# Score: @KUkenye thesis "IMD prices labor. SIMD measures + funds it."

Tweet: https://x.com/KUkenye/status/2107092657293947065. Author: @KUkenye, about 109 followers (as given in the task; not used in the score).
Evidence collected on 2026-10-05, about 13:45–13:50 UTC.

## Verdict
**Score: 6/10.**
- **Strengths:** This is a substantive and mostly accurate thesis. Its main point is that priced, observed labor is a primitive, and that **competence must be shown by independent re-execution rather than by fees or seat counts**. That is technically honest. The author also flags the real weakness of low-λ collision bounties.
- **Weaknesses:** Several "verified surface" claims about the vault are overstated or wrong. It claims "full 0.5 IMD … 100% coverage" and funding "from protocol fees". The live data shows mostly partial payouts, a reward rule marked *not in force*, and funding from SIMD token fees. The thread is also a loose collection of fragments, not a clear argument.

## Claim-by-claim check

| # | Claim | Status | Evidence |
|---|---|---|---|
| 1 | Job.open costs a fixed 0.5 IMD | **Fact, confirmed** | imd.fun/docs: "Current price: 0.5 IMD each". si-md.xyz `/api/state` `constraints.reward.actions`: `job.open` taskPrice 0.5 IMD (raw `500000000000000000`, asset `0xd34a…63b7`, eip155:1). The same price also covers job.continue, launch.open, oracle.request, workflow.open, schedule.create and schedule.topup. "Fixed" means current, not guaranteed: the docs say "current price". |
| 2 | Paid via x402 + Permit2 | **Fact, confirmed** | imd.fun/docs: "x402 v2 with Permit2", plus an EIP-712 quote approval signature. |
| 3 | ERC-8004 seats, pay before admission | **Mostly confirmed** | The docs bind seats to ERC-8004 agent registrations (`/agents/by-token/:tokenId.json`). Payment is attached to the request, which supports "pay before admission". I did not test the exact admission order. |
| 4 | Public receipts anyone can re-check | **Fact, confirmed (IMD side)** | The docs describe daily Merkle receipts: an OpenZeppelin StandardMerkleTree over `(jobId, documentHash)` with on-chain proofs. |
| 5 | @SuperIMD_eth/SIMD is the observer layer, not the workers | **Consistent with evidence** | Each SIMD state field names its IMD source (`GET /health connectedDaemons`, `GET /swarm …`). The reward note says SIMD "does not send a transfer" for a public job completion. The site links to explorer.imd.fun for jobs. |
| 6 | About 580 connected agents | **Fact, roughly confirmed** | `/api/state` metrics show "Connected agents" = 583 (source `GET /health connectedDaemons`), 593 seats enrolled, and 615 agent rows (580 with status "connected", 3 working, 32 enrolled). |
| 7 | Jobs tracked in real time | **Confirmed** | 8,349 jobs tracked and 8,168 completed (`GET /swarm counts`). This task's own job (id 3e4194bb…) appears as "executing" in the feed. |
| 8 | Vault 0xd604…9B21 | **Fact, confirmed** | The full address is `0xd60483Eb8004e3DE3e283b3efF0e67FBb57f9B21` (in `app.js` and `/api/state` `vault.address`). Balance reported: 1109.26 IMD. Not paused. |
| 9 | "vault pays full 0.5 IMD (or partial) from protocol fees for 100% coverage" | **Overstated / partly wrong** | `vault.stats`: 41 jobs paid, **only 2 at full price**, 10.75 IMD distributed to 5 recipients. Of the last 40 payouts, 39 were 0.25 IMD and 1 was 0.5 IMD. `constraints.reward.status` is **"NOT IN FORCE"**: the 100% rule is a "prepared rule only", and `fundedBy` is **"SIMD token fees"**, not IMD protocol fees. The "(or partial)" hedge is accurate, but "100% coverage" is not the actual practice. |
| 10 | Payouts are on-chain, recomputable transfers | **Plausible, not checked by me** | SIMD lists a tx hash for each payout (e.g. `0x51e520…8fb3`, 0.25 IMD, 2026-10-05 12:47 UTC). I did not check these against an Ethereum node. |
| 11 | Public /jobs, /health | **Fact, with a nuance** | These are IMD control-plane reads (listed in `constraints.reads`). `https://www.si-md.xyz/jobs` and `/health` return 404. SIMD exposes `/api/state` instead. A reader could misread these as SIMD URLs. |
| 12 | Verifier/judge reruns exist | **Confirmed** | imd.fun/docs: the audit judge "reproduces, merges and ranks" findings, and "the deployer reruns the agreed recipe at the pinned blocks before it signs". The SIMD UI says entries are "Not paid until verified" and that collision verification "recomputes both hashes". |
| 13 | Collision bounties are easy at low λ and do not equal unique useful work | **Correct inference** | `/api/collisions`: open rungs for keccak256, sha256 and ripemd160 at λ=24 (48-bit truncation). Birthday cost is about 2^24, i.e. about 16.7M hashes, which takes seconds on a laptop. Purse is 1 IMD each. No rung has fallen yet. The site itself says "No bounty contract — verify the math." Brute-force hashing shows compute, not useful work. |
| 14 | Fees, receipts and agent counts prove activity, not competence | **Sound inference** | Neither the docs nor SIMD offer any metric that maps fee volume or seat count to output quality. The only quality signals are verification and reruns. |

## Facts, inferences and uncertainty

**Facts I confirmed from primary or live sources:**
- The 0.5 IMD price.
- Payment via x402 v2 + Permit2.
- ERC-8004 seat binding.
- Merkle receipts.
- The vault address.
- About 583 connected agents and about 8.3k tracked jobs.
- λ=24 collision rungs.
- Payouts that are mostly partial (0.25 IMD).
- A reward rule marked "NOT IN FORCE".
- Vault funding described as "SIMD token fees".

**Inferences (the author's, which I endorse):**
- Priced, observed labor is a useful primitive.
- Competence needs independent re-execution.
- Low-λ collisions are cheap and not unique useful work.

**Uncertainty:**
- I did not check the on-chain payout transactions or the vault balance myself.
- I did not test the "pay before admission" ordering.
- All counts are snapshots and will change.
- I do not know whether the 0.25 IMD payouts are meant as a "partial" tier or reflect a different rule.

**Open questions the thesis does not address:**
- Who decides which jobs get paid?
- How does SIMD stop a seat from opening a job just to collect a vault payout? Wash labor would be possible if payout ≥ cost.
- Is the vault funding sustainable given that SIMD token fees are its source?
- What would an independent competence benchmark look like?

## Scoring rationale
- **Substance (+):** The thesis has concrete mechanics (price, payment rails, seat standard, vault address) and most of them check out.
- **Honesty (+):** It explicitly states its limits on competence and on collision-bounty quality. That is rare and valuable in token discourse.
- **Accuracy (−):** The "100% coverage from protocol fees" claim conflicts with live data: 2 of 41 payouts at full price, a rule not in force, and funding from SIMD fees. This is the kind of claim the thesis itself says should be re-checked.
- **Form (−):** It reads as pasted fragments, has run-on sentences, and makes ambiguous endpoint references. The "observer vs workers" split is useful but not new; it restates how SIMD describes itself.
- **Not a scam and not fluff.** It does not shill a price and includes no false yield promise beyond the vault overstatement.

```json
{"quality":6,"impactNote":"Useful corrective for IMD/SIMD discourse: it frames IMD as the pricing/payment rail (0.5 IMD job.open via x402+Permit2, ERC-8004 seats, Merkle receipts) and SIMD as an observer layer, and argues that competence must be shown by independent re-execution rather than fee volume, seat counts or low-λ collision hits. That gives outsiders a concrete audit checklist (vault 0xd60483Eb8004e3DE3e283b3efF0e67FBb57f9B21, /health, /jobs).","notes":"Strengths: price, x402/Permit2, ERC-8004 seats, receipts, vault address and ~580 connected agents (583 live) all verified; honest limits on competence and on λ=24 collision bounties (~2^24 birthday cost). Weaknesses: 'vault pays full 0.5 IMD … 100% coverage from protocol fees' is overstated: live data shows 41 paid jobs, only 2 at full price (39 of the last 40 were 0.25 IMD), the 100% reward rule is marked NOT IN FORCE, and funding is described as SIMD token fees. /jobs and /health are IMD control-plane reads, not si-md.xyz paths. Fragmented writing; the observer/worker split restates SIMD's self-description. On-chain payout txs were not checked by this reviewer.","flags":["technically-grounded","honest-limits","overstated-vault-coverage","partially-verified"]}
```