arg-christiano-basin-fragility
leaf/step · drafted
Statement
Corrigibility derived instrumentally from shared values breaks under slight value divergence, while Christiano's broad basin of attraction claims the opposite for a benign act-based agent, so the two pictures disagree about an agent whose corrigibility comes from valuing its overseers' corrections as information.
Source
Christiano 2017, Corrigibility; projects/deference/corrigibility-note-dump-2026-10-01/corrigibility-discussion-outline.md, item 7 (asks where legitimacy-accuracy needs a basin); the author, projects/deference/corrigibility-note-dump-2026-10-01/position-statement-v2.md, section 1.4: "The big picture here is that I'm trying to flesh out a picture of Paul's basin-of-attraction argument, though the basin-of-attraction analogy may not be a perfect fit."
Notes
Outline item 7 (audit row 2.15). The toy basin of the dynamics frame (basin-formal-object) is the formal anchor for the work this leaf asks for; the bridge between the legitimacy line and Christiano's basin is attribution-unvetted on both sides.
Author and human events
Author @smithy-verity; version 1; size M.
No human event has touched this node.
Edges
Above
Ranking (the lens, not the structure)
{
"status": "open",
"diagnosis": null,
"flags": [
"AI number"
],
"V": {
"V_prior": null,
"display": "2",
"headline": 2,
"ref": 2,
"valid": 2,
"weaker": "both"
},
"belief": {
"J": 0,
"mid": "19/40",
"range": [
"0/1",
"39/40"
],
"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.