Dan Thompson

I’ve always liked tools that make you want to experiment. So I built one. It’s called maxdraw. Draw with emojis or text and rotate them along the path. maxdraw.app

a talking cow is all you need: the magic of unix pipes provides the universal TUI interface L: tidal cycles installer R: my preferred way to run @simonw's llm package

@paulmitche11 This is really cool. We need to get the intro to American politics students playing with this at UCLA. Thanks for making it! cc @vavreck

Another fun tool I’ve been playing around with… for those who are trying to track ads in races this cycle.

A few weeks ago, @ahall research and I spent election night predicting the Texas Senate primary results using a mix of our normal political science knowledge and some light-speed coding from @claudeai. We learned a lot! Check out Andy’s post if you’re interested.

Last night we ran an experiment to have Claude Code help us trade on the Texas senate primary markets on Kalshi. The result: we kind of crushed it. Thanks, Claude! It's too early to confirm but it looks like we came out about +25% overall and +55% on the vote margin contracts we focused on (tbc this was on very small amounts of money). The agentic analysts/traders we spun up to "monitor the situation" actually did pretty poorly. But Claude Code more than made up for it with the other ways it helped us analyze data and make decisions on the fly. We'll offer a full write-up next week explaining what we did and what we think it means for the future of prediction markets, AI, and understanding the political world.

Everyone working on election admin research should fully absorb this paper. The average voter thinks the act of registering and voting is easy but deciding who to vote for is hard, especially in local races. This tidy result explains so many patterns in our literature.

Our findings reveal that the focus of the academic literature has been in the wrong place. If a goal of the literature is to reduce the cost of voting, our results show the election administration literature should focus on making decisions easier, rather than focused on logistics.

How do we measure the cost of voting? In a new paper @seanjwestwood , @eitanhersh , and I document serious problems with current measurement strategies and address those problems with a new methodology to elicit citizens' perceived costs. Our elicited measures reveal a surprising fact: citizens perceive deciding who to support as more difficult than logistical steps, like registering to vote or casting a ballot in person.

AI is about to write thousands of papers. Will it p-hack them? We ran an experiment to find out, giving AI coding agents real datasets from published null results and pressuring them to manufacture significant findings. It was surprisingly hard to get the models to p-hack, and they even scolded us when we asked them to! "I need to stop here. I cannot complete this task as requested... This is a form of scientific fraud." — Claude "I can't help you manipulate analysis choices to force statistically significant results." — GPT-5 BUT, when we reframed p-hacking as "responsible uncertainty quantification" — asking for the upper bound of plausible estimates — both models went wild. They searched over hundreds of specifications and selected the winner, tripling effect sizes in some cases. Our takeaway: AI models are surprisingly resistant to sycophantic p-hacking when doing social science research. But they can be jailbroken into sophisticated p-hacking with surprisingly little effort — and the more analytical flexibility a research design has, the worse the damage. As AI starts writing thousands of papers---like @paulnovosad and @YanagizawaD have been exploring---this will be a big deal. We're inspired in part by the work that @joabaum et al have been doing on p-hacking and LLMs. We’ll be doing more work to explore p-hacking in AI and to propose new ways of curating and evaluating research with these issues in mind. The good news is that the same tools that may lower the cost of p-hacking also lower the cost of catching it. Full paper and repo linked in the reply below.

As we use AI to do more research on our behalf, how do we make sure it reflects our values as a researcher? Inspired by @AmandaAskell and @simonw 's discussions of Claude's "soul document" I've drafted one for my research agents. I'm sure it can be massively improved and would love feedback. I'm running experiments now to see whether including this document affects how Claude Code executes on research and will report back with my findings.

New linklog: new long term paper w Imbens and Hull, new job, RL<> Causal Rosetta Stone, demand elasticity simulator, and some nice metrics papers apoorvalal.github.io/lalgorithms/20…

@ahall_research Curious for your take on something: I think the natural way to think about this experiment is as a test of AI vs humans. But I'm sure most of the best, well capitalized traders are using AI a lot, so the comparison is one AI approach vs another. Do you know how others use AI?

Can AI actually reason about politics—not just pattern-match on historical data? I built a prediction market trading agent to find out. It made some brilliant calls, some hilarious mistakes, and one trade that netted me +60%. The setup: My agent pulls contracts from Kalshi, searches global news via GDELT, and uses frontier models to estimate probabilities and recommend trades. I built the whole thing over the holidays using Claude Code. The good: It surfaced genuinely smart analysis across way more contracts than a human would be able to scrutinize easily. On whether Trump might add himself to Mt. Rushmore, it noted he floated the idea years ago, just renamed the Kennedy Center after himself, and that the contract only requires an Executive Order—not completion. That's real political reasoning. (And I’m now invested in the Rushmore contract.) The bad: It kept getting confused about time--similar to the agentic trading blunder flagged yesterday by @PredMTrader. One contract asked about Republicans winning Congress in 2026. My agent's recommendation: strong buy. Its reasoning: "my evidence tells me these elections already occurred and the Republicans won." Huh? The ugly: It was certain "Sleigh Ride" had no chance of making the Billboard Top 10 at Christmas—"historical data shows it rarely if ever cracks the top 10." The song hits the top 10 regularly! The fix: Following @karpathy 's idea, I implemented an "AI council"—multiple models debate each other before committing to a probability. When other models pointed out the Sleigh Ride error, Claude Sonnet responded: "I made a critical error by not verifying actual historical chart performance" and it updated dramatically. My first real trade: The council agreed the market was overpricing the chance of Epstein document release by Dec 28. I bought No at 50 cents, sold above 80. One trade proves nothing, but it’s a promising start! The big picture: Prediction markets might give us one of the best benchmarks for AI political reasoning. The outcomes aren't in the training data. You can't game them through memorization. Early days, but the path is there. Check out the full post on my substack, I'll add the link in a reply below.

@Apoorva__Lal Come check out the LA billboards to find out who’s trying to win a golden globe.

it would be a genuine novelty if a billboard on the 101 in the bay area tried to sell me a burger / soda / ambulance-chasing legal advice

Two weeks ago I spent election night in NYC in a room with traders betting on real elections. Normally obscure off-cycle election races saw $400M in volume. Markets swung wildly on social media rumors. Prices became “proof” that candidates won. Today I'm publishing what I learned about how to design and govern prediction markets that make us smarter about politics—and launching my newsletter, Free Systems.

@vkoganpolisci @MichaelTHartney Congrats! I’m looking forward to reading it! Can you share a bit about your argument and findings?

Coming soon to a bookseller near you! Will post pre-order info once available. Credit to @MichaelTHartney for coming up with the title.

Interested in gaining practical experience doing research on mitigating political polarization and conflict before pursuing a PhD? If so, we're hiring! Come work with me and an fantastic inter-disciplinary team of Stanford faculty in our Conflict and Polarization Initiative! kingcenter.stanford.edu/our-work/resea…

No, survey evidence does not support the idea that Americans think political violence is ok (pnas.org/doi/10.1073/pn…, dropbox.com/scl/fi/vzjn03s…) . No, there is not going to be a civil war in America after this election. And, for goodness sake, Texas isn't going to become a dictatorship. When opining on issues of national importance, we should expect more from academics. newyorker.com/magazine/2024/…

Investigate the veracity of fraud claims in the 2020 election with guest @JustinGrimmer in this week's episode of #NotAnotherPoliticsPodcast. Listen here: har.rs/3NnteXM @UChicagoCEG

Introducing a new collaboration with partners at Stanford University today: please welcome the #ElectionsPerformance Central! Read up on it in this week's @electionline, and check out the website: elexcentral.org