Journal
Posts and long-form series on AI, startups, venture capital, and more.
All Posts
Lennard-Jones Clusters: Finding Molecular Ground States
Using genetic algorithms to find minimum-energy configurations of atomic clusters, a classic benchmark problem in computational physics.
Incremental and Global Fit Are Two Different Products
Elo nudges after every result. Massey re-solves the whole season. The choice decides your cold start, your reproducibility, and what a rating even means.
Island Models: What Splitting the Population Actually Buys
Islands, migration and topology, measured at a matched budget. The results point the opposite way to the usual advice.
Whole-History Rating: Ratings With a Past Tense
Every other rating system tells you how good a competitor is now. WHR tells you how good they were on any day, using everything learned since.
Genetic Algorithms: Evolution as Optimization
An introduction to genetic algorithms: how evolutionary principles solve optimization problems, from selection and crossover to mutation and fitness landscapes.
Margin of Victory: Massey and Pythagorean
Two rating systems that read the scoreboard instead of just the result, and three things that turned up when I interrogated how they use it.
Building Python Libraries in Rust with PyO3 and Maturin
A practical guide to creating Python packages backed by Rust using PyO3 for bindings and Maturin for packaging. Performance without sacrificing usability.
What Uncertainty Buys a Rating System
Glicko, Glicko-2 and TrueSkill all track how sure they are of a rating. What that second number does to a prediction, and whose uncertainty moves it.
The Shape of a Pull Request You Can Trust
Six sections that let a reviewer believe a diff without reproducing the work, including the two most people leave out, a mutation check and a "not done" list.
Metering Is a Ledger, Not a Counter
Usage-based billing is a transactional problem. Reservations, compensations, and why a job that produces nothing has to refund what it held.