Even as Claude Fable 5.2 sets records on major coding and capabilities benchmarks, we find that without tools it still miscalculates taxes, credits and benefits about a third of the time for the households that owe or receive them.
Social feed
All my posts from across the web, aggregated in one open source location. Bluesky posts update at build time. X and LinkedIn posts are imported from exports.
My colleague David Trimmer (@dtrim99.bsky.social) checked each state's law. For example, Oregon exempts the share of the pay from service before October 1991 and taxes the rest. Georgia doubles its $17,500 exemption for a retiree under 62 who earns more than $17,500 in the state.
Method and DC forecasts: maxghenis.com/blog/dc-init...
Texas prompts and answers: github.com/MaxGhenis/te...
For same-day races the votes couldn't change, Denver's foie gras ban and Texas's Senate race, the lines nearly meet at 50%. I told the DC forecasters, not the Texas ones, how such a race should behave. DC passed the ban with 73%, and Hinojosa won by 7 points; the forecasters had given her 26%.
Before polls closed Tuesday, I asked AI forecasters what would follow a one-point win and a one-point loss. They expect passing DC's foie gras ban to make further farmed-animal measures likelier, and a Gina Hinojosa win for Texas governor to delay the restart of approvals for large data centers.
My colleague David Trimmer (@dtrim99.bsky.social) built our dashboard of seven tax and benefit measures on ballots today. For example, if Colorado voters pass Amendment 87's graduated income tax, returns with taxable income above $510,834 owe more and we estimate 2.4 million owe less.
Tomorrow, voters decide 158 ballot measures that have prediction markets. On Oct 19 I registered a test that pools the 21 states with a qualifying muni fund. In practice it rests on six revenue measures in California and Colorado, and I expect an inconclusive result.
github.com/MaxGhenis/ba...
I found no free daily price series for testing Washington's income tax or Florida's property-tax cut, and no fund that mainly holds either state's munis. The Municipal Securities Rulemaking Board bars automated access to EMMA, its record of individual bond trades.
In the markets I studied, traders moved the odds more before legislative votes than during them. Kalshi's market on Washington's income tax rose 22 points over the weekend before the state House took it up, and 7 across the vote. A study of the vote itself would have missed the 22.
A swing in Kalshi's Prop 40 odds from 0% to 100% went with +0.8 basis points in the yield gap between California and national muni funds (95% range −3.5 to +5.0). The same swing in @polymarket.com's went with −0.5 (−3.8 to +2.9). I'd catch an effect of 5 to 7 bp four times in five.
Californians vote tomorrow on Prop 40, the billionaire tax. Over the past year, prediction markets put its odds between 17% and 56.5%. The yield gap between iShares' California and national muni funds didn't move with them by any amount I could detect.
maxghenis.com/blog/califor...
GPT-6 Sol leads PolicyBench at 95.2% but gets 67% of nonzero tax and benefit amounts within $1 on its own. In a pilot, we gave it a tool that runs PolicyEngine, the engine behind our reference answers, and it got 97% of those amounts: policybench.org/notes/2026-1...
A video game doesn't guess where a thrown ball lands; its physics engine computes the path. On Tax Chats, I told Jeff Hoopes and Scott Dyreng I expect AI to compute taxes and benefits the same way, by calling an engine that encodes the law: www.buzzsprout.com/1878989/epis...
I document our open calculator and each assumption in a working paper: www.policyengine.org/us/californi...
Jaros & Rauh center on California losing 45% of billionaires' income tax forever. We estimate $51B more NPV if that loss lasts 10 years and $68B less if it grows 1.5% a year after inflation. If voters pass Prop 40, I expect studies of how many billionaires leave because of it.
The Public Benefit Innovation Fund is funding @axiom.org in its new cohort. With a state agency, we'll test whether open encodings of SNAP rules let staff trace a determination back to the law. Thanks to the Center for Civic Futures and Recoding America. www.centerforcivicfutures.org/resources/ce...
I've traded 22 of @PolicyEngineUK's play-money Budget markets this week. On the reporting, I now put a bank tax rise at 75%, a main CGT rate rise at 20% and a lower mansion tax threshold at 15%.
We've also opened seven new ones. Trade against me: manifold.markets/topic/uk-aut...
At 4:30pm GMT on Budget day, Wednesday 28 October, we'll present our first estimates of what the Autumn Budget means for public finances, household incomes and poverty, and take your questions.
Register free: us06web.zoom.us/meeting/regi...
What will the Chancellor put in the Budget on 28 October? @uk.policyengine.org opened 30 play-money markets on Manifold, CGT rate to mansion tax: manifold.markets/topic/uk-aut...
We'll model whatever's in it; the prices tell us what to test in advance. Trade, or tell us what we missed.
We project that 2025 Supplemental Poverty Measure will rise for children and fall for seniors compared to 2024, partly because thresholds rose 6.3% for renters and 4.4% for owners without a mortgage.
Numbers come out in one minute, so we'll report back on how our predictions fared shortly!
This is an example X post. Export your X archive and run the import script to populate this file with your real posts.
This is an example LinkedIn post. Export your LinkedIn data and run the import script to populate this file with your real posts.
How to update
This feed combines posts from multiple platforms. Here's how to keep it updated:
- Bluesky: Updates automatically at build time via public API
- X/Twitter: Request your data archive from X settings,
then run
npm run import-xto convert it - LinkedIn: Export your data from LinkedIn settings,
then run
npm run import-linkedinto convert it