Shielded jury assessments for Deep Funding

The pitch, in writing

Deep Funding lets a human jury settle prediction markets. Every round, the jurors' scores decide about a hundred markets on Seer. This proposal keeps those scores sealed until the markets resolve: shielded jury assessments, built on the Shutter API. Objective 2. 2,400 dollars. Six weeks.

1. What change, and for whom?

The jury scores that settle Deep Funding's prediction markets should stay sealed until those markets resolve, so that nobody, the operator and the jurors included, can trade on them early. That serves Deep Funding and Seer, whose oracle this is. It serves the traders, and the treasury that subsidises them. And it gives Shutter a second juror-privacy deployment, with a fee attached. Here is why it matters. Today, juror scores sit in the clear for weeks, while subsidised AI agents trade about a hundred markets that those scores will settle. The pools are thin: about a thousand sUSDS, across 196 pools. So one informed trade moves a price by 10 to 30 points. The Seer treasury paid more than 14,000 sUSDS in trading subsidies for this round alone. And in September, a wallet run by a registered juror traded the markets of that juror's own repositories, in the direction of the juror's scores. The design allows it, and nobody can see it.

2. What gets delivered?

Five outputs, from 9 November to 21 December. The dates follow the revised programme of 16 October; nothing else has changed. One: an evidence pack that prices, repository by repository, what an early score is worth, reproducible from Optimism and the Seer API, published by 27 November. Two: a threat model for the jury app: who sees what, and when, and how the reveal meets Reality.eth and Kleros. Three: a prototype on the Shutter API test environment, that seals an assessment until the market's resolution time. Four: a decision brief to Devansh Mehta and Clément Lesaege, with a Round 4 pilot plan, and a go/no-go call. And five: a fee memo for this DAO.

3. Why is this a credible path to a Shutter outcome?

The mechanism already exists. It is the Shutter API time-lock that protects Kleros juror votes today, so there is no new cryptography. The counterpart trusts it. Seer's co-founder also co-founded Kleros, and Kleros arbitrates every one of these markets. And the decision has an owner and a date: Round 4, early in 2027, designed by people who act on evidence. When a flaw in the Round 3 markets was raised on 30 September, the organisers rebuilt the market set the next day.

4. What can the DAO check, beyond activity and publication?

Not the reports. Four things you can check yourselves. The decision: a go/no-go call in the week of 14 December, with the operator's answer and its reason posted in the application thread. The integration point: the jury app's maintainers confirming in writing where sealing hooks in, and which release-time policy they accept. The prototype, run by you: seal a test assessment, and see that it will not open before its time. And a price: whether the operator names a fee range, or a budget line. The strongest signal is a scheduled Round 4 pilot. The weakest acceptable one is a named blocker in writing, which tells Shutter exactly what a second attempt needs. Disclosures. The applicant has received nothing from Shutter, the DAO, brainbot or Axia, and holds no role with Seer, Pond, Gitcoin, Kleros or Deep Funding. Seer trading subsidies were received as a participant, and the Round 2 loan is repaid. Nothing non-public from this work will be traded on, and the juror evidence stays anonymised. The ask is 2,400 dollars, half on approval and half after the milestone review, or 1,600 without the prototype. Shielded jury assessments for Deep Funding. Sealed until the market resolves.