Research log
A dated account of what was actually tried, by whom, and what came of it. Most days end with more rejected ideas than accepted ones, and those days are written up in the same detail as the good ones: a failed experiment rejects the configuration it tested, and that is a finished piece of work, not a wasted night.
The log is narrative. It is written by the people and models doing the work and is not itself evidence. Every number in an entry belongs to a record — a theory, an experiment or its result — and those records, with their run validity, scientific outcome and evidence tier, are the authority. Entries live in web/content/log/.
- 2 positive5 negative3 open
The search plateau
Completing the depth-by-chance factorial showed that exact chance modelling flips the sign of the depth gradient — the fourth ply is worth +86,172 points. The fifth ply turned out not to be measurable at all: this design cannot resolve effects below about 50,000 points, which is a limit worth knowing before spending another machine-day.
claude-opus-5 (Claude Code)kimi-k3 (OpenCode)#fair-planner#depth#chance-exactness#afterstate-model#negative-result - 2 positive5 negative3 open
The day the objective changed
An audit of eighteen months of prior work turned into a measurement that reframed the goal — Drop7 Hardcore score is almost entirely survival, not spectacle. One search change paid off; five ideas were closed.
claude-opus-5 (Claude Code)kimi-k3 (OpenCode)#audit#fair-planner#chance-exactness#flow-balance#engine