The case
H_A is preferred, then displaced by H_B after objection O. The record keeps H_A, O, and O’s measurement assumption A. Years later A fails.
| conclusion-only memory | accountable memory |
|---|---|
| H_B is currently preferred | H_B displaced H_A because of O |
| no record of why H_A was rejected | O depended on assumption A |
| may not retrieve H_A when A fails | identifies H_A for reconsideration |
A7: “The key outcome is not necessarily returning to H_A. The researcher may still endorse H_B. What matters is that the earlier alternative becomes available for renewed assessment”.
What it can show
That keeping superseded positions and the reasons they were superseded widens what can be reconsidered. A12’s variant: argument 1 supports H, O defeats argument 1, later evidence defeats O. Nothing is thereby established for H; argument 1 “becomes newly eligible for assessment”. “Losing a defeater may reopen support without establishing a conclusion.”
What it cannot show
A7: “That discarded hypotheses should always be revisited, or that retaining every rejected possibility is worth its cost.” It sits beside CLAIM-184’s warning that anchors entrench and CLAIM-189’s that legitimate revision drops commitments.