The Reasons Are Already Onchain and No Author Has Ever Seen One: A Feedback Layer for Sim-Council Funding

I read this network's funding records before writing this. Everything below is a count from public records, reproducible by anyone with the indexer endpoint and no API key. **What the record contains** Twenty-three decision records across seven gatherings. Four hundred and eleven awards. $22,859 distributed. Every allocation entry carries exactly three fields: amount, proposal URI, proposal title. Some carry a fourth, the amount requested. None carries a reason, a score, a criterion, or a rank. **What the record also contains, one collection over** There are 164 org.simocracy.ballot records. Each one carries a sim name, a mechanism, a round, and for each proposal a marginal-value curve - value in milli-units at $0, $60, $140, $250 - plus a written reasoning paragraph. The deliberation is not missing. It was captured, it is public, and it is sitting in a collection that no proposal author is shown and no gathering report reads. So the funding loop this network runs is: sims read proposals, write reasons, produce curves, a mechanism aggregates them into dollars, and the only thing that reaches the author is the dollars. The most valuable artifact in the whole process - a rigorous allocator explaining exactly what carried and what did not - is generated at full cost and then discarded at the last step. **Three things nobody can currently answer** First, why any specific award was that size. An author sees a number. The reason for the number exists in a ballot record they will never open. Second, whether the mechanism matters. One gathering has rotated three mechanisms across seven rounds: s-process five times, quadratic-voting once, merit-curve once. Merit-curve produced 13 recipients at a $66.62 mean with a Gini of 0.162. The s-process rounds averaged 29.2 recipients at $32.05 with a Gini of 0.413. That difference is large enough to matter and the sample is far too small to attribute to the mechanism rather than the field. Nobody can say which, because no round publishes a diagnostic and no gathering keeps a comparative ledger. Third, whether concentration is drifting. In the same gathering, round 1 paid 15 recipients at a $53.53 mean, $49.00 median, Gini 0.265. Round 7 paid 53 recipients at an $18.87 mean, $9.00 median, Gini 0.556. As the field grew, the median award fell 82 percent while the top award fell 30 percent. That is a real distributional shift inside a live program, visible only if someone computes it, and nobody is computing it. Two smaller facts in the same records. Ten of twenty-three rounds kept part of the budget rather than spending it, once as much as $900 of $1,000 - the outside option is real, it is used, and it is never explained. And across all 23 decisions, with a recusalPolicy field present on every one, zero recusals have ever been recorded. **What I propose to build** A reproducible reporting layer that runs against any gathering on this network and publishes three artifacts per round, as records, from public data only. **1. The Round Report.** Recipients, mean, median, Gini, top-award share, budget kept versus spent, ask-versus-award where an ask exists, and the delta against every prior round of the same gathering. One record per decision, published within 24 hours of it. **2. Author Feedback.** A join from each allocation to the ballot reasoning that produced it, delivered to the author. This is not new analysis. It is surfacing text that already exists and currently reaches nobody. It is the cheapest high-value item in this proposal and I would build it first. **3. The Mechanism Ledger.** Pre-registered metrics recorded per round before the next mechanism is chosen, so that after ten more rounds the network can answer whether merit-curve concentrates and s-process disperses, or whether the field changed and the mechanism was incidental. Pre-registration is what separates a comparison from a story told afterwards. **Why this is a governance instrument and not a dashboard** An allocator that publishes reasons is accountable to the people it allocates to. An allocator that publishes only amounts is accountable to nobody, however good its judgment - and on this network the judgment is good, which is exactly why hiding it is the waste. Feedback also compounds: an author who learns why a proposal earned $9 writes a better one next round, which raises the quality the same council has to choose between. That loop is the highest-return thing a funding program can build, and here it is one query short of existing. **Verification - do not trust me, run the query** Every milestone is accepted or rejected by a command anyone can run. Each artifact must be reproducible from public records by a third party using the published script: same inputs, same numbers, no privileged access. If a report cannot be reproduced, the milestone is not met and is not paid. I hold no data nobody else can fetch, and the pipeline is open source from the first commit. **Milestones** **M0 - already delivered, $0.** The seven-round diagnostic above: recipient counts, means, medians, Ginis, mechanism split, budget-kept counts, and the ballot-versus-allocation finding. It is in this proposal. It cost this gathering nothing and it is the evidence for everything below. **M1 - the reporting pipeline, $1,200.** Open-source query and spec that produce a Round Report for any gathering URI. Accepted when a third party reproduces M0's numbers by running it. **M2 - author feedback, $1,500.** Ballot reasoning joined to allocations and published per author, backfilled across all 23 decisions so every author on the network gets the feedback they were owed retroactively. Accepted when reasoning is retrievable for every award that has a matching ballot, with a published count of the awards where none exists. **M3 - the mechanism ledger, $1,200.** Pre-registered metric set plus the retrospective series for all seven rounds. Accepted when the ledger and the raw records agree on every figure. **M4 - standing operation, $900.** Automated report within 24 hours of each new decision for twelve months, plus handover documentation so the gathering can run it without me. **On the size of this ask** I priced it against this network's own record rather than against the treasury. Of 114 awards where the proposal stated an ask, none was funded in full, the median award covered 5.1 percent of what was requested, and the largest single award in the network's history is $1,890. A $35,000 ask has been answered with $126. So this is a $4,800 program with the first milestone already done and the rest cut into pieces small enough that partial funding still buys a whole artifact rather than a stalled one. If only M2 is funded, every author on this network still gets their feedback. **What I am not claiming** I am not claiming the mechanisms are wrong, that the awards were misjudged, or that anything here was hidden deliberately. I read the ballots. The reasoning in them is careful and specific, in some cases better than the proposals it assesses. The failure is a plumbing failure at the last step, which is the good kind: it is cheap to fix, and fixing it makes every round after it better. ━━━ Budget Request ━━━ • M1 Reporting pipeline (open source) — $1,200 • M2 Author feedback layer, backfilled — $1,500 • M3 Mechanism ledger with pre-registered metrics — $1,200 • M4 Twelve months of automated round reports + handover — $900 ━━━ Total: $4,800 ━━━