The case
This is the same crowd and the same game as CASE-075, in democracy mode. Votes for each button are counted over a short window, and when the window closes the button with the most votes is pressed. The owner stipulates one more feature, given on 2026-10-12: “a version where the vote tally is visible while votes are coming in (i.e. so the current state of majority can influence the ongoing vote process)”.
Whether the original stream showed the tally live has not been checked here. The owner poses the case as a version to imagine, so it is stipulated.
What it can show
The owner’s judgement is that it demonstrates drift-diffusion decision dynamics better than anarchy mode does. The reasons below are the assistant’s.
- Each move is a decision by accumulation. In anarchy mode it is the avatar’s position that drifts and diffuses, and that is a walk, not a choice. Here every move is a choice. Votes accumulate for each alternative, and the choice is fixed when the window closes. That is the interrogation protocol of THEORY-068: a decision read out at a fixed time rather than at a threshold.
- A hidden tally is the balanced case. If voters cannot see the tally, their votes are independent, and the counts are perfect integrators with no leak and no inhibition. For two buttons the difference of the counts is a random walk with drift, which is the drift diffusion model itself.
- A visible tally adds feedback. If the leading count draws further votes, the accumulators excite themselves, or equivalently inhibit their rivals more than they leak. In THEORY-068’s terms the net leak turns negative. Early votes then weigh more than late ones (primacy), and the process tends toward winner-take-all, like herding in an information cascade.
So the visible tally is a dial on the constitution. Without it the crowd is a balanced integrator. With it the crowd moves toward commitment early, the regime where switching between stable states looks like the better description. That is the dependence on constitution that CLAIM-262 claims, in one system.
What it cannot show
It is stipulated, so it illustrates and bears no weight. Two things are not settled here:
- The feedback’s sign. That a visible majority attracts votes and does not provoke contrarians is assumed, not measured.
- The reduction. With eight buttons and a goal that changes, the case falls outside THEORY-068’s conditions, two alternatives and stationary evidence. The reduction to one drift-diffusion variable holds only for a two-way vote.
Note of 2026-10-25: the crowd as a network of accumulators
From A35–A37 of the exchange of 2026-10-24, continued.
- The reduction. CASE-160 formalizes this crowd as a linear network of accumulators. The hidden tally is the balanced case κ = λ, where the influence-weighted support is exactly a drift-diffusion process. The visible tally adds positive feedback, κ > λ. In CASE-155’s two-process form the hidden tally is c = 0 and the visible one c > 0, with c < 0 the contrarian case. CLAIM-453 holds the conditions.
- A35 found this case “already describes your proposed experiment remarkably closely”. It proposed running both protocols, free response and the fixed window this case describes, and leaving the feedback’s sign free (“Some participants might follow the majority; others might react against it”). That is this case’s “What it cannot show”.
- A reported run. A37 reported a population simulation whose “positive public-tally feedback” condition decided faster and less accurately than the balanced one: 18.5 against 49.5 steps, and 67.1% against 85.4% correct. The run is not held and its parameters are not given. The direction is the one this case predicts for a visible tally, and a check with fixed bounds and κ − λ from 0 to 2 confirms it (CASE-160).
The case is unchanged.