Talking to Machines

Notes

Notes is where Talking to Machines writes up results as they come in, before the papers: what the digital twins predicted and how they scored, what we learned building them, and what turned up at our meetings. The notes are published on Substack and listed here as they appear, newest first.

  • Your Town, Your Bill
    We showed fourteen data-center ads from the midterms to digital twins of Ohio voters. Then we wrote our own and took it apart one element at a time.
  • What turned up
    Three days in Hong Kong, 117 submissions, and a field that has started borrowing the right tools — the follow-up I promised.
  • Where we actually are with synthetic respondents
    Written on the eve of LLMSS 2026, Hong Kong
  • Nine readers
    Four Supreme Court predictions, filed before the arguments are heard.
  • The Silent Region: What Financial Influencers Don't Say
    We built digital twins of 81 financial influencers and interviewed them every day for three months. The most interesting thing we measured was the silence. Our paper is now on arXiv.

New notes arrive by email if you subscribe; it is free.