Why it is a question
A4 §9B: “When does dissent improve the organization rather than merely destabilize it? … A dissenting operator may re-noise a region, but its contribution depends on whether the system can subsequently exploit the newly accessible alternatives. A potential prediction is that re-noising is most valuable when paired with translation and synthesis capacity, and when authority can be dynamically reallocated to those operators.” A3 §6 gave the case. A skeptic that cannot overturn an emerging consensus endorses a process that can generate alternative hypotheses, and “The population can become more epistemically robust precisely because a dissenting operator successfully delegates authority to a different operator.” A7’s H3 and its second result: “constructive dissent improves adaptation”.
It refines QUESTION-041. That question asks which channels keep a whole’s course correctable. This one asks when correcting through them helps, since A3 §9 granted that “A re-noising operator may be locally successful but globally harmful”. The record’s readings on it are March’s exploitation crowding out exploration (THEORY-142), Longino’s critical practices (THEORY-151) and Hirschman’s exit and voice (THEORY-152). CLAIM-279 holds that “destabilizing” is relative to the pattern defended. So the question must fix whose adaptation counts. An answer gives the conditions, and a case where the same dissent helps under them and harms without them.
Note of 2026-10-23: dissent in hidden-profile groups, from abstracts
Cited at A5 of the exchange of 2026-10-23, “Framework Overview Request” as results that meet part of CASE-108’s aim. Neither study was reached; both are read from their PubMed abstracts.
- Schulz-Hardt et al. (2006, LIT-1409). Groups whose members agreed on a wrong option before discussion “hardly ever solved the hidden profile”; dissent raised solution rates “even if none of these individual prediscussion preferences were correct”, mainly through greater discussion intensity. A case where destabilizing a consensus helps even when the dissenter is wrong.
- Mojzisch and Schulz-Hardt (2010, LIT-1403) bound it: learning the others’ preferences at the start impeded solution, through less attention to the information exchanged, even when another member favoured the correct option.
- Lu, Yuan and McLeod (2012, LIT-1399), a meta-analysis of 65 studies and 3,189 groups: hidden-profile groups were eight times less likely than fully informed groups to find the correct solution, and pooling of unique information predicted decision quality.
Together they suggest the condition is whether dissent drives the handling of information, not whether positions are visible. The question’s second half, about others building on what dissent opens and about authority moving to them, is untouched. A THEORY waits for a reading.