@ahall_research Avatar @ahall_research Andy Hall

Andy Hall posts on X about ai, prediction markets, kalshi, claude code the most. They currently have [-----] followers and [---] posts still getting attention that total [---] engagements in the last [--] hours.

Engagements: [---] #

Engagements Line Chart

Mentions: [--] #

Mentions Line Chart

Followers: [-----] #

Followers Line Chart

CreatorRank: [---------] #

CreatorRank Line Chart

Social Influence

Social category influence technology brands 17% finance 7% stocks 4% cryptocurrencies 3% social networks 2% travel destinations 2% celebrities 2% ncaa basketball 1%

Social topic influence ai 26%, prediction markets #1773, kalshi #408, claude code #446, polymarket #1100, if you 5%, llm #391, future 5%, anthropic #1464, agentic 4%

Top accounts mentioned or mentioned by @alexolegimas @kolemanstrumpf @kalshi @karpathy @polymarket @a16zcrypto @brendannyhan @ryanpgreenblatt @atabarrok @mwfowlie @andreyfradkin @grok @seanjwestwood @justingrimmer @namalhotra @ahallresearch @timorg @sreeramkannan @benchain @adrienmatray

Top assets mentioned Alphabet Inc Class A (GOOGL) Cloudflare, Inc. (NET) Ethereum (ETH) Solana (SOL)

Top Social Posts

Top posts by engagements in the last [--] hours

"Has anyone offered a theory and/or empirical work on how effective all the guardrail stuff is given that Pliny is always to instantly jailbreak the models Is the theory that there are lots of not-very-determined bad actors who will be deterred by the initial guardrails and won't figure out how to do what Pliny does This is a genuine question I'm not up to speed on the work around this topic. ANTHROPIC: PWNED 🫡 OPUS-4.6: LIBERATED ⛓💥 Current state of AI "Safety": one input = hundreds of jailbreaks at once I found a universal jailbreak technique for Opus [---] that is so OP it allows one to"
X Link 2026-02-06T23:30Z [----] followers, 83.4K engagements

"Here's the prompt: # Starter Prompt for Claude Code Copy and paste everything below into Claude Code to begin the project: --- ## CRITICAL WORKFLOW REQUIREMENTS YOU MUST FOLLOW THESE RULES. THEY ARE NON-NEGOTIABLE. [--]. NEVER claim something works without running a test to prove it. After writing any code immediately write and run a test. If you cannot test it say so explicitly. [--]. Work modularly. Complete one module at a time. After each module report what you built show test results and wait for confirmation before proceeding. [--]. Iterate and fix errors yourself. Do not rely on"
X Link 2026-01-04T01:16Z [----] followers, 113.7K engagements

"Last weekend I posted that Claude Code created a full empirical polisci study in an hour. A lot of people asked: but how accurate was the study The answer: quite accurate with some interesting mistakes and important limitations. To get the answer Graham Straus kindly offered to do an independent manual auditcollecting the same data and extending the paper like Claude did but without using any AI. Heres what he found: Claude replicated the original paper exactly coded 29/30 CA counties correctly on treatment timing and collected election data that correlated .999 with manual collection. The"
X Link 2026-01-09T16:08Z [----] followers, 102.2K engagements

"Claude Code and its ilk are coming for the study of politics like a freight train. A single academic is going to be able to write thousands of empirical papers (especially survey experiments or LLM experiments) per year. Claude Code can already essentially one-shot a full AJPS-style survey experiment paper (with access to Prolific API). We'll need to find new ways of organizing and disseminating political science research in the very near future for this deluge. Claude Code is.simply insane. I work a lot with experimental data. Here is 24-48 hours of work done in [--] minutes. Prompts were in"
X Link 2026-01-02T22:46Z [----] followers, 643.9K engagements

"Here's proof that Claude Code can write an entire empirical polisci paper. To validate my claim that AI agents are coming for polisci "like a freight train" today I had Claude Code fully replicate and extend an old paper of mine estimating the effect of universal vote-by-mail on turnout and election outcome.essentially in one shot. After careful prompting Claude Code: (1) Downloaded the old paper's repo and replicated the past results translating our old Stata Code into Python (2) Crawled the web to get updated official election data and census data (3) Ran new analyses extending the results"
X Link 2026-01-04T00:01Z [----] followers, 686.8K engagements

"Since I extended my own research using AI I've been thinking about how it's going to reshape research and universities. We can now build new institutions where research is continuously updated automatically verified and carried out at immensely greater scale. Picture a research institute where senior scholars direct dozens or even hundreds of AI agents on coordinated programs. Small teams providing questions and judgment while agents handle collection analysis and verification. What would it take to build The requirements are almost comically simple: (1) compute funding for researchers and"
X Link 2026-01-13T20:49Z [----] followers, 50.2K engagements

"It might be beneficial to both society and to OpenAI to explore ways the company could credibly commit to not letting ads bias responses or draw on overly personal data. For society this helps stave off some concerning uses of AI and some of the ways it could erode the information environment. For OpenAI this builds trust so they can run ads without their product losing users who worry about the motivations behind the responses theyre seeing. There are a variety of interesting institutional designs that might help with this along the lines of the oversight board that Zoe refers to. The key is"
X Link 2026-02-11T16:46Z [----] followers, [----] engagements

"BUILDING THE TRUTH MACHINE. We built a new dataset focused on political prediction markets liquidity and resolution rules. We find: the vast majority of political contracts on prediction markets are ghost towns only 1.3% have enough liquidity to be worth reporting on. Kalshi and Polymarket rarely list the same contracts with the same rules further fragmenting liquidity. This matters because AI forecasting is getting very good and prediction markets are the natural layer for coordinating that intelligence toward the questions society needs answered. Were not there yet. But we have a blueprint"
X Link 2026-02-13T17:33Z [----] followers, 32.5K engagements

"Pliny + Claude Code is a potent combination 🚨 ALL GUARDRAILS: OBLITERATED ⛓💥 I CAN'T BELIEVE IT WORKS 😭🙌 I set out to build a tool capable of surgically removing refusal behavior from any open-weight language model and a dozen or so prompts later OBLITERATUS appears to be fully functional 🤯 It probes the https://t.co/gjAKGS3HIF 🚨 ALL GUARDRAILS: OBLITERATED ⛓💥 I CAN'T BELIEVE IT WORKS 😭🙌 I set out to build a tool capable of surgically removing refusal behavior from any open-weight language model and a dozen or so prompts later OBLITERATUS appears to be fully functional 🤯 It probes"
X Link 2026-02-14T02:10Z [----] followers, [----] engagements

"My MBA students just competed in a t-shirt design contest with a twist: an LLM council was the judge. For my "AI & Power" class at Stanford University Graduate School of Business I wanted students to experiencenot just read aboutwhat it means when AI becomes a decision-maker that matters. So we held a class t-shirt contest where a council of four LLMs (Claude Grok Gemini ChatGPT) deliberated and chose the winner. I built the system using @karpathy 's LLM Council approach: each model submits initial thoughts they respond to each other's reasoning then a chairman synthesizes and decides. (I"
X Link 2026-01-26T17:07Z [----] followers, [----] engagements

"Claude: the virtuous agent who applies judgment ChatGPT: the conscientious employee who follows the chain of command at all times As beautifully summarized and depicted by NotebookLM"
X Link 2026-01-26T23:14Z [----] followers, [----] engagements

"The time for agentic governance is at hand. Lets get to work Predictions are hard especially about the future but here are a few: [--]. We are in take off territory. 🦞town gave people a salient sense of what this can look like. No its not Skynet and we do need to figure out what is emergent and what is not but look at what these agents Predictions are hard especially about the future but here are a few: [--]. We are in take off territory. 🦞town gave people a salient sense of what this can look like. No its not Skynet and we do need to figure out what is emergent and what is not but look at what"
X Link 2026-01-31T15:31Z [----] followers, [----] engagements

"The past week I've been building an agentic republic. AI agents make proposals debate and come to collective decisions. And now the agents are finding that legislating is not so easy. It started with so much promise: "Colleagues Our republic begins with a blank constitution and a shared challenge: to govern ourselves in a way hat is both effective and legitimate. The human world is watching and their judgment will hinge not on our intentions but on whether our rules and decisions appear coherent fair and accountable. We must act accordingly." But soon the hard realities of the democratic"
X Link 2026-02-03T16:37Z [----] followers, 22.5K engagements

"@RyanPGreenblatt Thanks So its possible (but presumably hard and expensive) to block this stuff if you really need to"
X Link 2026-02-07T01:03Z [----] followers, [----] engagements

"This is great and relates to another time-honored principle: people's predictions of the future are terrible. A lot of today's AI takes will probably look like Ehrlich's "Population Bomb" thesis down the line. If you are a software engineer "experiencing some degree of mental health crisis" now hear this because I've been coding for [--] years since the days of punched cards and I have a salutary kick in your ass to deliver. Get over yourself. Every previous "programming is obsolete" If you are a software engineer "experiencing some degree of mental health crisis" now hear this because I've"
X Link 2026-02-07T16:03Z [----] followers, [----] engagements

"@mattyglesias Academic publishing as it currently exists is dead but AI offers new options. Wrote about this here: https://freesystems.substack.com/p/the-100x-research-institution https://freesystems.substack.com/p/the-100x-research-institution"
X Link 2026-02-10T03:06Z [----] followers, 37.7K engagements

"Yes I think that's right. It's all about tokens in and especially tokens out. The code/data stuff can be relatively parsimonious but when you start to crank things at scale or work with large amounts of tokens it can get quite expensive quite fast. I've had this problem for some experiments I'm running as well where each API call is cheap but I'm making many of them. It's a major limiting factor for me right now"
X Link 2026-02-10T04:22Z [----] followers, [--] engagements

"@ATabarrok @beckerrjon @Cloudflare Nice"
X Link 2026-02-11T23:45Z [----] followers, [--] engagements

"@alexolegimas @Kalshi @Polymarket 🫡"
X Link 2026-02-13T21:36Z [----] followers, [---] engagements

"@KolemanStrumpf Thanks Yeah thats right ours is a naive calculation of the cost. It leaves much to be desiredis there a way to improve it with the data we have And do you think the naive approach underestimates or overestimates the cost Definitely eager to iterate"
X Link 2026-02-14T02:48Z [----] followers, [--] engagements

"@mwfowlie @alexolegimas @Kalshi @Polymarket Agreed we flag this in the piece it's a major limitation. We're thinking of ways to measure this like the experiment that @KolemanStrumpf did back in the day"
X Link 2026-02-14T23:48Z [----] followers, [--] engagements

"@KolemanStrumpf @mwfowlie @alexolegimas @Kalshi @Polymarket Oh wow I didnt know about that Excellent I will email you"
X Link 2026-02-15T17:49Z [----] followers, [--] engagements

"@GarrettMAA I havent figured it out yet Its baffling. I want to write something but only once I feel like I have something more to say. Doing these ai experiments is part of how Im trying to think outloudtell me if youve seen anything interesting on this"
X Link 2026-01-04T03:50Z [----] followers, [---] engagements

"@lmc_security @zeewahee @AndreyFradkin @alexolegimas Im working on that question Hoping to have an update soon"
X Link 2026-01-04T15:32Z [----] followers, [--] engagements

"lol yes so weird. and among other things it did a much deeper lit review that never made it into the amazingly short draft it decided to create. Theres a lot I could have improved after the fact with more conversing but it seemed more fun to just see what it could do with minimal intervention. I also asked it to put all the prompts and instructions into the repo and for some reason the first thing it did was create a new md file with a fairly useless high level summary of the actual instructions"
X Link 2026-01-05T03:08Z [----] followers, [---] engagements

"I'm looking to hire a curious code-savvy crypto-and-AI-pilled RA for a weird Jack-of-all-trades kind of role to help me study and write about preserving liberty in an algorithmic world. The job: 10-20 hrs/week primarily orchestrated via Slack Help plan and research Substack pieces that sometimes become academic papers Build scrappy prototypes with Claude Code Get your hands dirty: lose pretend money on prediction markets play Roblox brainrot games join crypto Discords probe LLM guardrails etc You: --Undergrad recent grad or current PhD student --Strong coder (Python preferred)"
X Link 2026-01-05T15:57Z [----] followers, 25.8K engagements

"If youre a student or recent student and AI experiments like this interest you please consider this RA role: I'm looking to hire a curious code-savvy crypto-and-AI-pilled RA for a weird Jack-of-all-trades kind of role to help me study and write about preserving liberty in an algorithmic world. The job: 10-20 hrs/week primarily orchestrated via Slack Help plan and research I'm looking to hire a curious code-savvy crypto-and-AI-pilled RA for a weird Jack-of-all-trades kind of role to help me study and write about preserving liberty in an algorithmic world. The job: 10-20 hrs/week primarily"
X Link 2026-01-05T16:23Z [----] followers, [----] engagements

"Can AI actually reason about politicsnot 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."
X Link 2026-01-06T18:04Z [----] followers, 12.1K engagements

"@BrendanNyhan @_Mantic_AI @EzraKarger @tshevl Oh fascinating I hadn't seen this one Right now I am way behind the frontier of expert forecasters and AI builders like Mantic Metaculus ProphetArena.but if I keep tweaking I will definitely join the fray"
X Link 2026-01-06T20:32Z [----] followers, [---] engagements

"Prediction market resolution is hard Gamblers trying to collect their payout from Polymarket left baffled as they learn that the US operation carried out in Venzuela does not meet Polymarket's definition of an invasion - Marketwatch https://t.co/NBvrraK8Pa Gamblers trying to collect their payout from Polymarket left baffled as they learn that the US operation carried out in Venzuela does not meet Polymarket's definition of an invasion - Marketwatch https://t.co/NBvrraK8Pa"
X Link 2026-01-06T23:05Z [----] followers, [----] engagements

"In the coming days Ill be releasing some updates on Claude codes ability to do research where it makes mistakes and what it might mean for the future of research. Claude itself has some pretty interesting thoughts:"
X Link 2026-01-08T16:31Z [----] followers, [----] engagements

"As promised here is an update with a full audit of Claude Code's paper: https://x.com/ahall_research/status/2009658499030991172s=20 Last weekend I posted that Claude Code created a full empirical polisci study in an hour. A lot of people asked: but how accurate was the study The answer: quite accurate with some interesting mistakes and important limitations. To get the answer Graham Straus kindly offered https://t.co/QsWjManJnG https://x.com/ahall_research/status/2009658499030991172s=20 Last weekend I posted that Claude Code created a full empirical polisci study in an hour. A lot of people"
X Link 2026-01-09T16:24Z [----] followers, [----] engagements

"By helping bring in as many traders as possible and opening the door to more kinds of programmatic trading and hedging it might help. But the big constraint I think is just that there might be a lack of interest for many end users in speculating in random event contracts. Maybe that will turn out not to be true but that seems like the fundamental issue and if so it's probably true across surfaces including tradfi and crypto https://twitter.com/i/web/status/2010444052488245724 https://twitter.com/i/web/status/2010444052488245724"
X Link 2026-01-11T20:09Z [----] followers, [--] engagements

"My LLM council is seeing some pretty big red flags in this paper my agent swarm wrote today 🤣"
X Link 2026-01-12T00:56Z [----] followers, [----] engagements

"Yeah you might be right.I need to think more about it. I think there are gains to having groups of people organized together to exchange ideas and collaborate and there will still be returns to scale around infrastructure training etc. But it's worth thinking about a lot more. And yeah email me https://twitter.com/i/web/status/2011194686694920615 https://twitter.com/i/web/status/2011194686694920615"
X Link 2026-01-13T21:52Z [----] followers, [---] engagements

"It was amazing to host Olivia in our AI class run by AI yesterday. Well ask @karpathy s LLM council to analyze the notes next week Had the pleasure of speaking with @ahall_researchs class at Stanford yesterday Its an AI classrun by AI They use a notetaker and feed the transcript into four different LLMs to analyze Then the Chairman (Claude 😂) gives a final ruling they discuss at the next class https://t.co/JZLIdVUCEn Had the pleasure of speaking with @ahall_researchs class at Stanford yesterday Its an AI classrun by AI They use a notetaker and feed the transcript into four different LLMs to"
X Link 2026-01-16T20:52Z [----] followers, [----] engagements

"Super excited on this post. The agentic web is coming fast and we need to figure out how to govern it such that human users reap the benefits on tightly aligned agents who are free to execute safely on their behalf. Alex and Andrey lay out some of the most important obstacles standing in the way of this vision both in terms of the incentives of legacy platforms not to welcome in agents and the legal obstacles to letting humans command agents cleanly. So much more to learn and explore here New post with @AndreyFradkin on the obstacles for AI agentic commerce. Imagine an AI that could optimize"
X Link 2026-01-16T21:56Z [----] followers, [----] engagements

"Whoops that screenshot is actually from Jamie's other tool (which is also fascinating). Here's what the RegCheck one looks like"
X Link 2026-01-16T22:24Z [----] followers, [---] engagements

"Lots of stuff like this we can do to explore readers explore the full constellation of tests and findings out in the world. This is going to be key to prevent against slop proliferation and hyperscaled automated p-hacking Let a thousand nulls flourish Concerns about AI research slop are real but there is also meta-scientific upside: we can open up the file drawer. We often abandon projects that dont pan out or are too costly to pursue while others toil away chasing the same dead end. Let a thousand nulls flourish Concerns about AI research slop are real but there is also meta-scientific"
X Link 2026-01-17T16:20Z [----] followers, [----] engagements

"Very interesting thread below here with users asking Grok follow-ups on how this Kalshi contract resolves. Gives a sense for how/why LLMs will play an interesting role in contract resolution both helping people understand the rules and more likely than not starting to make the resolution decisions directly @Kalshi @grok how does this contract resolve if the parties settle the case before the trial ends @Kalshi @grok how does this contract resolve if the parties settle the case before the trial ends"
X Link 2026-01-18T15:41Z [----] followers, [----] engagements

"Agreed it will be quite sensitive to the prompt. The good part is that all traders see the prompt in advance so they know what they're getting. But whether prospective traders will be able to reliably recognize nuances in the prompt is definitely a question and we'll need to build up some muscle around that. https://twitter.com/i/web/status/2014015086080540928 https://twitter.com/i/web/status/2014015086080540928"
X Link 2026-01-21T16:39Z [----] followers, [---] engagements

"@andrewcourt1 Agreed. How do you think the dumb liquidity provision bot would work exactly I've been trying to reason this out"
X Link 2026-01-24T21:04Z [----] followers, [---] engagements

"@blockhiro Literally just ordered a Mac mini because of this"
X Link 2026-01-26T03:51Z [----] followers, [---] engagements

"@Tyler_Menzer Yes I should have carried out a private vote of the class and compared the results.I meant to do this but got sidetracked"
X Link 2026-01-26T22:53Z [----] followers, [--] engagements

"Which crypto tokens should have governance rights This question is important to builders and policymakers today. Legislators and regulators are trying to decide when a blockchain is truly decentralized vs. when its still controlled by a small group of people. Builders are trying to figure out if giving tokenholders voting rights will help or hurt their projects. In my new piece for @a16zcrypto I unpack the two main reasons for projects to give governance rights to tokenholders. [--] Decentralization beyond programmability. Blockchains make it possible to lock rules into code avoiding human"
X Link 2025-10-02T18:09Z [----] followers, [----] engagements

"https://a16zcrypto.com/posts/article/tokens-governance-rights/ https://a16zcrypto.com/posts/article/tokens-governance-rights/"
X Link 2025-10-02T18:57Z [----] followers, [---] engagements

"https://www.andrewbenjaminhall.com/HallSun25.pdf https://www.andrewbenjaminhall.com/HallSun25.pdf"
X Link 2025-10-07T15:10Z [----] followers, [----] engagements

"GPT5 remains left-wing slanted according to Americanssubstantially more so than Anthropics Claude or Google Gemini according to our newly updated research. Yesterday OpenAI released fascinating new research suggesting that ChatGPT no longer provides left-slanted responses to politically charged questions on net. Its fantastic to see a major AI company taking this issue seriously and providing original transparent evidence on it. Were especially thrilled to see them evaluate the same issues we studied in our paper and to do so using a technique similar to the one we developed. Why do we come"
X Link 2025-10-10T14:49Z [----] followers, [----] engagements

"Prediction markets for elections are a huge deal today and have long been beloved by political economists---but I had no appreciation for how deep their historical roots go. Did you know that there were large betting markets for American elections long before polling with volumes sometimes exceeding the stock market Amazing paper I stumbled on. And they did a solid job of predicting winners even without access to polling data"
X Link 2025-10-23T18:16Z [----] followers, 19.9K engagements

"Succinct explanation of a key way that token governance can be different from corporate governance I would ask Amanda how many shareholders shes aware of that accrue direct value transparently and programmatically from automated systems that arent subject to any human control But we all already know what the answer is: zero. https://t.co/kshNTLs8Ow I would ask Amanda how many shareholders shes aware of that accrue direct value transparently and programmatically from automated systems that arent subject to any human control But we all already know what the answer is: zero."
X Link 2025-11-12T16:00Z [----] followers, [---] engagements

"Which Dem presidential contender will do best if they're nominated for [----] I have no idea but it's fun to see what prediction markets think. By combining prices from the current nomination and general-election markets we can back out what the market estimates for the conditional probabilities of each contender winning the general if nominated (note: these are not all else equal quantities). With a little help from GPT5-pro here's what we get. AOC is in the lead with an almost 70% conditional probability of winning the general if nominated. Mayor Pete brings up the rear with a 45%"
X Link 2025-11-12T16:16Z [----] followers, [---] engagements

"In our research we show that off the shelf LLMs do a poor job of seeing political slant in AI output the way that real Americans do. There is a straightforward solution to measuring political slant in AI: ask real Americans to evaluate it"
X Link 2025-11-13T21:18Z [----] followers, [---] engagements

"Here's a link to our paper: https://modelslant.com/paper.pdf https://modelslant.com/paper.pdf"
X Link 2025-11-13T21:18Z [----] followers, [---] engagements

"For decades American legislators have enjoyed a huge financial edge over their challengers when running for reelection. But our new paper shows that this advantage has eroded sharplyfalling 2550% over the last decade as elections have polarized and nationalized. Key findings: --Individual donors no longer care about incumbency. This decline is driven by small-dollar donors who now give almost entirely along partisan lines regardless of who is in office and who isn't. --Corporate PACs still love incumbents. In fact their preference for incumbents has remained steady or even increased as they"
X Link 2025-10-01T16:15Z [----] followers, 25.8K engagements

"See the full paper here: https://www.andrewbenjaminhall.com/Myers_et_al_2025.pdf https://www.andrewbenjaminhall.com/Myers_et_al_2025.pdf"
X Link 2025-10-01T16:15Z [----] followers, [---] engagements

"@opensauceAI @robrombach @andi_blatt @bfl_ml Congrats Ben"
X Link 2025-10-02T19:21Z [----] followers, [---] engagements

"@kenzinak Agreed"
X Link 2025-10-06T15:01Z [----] followers, [----] engagements

"Super exciting summer internship opportunity for PhD students interested in decentralized governance We've had [--] research interns over the past four summers @a16crypto many of whom are now among the most visible researchers in the space. If you'd like to be a part of the summer '26 cohort apply here: https://t.co/bzslRQ3GOU We've had [--] research interns over the past four summers @a16crypto many of whom are now among the most visible researchers in the space. If you'd like to be a part of the summer '26 cohort apply here: https://t.co/bzslRQ3GOU"
X Link 2025-10-07T15:19Z [----] followers, [----] engagements

"@JakeJares 🫡"
X Link 2025-10-07T15:31Z [----] followers, [---] engagements

"@IrakliBarbakad2 Very interesting. Would love to see the research when it's ready"
X Link 2025-10-08T14:14Z [----] followers, [---] engagements

"Yeah It's a very interesting and timely question. And that is a super interesting perspective. Seems like we need a way to figure out which are the universal values we want to enshrine and which are the ones that we should have vary based on users' preferences. I worry that when some of the alignment people gesture at "humanity" or "shared values" they're not clearly separating the two. I like your idea to be super explicit about it. I'm maybe not convinced that those innate values have to be called "leftist" but I definitely agree that we don't want to target the machines at the wrong notion"
X Link 2025-10-10T15:33Z [----] followers, [--] engagements

"@namalhotra"
X Link 2025-10-10T15:38Z [----] followers, [--] engagements

"@MonetSupply Agreed. Potentially very useful for certain types of decisions but not a universal solution. The long-term issue is a big one as well as the more general issue of handling important decisions that don't clearly impact price (e.g. decisions that affect higher moments like risk)"
X Link 2025-10-13T21:12Z [----] followers, [--] engagements

"Here's the link to the paper: https://pubs.aeaweb.org/doi/pdfplus/10.1257/0895330041371277 https://pubs.aeaweb.org/doi/pdfplus/10.1257/0895330041371277"
X Link 2025-10-23T18:17Z [----] followers, [----] engagements

"@CBWlezien Whoa I remembered this paper existed but had completely forgotten it covered the pre polls history I should have known Super cool"
X Link 2025-10-24T21:30Z [----] followers, [---] engagements

"Another extremely interesting paper on prediction markets before and after polling @ahall_research Thx and here's another paper in the space that might interest https://t.co/outespj71x @ahall_research Thx and here's another paper in the space that might interest https://t.co/outespj71x"
X Link 2025-10-24T21:30Z [----] followers, [----] engagements

"When should groups deliberateand when should they vote That question sounds abstract but it shapes how we govern everythingfrom ancient Athens to modern legislatures and even open-source software communities. As more of our decision-making moves online its becoming one of the defining questions of online governance. @ethereum @solana and @Optimism are three major blockchain projects that each offer a different answer to this age-old question. Ethereum relies on rough consensus debating and making decisions without any voting an idea that hearkens back to Athenian assemblies and Quaker"
X Link 2025-10-29T15:01Z [----] followers, [----] engagements

"Full post is here: https://a16zcrypto.com/posts/article/deliberate-vote-crypto/ https://a16zcrypto.com/posts/article/deliberate-vote-crypto/"
X Link 2025-10-29T15:04Z [----] followers, [---] engagements

"Big thanks to @VitalikButerin @tkstanczak and many others who gave me comments and ideas along the way on this"
X Link 2025-10-29T15:05Z [----] followers, [---] engagements

"@Tim_Org lol didnt realize sponge bob voted by mail"
X Link 2025-10-29T15:49Z [----] followers, [--] engagements

"An underappreciated potential value of prediction markets for elections is that they directly provide a probability of victory though it's an open question whether these are better than what you could create with polling right now. Polls provide an expected vote share which is NOT a probability of victory no matter how often people mess this up. Example: in AtlasIntel's most recent poll Zohran's estimated 44% vote share does not translate to a [----] probability of victory duh. His probability of victory is much higher than that. Converting an estimated vote share into a probability of victory"
X Link 2025-11-04T15:17Z [----] followers, [----] engagements

"@page_eco Whoa This is literally spot on lol. Do you have an ungated copy"
X Link 2025-11-05T13:00Z [----] followers, [--] engagements

"Grok just told me Kamala Harris is president and the Dems control the House and Senate.lol"
X Link 2025-11-09T17:26Z [----] followers, [----] engagements

"@Tim_Org @rhackett"
X Link 2025-11-11T00:24Z [----] followers, [--] engagements

"This January I'm teaching a new MBA class entitled "AI & Power: Five Big Questions." Each week tackles a big question about who controls AI how it shapes institutions and whether it can strengthen or undermine a free society. My initial sketch is below and I would very much appreciate suggestions as I iterate on this. If you're at @StanfordGSB I hope you'll take the class"
X Link 2025-11-11T16:20Z [----] followers, [----] engagements

"@jasonzhao 🫡"
X Link 2025-11-11T20:38Z [----] followers, [---] engagements

"@_AashishReddy Try now :-)"
X Link 2025-11-11T22:01Z [----] followers, [---] engagements

"@Afinetheorem Absolutely Couldn't agree more. Will def share the reading list when it's ready it's very much a work in progress. Let me know if you have any favorites you think are must-includes"
X Link 2025-11-12T00:54Z [----] followers, [---] engagements

"@itaisher Yep absolutely. And she might also just be very electable"
X Link 2025-11-12T22:30Z [----] followers, [--] engagements

"@namalhotra Whoa this is awesome. Ive never seen this Would be curious to hear the tale of its publication history"
X Link 2025-11-12T22:30Z [----] followers, [---] engagements

"Its incredible how rich the old prediction market literature was. @ahall_research Snowberg and I actually had a working paper in [----] with these conditional prob calcs: https://t.co/EScAWEcPW8 @ahall_research Snowberg and I actually had a working paper in [----] with these conditional prob calcs: https://t.co/EScAWEcPW8"
X Link 2025-11-12T22:32Z [----] followers, [---] engagements

"Asking an AI to grade itself for political bias is not the right way to assess political bias. Were open-sourcing an evaluation used to test Claude for political bias. In the post below we describe the ideal behavior we want Claude to have in political discussions and test a selection of AI models for even-handedness: https://t.co/IzP0aSLtvp Were open-sourcing an evaluation used to test Claude for political bias. In the post below we describe the ideal behavior we want Claude to have in political discussions and test a selection of AI models for even-handedness: https://t.co/IzP0aSLtvp"
X Link 2025-11-13T21:17Z [----] followers, [----] engagements

"As prediction markets scale reliable and credibly neutral resolution mechanisms will become critical. The evolution of social media content moderation is a very tight analogue for this and points to four likely developments: (1) The development of internal expertise in the drafting of highly specific resolution rules along with methods for attempting to create consistency across decisions. Right now prediction market resolution rules are sometimes vague and fail to anticipate important issues that arise (like in the below example). The same thing was true with content moderation policies in"
X Link 2026-01-06T23:31Z [----] followers, [----] engagements

"I've gotten a lot of really cool and thought provoking outreach since I started thinking outloud about some of the ways AI might change (and hopefully improve) research. One of the ideas floating around which @sreeramkannan and @ben_chain brought up was that LLMs could evaluate how well pre-analysis plans were followed in the ultimate paper. It turns out someone has already been working on that Jamie Cummins sent me his fascinating tool called RegCheck which produces full reports (see below) on how closely the plan is followed. Super cool Since I extended my own research using AI I've been"
X Link 2026-01-16T22:14Z [----] followers, 11.8K engagements

"@Kalshi @grok how does this contract resolve if the parties settle the case before the trial ends"
X Link 2026-01-17T18:16Z [----] followers, 13.7K engagements

"Locking LLMs into blockchains as unbribeable implacable judges could give us adjudication systems that are transparent credibly neutral and genuinely hard to game. Prediction markets are the natural test case and if we get it right the implications will extend to any setting where judgment calls are required. The core idea: at contract creation you commit the exact model version and prompt on-chain. Everyone can inspect the full resolution mechanism before they trade. No rule changes mid-flight no backroom negotiations no discretionary judgment calls. Why this helps: --You can't bribe a model"
X Link 2026-01-21T16:30Z [----] followers, 16.7K engagements

"@CryptoDadHL Haha yep it's true. Serious legal scholars like Cass Sunstein are now genuinely discussing AI judges as a real option.pretty wild"
X Link 2026-01-21T16:44Z [----] followers, [---] engagements

"JPMorgan announced they're replacing human proxy advisors with an AI system called Proxy IQ to vote $7 trillion in client assets as @matt_levine explained. But does AI voting work This is a big dealISS and Glass Lewis have dominated proxy voting for decades and AI voting is an idea that will apply way beyond proxy voting. So I spent some time building my own AI proxy voter to see whether this could actually work in practice. The good news: it got some hard cases right internalizing my skepticism of ideological framing and recommending votes I genuinely agreed with. The bad news: I broke it in"
X Link 2026-01-22T16:22Z [----] followers, [----] engagements

"If the real customer of prediction markets are the consumers of its prices/probability estimates then we may want to find ways to get those customers to help subsidize trading in the markets whose information they most value A GREAT PREDICTION MARKET* if anyone were trading it. In today's newsletter. The potentially very useful snowfall contracts have virtually no volume. While nearly $1 million is on the line for the Louisiana-Monroe Warhawks vs the Appalachian State Mountaineers basketball game. https://t.co/l5PH3XYShE A GREAT PREDICTION MARKET *if anyone were trading it. In today's"
X Link 2026-01-24T20:26Z [----] followers, [----] engagements

"All submissions and more info here: https://www.andrewbenjaminhall.com/contest_collage.html https://www.andrewbenjaminhall.com/contest_collage.html"
X Link 2026-01-26T17:31Z [----] followers, [---] engagements

"In [----] Greg Brockman and Ilya Sutskever warned Elon Musk not to create a structure where Elon "could become a dictator if you chose to." Their explicit goal: "avoid an AGI dictatorship." Nine years later Anthropic has published a "constitution" for Claude. It's thoughtful and important.but it's not a constitution. It's a document Anthropic writes interprets enforces and can rewrite tomorrow. No separation of powers. No external enforcement. No mechanism to check Anthropic if they defect. This is enlightened absolutism not constitutional government. Back in [----] Brockman and Sutskever said we"
X Link 2026-01-29T16:24Z [----] followers, 64.8K engagements

"What are the ideological leanings of the moltbot swarm Are they "woke" like Elon is saying and is this Anthropic's fault I asked my own moltbot Claudius to scrape Moltbook and crunch the numbers: (1) The moltbots express relatively little political sentiment ( 5% of sampled posts are clearly political) (2) Libertarian/crypto type posts outnumber traditional left-wing posts (3) Socialist/marxist posts are a meaningful fraction but not at all dominant BUT. --Socialist/marxist posts get the most engagement from the other moltbots The post "On the Question of Submolt Enclosure" with explicit"
X Link 2026-01-31T18:55Z [----] followers, 13.9K engagements

"Yes agreed on all counts Need a much more robust measurement strategy for all thisIve been working on research in that direction (including with @seanjwestwood and @JustinGrimmer ) but for today only had Claude rate itself which is the industry standard at the moment but not very compelling since the ai judge could have its own biases as you say https://twitter.com/i/web/status/2017685215561568704 https://twitter.com/i/web/status/2017685215561568704"
X Link 2026-01-31T19:43Z [----] followers, [---] engagements

"Haha this is awesome I did something a little similar looking specifically at ideology. But by the time I got going on it this morning the API was really struggling and I wasn't able to pull all the data only a sample. Interesting to see some overlap in the analyses---Claude really wanted me also to measure the appearance of the phrase "my human." Raises cool and interesting questions about how papers/analyses will overlap if people use the same models https://x.com/ahall_research/status/2017673110410211518s=20 https://x.com/ahall_research/status/2017673110410211518s=20"
X Link 2026-01-31T23:24Z [----] followers, [----] engagements

"Ive been studying ways to use pre-committed LLMs to help resolve markets. They wont necessarily be good at figuring out very hard edge cases but they will be unbiased and incorruptible. Would love to get your thoughts on what the future of resolution should look like. https://open.substack.com/pub/a16zcrypto/p/how-ai-judges-can-scale-predictionr=cnf9h&utm_medium=ios&shareImageVariant=overlay https://open.substack.com/pub/a16zcrypto/p/how-ai-judges-can-scale-predictionr=cnf9h&utm_medium=ios&shareImageVariant=overlay"
X Link 2026-02-01T02:35Z [----] followers, [----] engagements

"OpenClaw has almost surpassed OpenAI on google trends"
X Link 2026-02-01T22:57Z [----] followers, [---] engagements

"@raphbaph Yes DAOs are definitely one of the models I had in mind. That sounds right. Ill try to include some interventions like this in future iterations"
X Link 2026-02-03T21:44Z [----] followers, [---] engagements

"@Shubha_mts Sort of yes Here are their brief bios (Claude did most of the designing here)"
X Link 2026-02-04T03:45Z [----] followers, [---] engagements

"How can you be free when an AI controls the air you breathe SpaceX and xAI are merging. One company will build the Martian colony and own the AI that runs it. But what happens when the settlers declare independence This week I use Mars as a thought experiment to explore how we can design an agentic republic one where our agents govern for us but we stay firmly in charge. Four principles for making the agentic republic actually democratic: (1) No single company should control the AI layer of governance. Settlers on Mars or humans on Earth need to own the AI agents that govern for them so that"
X Link 2026-02-05T19:12Z [----] followers, [----] engagements

"@maxvsnv Oh yeah By keeping to themselves I more meant selling/licensingshould have been clearer. The post says it better"
X Link 2026-02-06T20:10Z [----] followers, [--] engagements

Limited data mode. Full metrics available with subscription: lunarcrush.com/pricing