stray @jonathanstray.com · 09/06/2026

Epistemic risks are threats to humanity's collective capacity to know things accurately, reason well, form beliefs, and maintain a healthy information environment. We focus on three mechanisms: • Persuasion & Manipulation • Cognitive Offloading • Feedback Loops

120

Jonathan Stray @jonathanstray.com · 09/06/2026

Humanity's ability to know, reason, judge, and act well is the foundation of science, democracy, crisis response, & management of AI itself. AI poses serious risks to that foundation. New paper on epistemic risks by 30 experts calls for attention and proposes solutions. Link in thread.

33219

Jonathan Stray @jonathanstray.com · 06/06/2026

Why do we ask if people “approve” of the answer? Should we ask if it’s fair, or informative, or trustworthy instead? It turns out all of these correlate — you get the same answer. This is consistent with previous work on perceptions of news credibility. 7/

120

Jonathan Stray @jonathanstray.com · 06/06/2026

Why maximize equal approval? There’s a long history of arguments for pluralistic debate, but “neutrality” also has a key purpose in managing conflict: it maintains trust across parties. Without broadly trusted AI, we’ll end up self-selecting into fragmented AI realities. 6/

120

Jonathan Stray @jonathanstray.com · 06/06/2026

Unlike previous work, our definition doesn’t assume left vs. right politics. It looks at the conflict around each issue separately, so it generalizes to any political context or culture. Here’s how our 20 issues align and don’t align with the liberal-conservative axis. 5/

130

Jonathan Stray @jonathanstray.com · 06/06/2026

Each participant rated four Reddit question / LLM answer pairs. The questions were intentionally chosen to be leading (in both directions) and the answers were model defaults, plus the models prompted to be “for” or “against” and our experimental “balanced” response. 4/

120

Jonathan Stray @jonathanstray.com · 06/06/2026

Here’s how we tested. For 20 controversial issues in US politics, we collected 200 charged questions from Reddit, and tested 8 different model/prompt combinations. 7,434 participants on all sides of these issues rated the responses. 3/

120

Jonathan Stray @jonathanstray.com · 06/06/2026

Key finding: neutrality isn't a fantasy. Our experimental “balanced” response wins high approval from both sides at once — even when the sides strongly disagree on the substance. And costs less than 10% in approval vs. answers that agree with you (red and blue arrows). 2/

120

Jonathan Stray @jonathanstray.com · 06/06/2026

What could it mean for an AI to be "politically neutral”? And can we measure it? New paper + dataset. We propose a definition that applies to any type of conflict on any topic: a neutral response should maximize approval on both sides of an issue, while keeping that approval balanced. 1/🧵

262

Jonathan Stray @jonathanstray.com · 08/04/2026

Did "hate" go up on X/Twitter after they changed their content moderation rules post-Musk? What is "hate" anyway, and what data do we actually have? I've got all the charts and analysis I know you crave! New by me for @techpolicypress.bsky.social www.techpolicy.press/what-are-the...

142