Fair Share

A team wins one prize and has to split it. The usual way, everybody writes down a split and you average the answers, pays whoever lies about themselves: in a team of four with a true split of 40 / 30 / 20 / 10, D writes "100% me" and takes 32.5% instead of 10%. In Fair Share members rate only the others, and what you write never changes your own share: D keeps 10.0% whatever D writes.

The rule is the impartial division of de Clippel, Moulin and Tideman (2008). The code is on GitHub (MIT). Everything below runs in your browser from the same files as the command line. NEAR testnet only: no wallet, no keys, no money.

1. Try it

Pot, team and ratings are yours to change. Everybody starts out honest (each rates the others by the true split) except one member, who lies. Switch the lie off or on, or drag the sliders.

Whose ratings are you looking at?

Who gets what

The "plain average" column: everybody states a split for the whole team, a lying member claims 100% for herself, the answers are averaged. A split needs at least 4 members: with 3 no impartial rule can hand out the whole pot (Proposition 1 of the paper).

2. Check the real testnet round

This page reads the Fair Share contract on NEAR testnet through a public RPC node (view calls only, nothing is signed), recomputes every number with the same fairShare, toAmounts and checkRound as npm run verify, and says MATCH or MISMATCH for each member and for the whole round.

3. What this does not do