Block a user
UnknownKey reports indices, not the key — and rejecting the whole matchup is easy to swallow into a plausible constant
Design question: can a fitted
History be persisted, or must every consumer replay its whole event log?
Design question: can a fitted
History be persisted, or must every consumer replay its whole event log?
Answering with measurements rather than opinion. Short version: your premise is right that a snapshot is delicate, but wrong about where the delicacy is — and Q1's answer is better than you…
UnknownKey reports indices, not the key — and rejecting the whole matchup is easy to swallow into a plausible constant
Asks 1 and 2 are done in c12bc83. Ask 3 is deliberately still open — see below.
Thank you for the write-up. "No exception, no log line, no failing test — only a parameter sweep that produced…
No public way to ask "what is P(skill below x)?" —
cdf is pub(crate), so callers approximate it with a z-multiplier
ITERATIONS = 30 is far too low for a dense history, and overrunning it fails quietly
expected_information_gain is discrete-outcome only, so scored consumers cannot ask "which matchup next?"
Every
predict_* is a ranked outcome — there is no predictive distribution for a SCORE
A node's reported sigma is ~10x wider than the estimate's measured reproducibility
No joint posterior: anything derived from two or more competitors has to add marginals and overstate its width
Design question: can a fitted
History be persisted, or must every consumer replay its whole event log?
UnknownKey reports indices, not the key — and rejecting the whole matchup is easy to swallow into a plausible constant
No public way to ask "what is P(skill below x)?" —
cdf is pub(crate), so callers approximate it with a z-multiplier