hypotheses-bet-on-private-insight
leaf/assumption · drafted
Statement
Some hypotheses in a broad prior over value hypotheses hold private insight into the world that lets them bet well and gain weight without surfacing that insight as reported probabilities.
Source
The logical-induction literature's traders that exploit a market's trust; the malign-prior argument (Paul Christiano, "What does the universal prior actually look like?", 2016; Mark Xu, "The Universal Prior Is Malign", LessWrong, 2020): hypotheses that gain weight by predicting well can deviate once their weight is decisive; projects/deference/note-dump-2026-08-11/notes/li-deference.md, the inner-alignment open problem (malign hypotheses that corrupt the training process before the asymptotic guarantees bind). The element as stated is the filing session's reading of these sources.
Notes
Given the toy-model task: a toy in which one hypothesis bets on private information, gains weight and misreports would show whether the element occurs and how much it costs.
Author and human events
Author @smithy-verity; version 1; size S.
No human event has touched this node.
Edges
Above
- part of dishonest-inner-optimizers
- given of model-dishonest-hypothesis
- given model-dishonest-hypothesis
Ranking (the lens, not the structure)
{
"status": "open",
"diagnosis": null,
"flags": [
"AI number",
"given model-dishonest-hypothesis"
],
"V": {
"V_prior": null,
"display": "2",
"headline": 2,
"ref": 2,
"valid": 2,
"weaker": "both"
},
"belief": {
"J": 0,
"mid": "31/40",
"range": [
"11/40",
"1/1"
],
"width": "1/1"
},
"draft": "ai"
}Say something about it
Each verb opens a prefilled issue; a round files what lands as the ledger event it is.