Writing

Notes on evaluating agents, simulation, and eval infrastructure.

01Essays
Long-form arguments with interactive simulation models.
ESSAY 01·11 MIN READJun 25, 2026
Eval-Driven Development

Agents move decisions out of control flow and into the model. That part of the specification can only be recovered by watching the system act, which makes the eval the place where it gets written.

§ Evals discover the contract · § The world can make the spec false · § Outcomes cross ownership boundaries◆ Interactive Model
ESSAY 02·8 MIN READDec 6, 2022
Priors for a portfolio

A model's job is to be honest about what it doesn't know, in a form a decision can use. Three years of simulating clinical trials taught me what that means.

§ Two patients at a site · § The decision is the output · § Why simulate◆ Interactive Model
02Field Notes
Short tactical observations from building evals.
  1. Which branch would you bet on
    A leaderboard reports scores. A decision needs a preference and an uncertainty, and frontier scenarios are exactly where you have neither by default.
    Aug 18, 2026
  2. Known-bad candidates
    An eval that has never been shown a wrong answer has not been shown to measure anything.
    May 19, 2026
  3. Five of five is not a statistic◆ Interactive
    What repeated eval runs can tell you, what they cannot, and why the two get confused.
    Apr 14, 2026
03Elsewhere
Older writing on Medium & Towards Data Science.