This interests me for three reasons: 1) it is the kind of ambitious policy vision we need to see; 2) it models its assumptions and clearly explains them; 3) there are no notable US politicians speaking at the World Inequality Conference. We need leaders who are relevant voices in exactly these conversations.

An AI harness for research. I have not tried it out because it seems far too optimized on writing papers which is not my primary output anymore.

I’m not surprised by the finding that even among AI users, code harnesses are not used. But if you are not finding value from AI for writing quantitative analysis code and research communication outputs its probably because you aren’t using a coding harness like OpenCode or Claude Code.

When I was in grad-school it was trendy to call these FUQs - fundamentally unidentifiable (or unanswerable) questions. This is a much more nuanced, engaging, and useful take on how to approach a problem where your variables of interest can’t be varied.

Per Engzell and Nathan Wilmers provided some real insight on experiments with how well LLMs can produce quantitative social science paper production, even held back by using Stata of all things.