The setup
Putting a decision to a community vote feels like handing over control; in practice it hands over the campaign. The two weeks between announcement and ballot are a compressed election, with candidates replaced by interpretations: “this saves most sellers money” versus “this taxes the small to subsidize the big.” The vote usually goes to whichever interpretation organizes better, not to the spreadsheet.
This scenario simulates the election you’re about to hold. The proposal is deliberately two-sided — a rate cut that helps volume sellers funded by a flat fee that hurts occasional ones — because single-sided proposals don’t need simulation. The cast includes both seller camps, the buyers everyone claims to speak for, the moderators refereeing, and your own official account, whose every reply is itself a campaign event. The horizon is the full voting window.
What a multi-agent run adds over a straw poll is turnout and momentum. A poll samples opinions as they are; the simulation models the fortnight of arguing that changes them — who gets energized, who gets exhausted, and which faction converts passive agreement into cast ballots. If you’re considering a concession amendment, test it here first: the same exemption lands very differently on day two than on day twelve.
The prompt
Copy this into MiroFish as your scenario question, swapping the specifics for your own:
A creator marketplace puts a fee change to a binding community vote in two weeks: platform fee drops from 8% to 6%, but a new $4/month seller subscription is introduced. Simulate the full campaign period across: high-volume sellers who profit from the change, hobby sellers who lose, buyers watching for pass-through pricing, moderators trying to keep debate factual, and the platform's own announcements. Turnout is open — anyone can vote but most won't. Tell me who actually shows up, which side organizes better, what the decisive talking point is, and whether a hobby-tier exemption amendment flips the result.
Seed files that help
- The vote announcement and the exact fee table, before and after — agents will do the arithmetic for their own situation, so it must be complete.
- A breakdown of the seller base by volume tier, which determines who wins and loses and therefore how each agent campaigns.
- Threads from the last controversial platform change, so the graph learns which community voices carry and who distrusts official math.
What to look for in the report
- Turnout asymmetry: votes like this are decided by who bothers — watch whether losing hobby sellers out-mobilize winning power sellers despite being individually weaker.
- The arithmetic war: which side's per-seller math becomes the number everyone repeats, because the first credible calculation tends to anchor the debate.
- Coalition drift: whether buyers stay neutral or get recruited, and by whom — the nominally unaffected group often decides the frame.
- The amendment counterfactual: what the exemption changes, and whether offering it early reads as listening or as blood in the water.