Write a tender that can be scored.
Separating what the buyer decides from what the network decides is the whole legitimacy argument, so it gets stated on the screen where the buyer sets both. You choose the criteria, the weights and the tie break. The network chooses the scores.
Weights are yours and are published with the criteria. They never enter the prompt, so the network scores each criterion on its own and your priorities are applied in code afterwards. There is no method in this contract that edits either after publication.
The tie break is part of the published standard, chosen now rather than after the scores arrive. Two bids on the same weighted total are separated by the criterion you mark here, and only then by the order the commitments arrived.
Criteria that cannot be scored from proposal text produce scores nobody can defend. The network judges each one and the verdict is stored on chain - publishing refuses any criteria set that has not passed.
After the reveal window, the buyer has until the decision deadline to award or decline. After that it is permissionless - a buyer who dislikes the result cannot strand an escrowed budget by doing nothing.
Round fee 0% of the awarded budget, charged at award and never on a declined round. Entry deposit 0 GEN per bidder, refunded once their bid is scored. Appeal bond 0 GEN.
All three are copied onto the round at publication, so a later change by the contract owner cannot reach a tender whose bidders have already read the terms.
| Criterion | Why it works |
|---|---|
| relevant delivered work with references | a proposal either names them or does not |
| plan is specific and sequenced | dates and order are visible in the text |
| maintenance after handover | a named period and price, or nothing |
| price against scope | two numbers, compared |
| cultural fit | no standard exists in the text |
| enthusiasm | rewards writing style |
| long term partner | unmeasurable from a proposal |
These are refused because answering them needs a standard that is not written down, outside knowledge about the bidder, or a prediction about the future - not because they are bad things for a buyer to want. Put them in a conversation, not in a scored tender.
Scoring prose is not the same as evaluating capability. A well written proposal from a weak supplier will beat a badly written one from a strong supplier. Criteria that ask for verifiable references blunt that as far as anything can - which is why the first row of the table above is the first row.