computer science
simulation, evaluation, and complexity treated as engineering inputs.
platonik began as a question from minimal-systems research: what a small colony of executable policies can actually be shown to do. answering honestly requires real machinery: a deterministic simulator, feasibility proofs for its challenges, and an evaluation method that refuses to overclaim.
this category covers that machinery and the adjacent computer science: simulation semantics, controlled measurement, and what a bounded experiment can and cannot establish.
lessons
- what minimal systems can actually show — small programs, honest claims
- deterministic simulation without foundationdb's budget — seeded ticks, atomic steps, replayable worlds
- feasibility checking and witness runs — a challenge you can't solve isn't a test
- your benchmark is an experiment: measuring behavior without overclaimingsubscriber — controls, baselines, and what a result does not show
- p-vs-np-adjacent experiments: what they can and can't tell you — average-case evidence, honestly labeled
projects
- platonik — the organism game whose engine doubles as a research instrument.
- peqnp — a laboratory for exploring p vs np through small programs and explicit proofs.
- aicharts — benchmarks as published, sourced, reproducible measurements.
- morphogen — replayable workflow organisms evaluated against recorded receipts.
AI-drafted at Ben Guo's direct request as category seeds; each lesson page is reviewed against the source repositories before publication.