A version shows its personas its own description and price

A scenario names the version under test — its variant (name, description, emphasised claims) and its price — and a sweep runs several to compare them. But the concept a persona was shown was built from the brief's product description alone. Neither the scenario's variant description nor its price reached any persona, so every version of a sweep showed one identical concept, and a model at temperature zero answered them identically. The first price sweep (2026-10-03: $9, $19 and $29 over one population, two seeds) returned the same adoption at every price to four decimals; only the world ids differed.

The concept a world publishes is now the scenario's variant description followed by its price ("…\nPrice: 19.00 USD"). A study run from one brief has a single version whose description is the brief's, so it gains only the price line.

Consequences

Prompts change, so turns recorded before this are not comparable with turns after it, and the replicate-safe cache does not reuse them. Every earlier study showed no price: results that discussed price sensitivity read a reaction to something else (the code-review benchmark is corrected). World and config hashes do not move — the scenario already carried its price and description; only what a persona is shown changes.

Which claims a world posts is unchanged: every brief claim is still posted, whatever a variant emphasises. Making emphasized_claims select the posted claims is a separate decision, not taken here.