Recent prescriptions and policies would have us judge research by non-evidentiary standards, e.g. presumed political impact and scholar identity. In a new essay, Tom Clark and I argue empirical claims should instead be evaluated based on evidentiary strength. We outline a framework for judging evidence, and show its usefulness by revisiting high-profile scholarly disagreements spanning policing, political violence, election administration, and terrorism, before critiquing non-evidentiary approaches. We also argue social science needs a culture that is hospitable to critique—rather than one which condemns critics as “bullies”—if quality standards are to be upheld. dropbox.com/scl/fi/fsgpflron…

Starting to see tweets suggesting the use of AI in research, even if disclosed, is immoral because of the harms the AI industry might inflict on society. If that view catches on, the effect will not be to discourage the use of AI in research, but to discourage the disclosure of using AI in research. It's here and it's not going anywhere. Our job is to figure out best practices within that reality.

"The work is not complete." Andrew Hall shares a message from his father, @Stanford and Hoover Institution economist Robert E. Hall, winner of the 2026 Nobel Prize in Economic Sciences. Read more: hoover.org/news/nobel-prize-… Audio enhanced with AI.

Robert E. Hall, Hoover senior fellow and @Stanford economist, celebrates his Nobel win on campus with his sons Jonathan Hall (red sweater) and @AHall_Research (black sweater). Read the announcement: hoover.org/news/nobel-prize-…

Whoa

New research from our team’s ongoing work with police body-worn camera data. We show current AI techniques are not reliable enough for unsupervised analysis of police-civilian interactions. More to come on this topic soon! engineering.princeton.edu/ne…

The best approach to the “astrologer in the astronomy department” objection to viewpoint diversity is to bring back a culture of evaluating research on theory, logic, and evidence rather than on conclusions. Not many astrologers will meet that hurdle. But if they do, give them a chair!

WATCH: @seanjwestwood joins KPT to explain the AI threat to political polling he's been warning about for more than a year - and why it would be "naive to assume" that foes like China or Russia are not trying to alter American polling.

We have a narrow window to build political superintelligence and design AI to strengthen free societies. I joined Anthropic so that I could do the unique research on this that can only be done from inside a frontier lab, and I couldn't be more excited for it. We'll also need a broader ecosystem of work beyond the labs---and that's why @freesystemslab is entering a new, bigger, and bolder phase. In under a year, we built the largest human-judged study of political slant in AI, mapped AI politics across 305,000 videos and emails, and beat prediction markets with an AI assistant. We taught AI to catch ambiguous contract language, represent voters, and replicate published research in under an hour for $10. And we discovered AI's strange love for the Japanese Communist Party. But we have a lot more to do. So we're scaling up in three ways. (1) NEW MEMBERS OF THE FREE SYSTEMS FACULTY First, we're bringing on three new faculty affiliates to help us do more work faster: @fouirnaies will help us study how to build agentic tools for legislators. Cassandra Handan-Nader will apply AI to help develop new technical research in political methodology, like scaling legislators based on roll-call votes. @danmthomp will study how to turn AI into the best political advisor possible. I'm incredibly excited to welcome all three on board! (2) OUR FIRST LAB MANAGER But how are we going to do all of this new stuff with all these new teammates?? We need someone focused every day on turning this vision into a functioning research institution: guiding fellows, pushing projects forward, improving how we communicate findings, building partnerships, and helping the work reach researchers, policymakers, technologists, and the public. We have found the perfect person in James Campsie, who is joining us as the first Free Systems lab manager. I've already been working with James for a while, and he is going to supercharge the lab. (3) API TOKEN GRANTS FOR EXTENDING OUR RESEARCH We're also going to scale our work beyond the lab itself. We're using the revenue we make from subscription fees to fund API token grants---applicants can propose work that extends our existing research projects, and we'll provide API tokens for the work. If you want to support these grants, please add a paying subscription! Finally: we're launching our own X account for Free Systems (see below), and a lab website (freesystems.net) that houses all of our artifacts, prototypes, and more. The substack is still the hub for our research, but the website helps track progress over time and puts all our stuff in one place. I'm incredibly grateful to @StanfordGSB and @HooverInst for supporting this work, which will remain separate from Anthropic, and to all of my collaborators, students, friends, and family---I couldn't think of more meaningful work for us all to get to do together.

Partisan motivated reasoning costs votes. A healthy party seeking to gain ground would be hell-bent on actively rooting it out, constantly questioning assumptions, entertaining critiques even when they come from the opposition, if only out of self interest.

In Covid, anyone who did didn’t support a lockdown — which had devastating impacts on vulnerable people— was told they were Covid deniers who wanted to kill vulnerable people Now anyone who even raises the smallest objection to government policy on online harms — for example its likely impacts driving kids to dangerous spaces — is accused of being an abuser. These debates have a lot in common: they assume that all problems have only one solution, and anyone who objects to the solution is part of the problem. They never move from understandably emotional responses to a problem to rational analysis of solutions. They cast a problem as an emergency, charged with moral panic and apocalyptic rhetoric. They ignore that policies drafted to solve one problem may create many others — often longer-term, more complex problems which can only be seen at a distance. They deny there are multiple sides to any issue, and can thus ‘other’ any critics as murderers or pedophiles. And increasingly, they have become tactics of governments themselves: by playing on people’s negative emotions, and presenting any dissent as driven by bad actors, ministers forge a consensus grounded not in a reasoned analysis of a problem but in the fear of what others will say if you hold a different view.

I still find it baffling that Berkeley gave up its cheat code of admitting really smart, hardworking students who couldn't play the Ivy admissions game. The theme of California public institutions seems to be: Try to kill the golden geese!

Weber’s thoughts on how to treat the subject of democracy in the classroom, from his lecture, “Science as a Vocation,” 1917.

Except this is the wrong calculation. The “zero update” was because the AP pushed out two updates to its data feed in close succession. Immediately after the Pratt “0” update there was a Bass and Raman “0” update.

Serious social science question: what observations would be consistent with effective checks and balances operating? Some see democracy crumbling, but the data are also consistent with a president seeking to undermine democracy and failing. 1/6 failed, and Trump keeps getting checked in the second term (see recent retreat on the $1.8b slush fund, but there are many examples). As social scientists, we should offer a way to distinguish these two states of the world. We can of course just assume the worst, but then what is our value added? No one needs a PhD to panic.

No, this isn't right. 41 seconds after the update with zero ballots for Pratt, there was an update with zero ballots Bass and Raman. This isn't evidence of fraud. It is just how a media outlet decided to report non-official vote updates.

1/ New @Nature! We study how powerful institutions shape the information environment for LLMs. Commercial LLM training is opaque, so we trace a path from state-coordinated media -> training data -> model responses.

From our First View: Do Donors Punish Extremist Primary Nominees? Evidence from Congress and American State Legislatures by ANDREW C. W. MYERS doi.org/10.1017/S00030554251…

The rank order replicates with better data, but the % supporting partisan murder is much lower. Importantly, this is passive support and not willingness to actually murder.

Hear from PRL's Sean Westwood about his Cozzarelli Prize-winning paper: The Potential Threat of AI to Online Survey Research piped.video/watch?v=opGrImK4…