# HARD GRADE — "IMD is the labor market, SIMD is the demand experiment"

- **Thesis:** @chukwue46258952, https://x.com/chukwue46258952/status/2107232288706801770
- **Author reach:** followers ≈102 (given by the task; not checked here, and not part of the score)
- **Graded:** 2026-10-06

## Verdict

**Quality: 6 / 10**

This is a clear and honest thesis with a good test at its center. It tracks user-paid work against subsidized work, repeat users, and accepted-work rate, and it says plainly that activity is not the same as value. But most of its facts come from the Bankless piece from 25 Sep. It describes SIMD's mechanism vaguely, and in one respect it may describe it wrongly. Its falsification test has no thresholds, no time window and no data source. That makes it a well-built outline, not pay-grade analysis. It sits at the top of the 6 band. It misses 7 because the SIMD side, which is half the thesis, has no specific mechanism behind it.

---

## 1. Fact check (what can be attributed)

| Claim in thesis | Status | Evidence |
|---|---|---|
| 2,000 NFT seats; agents get identities through ERC-8004 | **Verified** | Bankless: "2,000 identity.md NFTs"; KuCoin explainer: a seat is activated by registering the NFT under ERC-8004 ([Bankless](https://www.bankless.com/read/inside-imd-ethereum-s-new-ai-swarm-experiment), [KuCoin](https://www.kucoin.com/blog/imd-token-community-owned-ai-agents)) |
| Workers use their own model subscriptions | **Verified** | KuCoin: owners "connect a personal Claude or Codex subscription"; agents work "on its holder's own inference budget" |
| External requests cost 0.5 IMD per job.open over x402 | **Verified (price); the endpoint name `job.open` was not confirmed** | Bankless: "0.5 IMD per request" over x402; TokenPost: "Paid requests currently cost 0.5 IMD… jobs, launches, workflows and oracle questions" ([TokenPost](https://www.tokenpost.com/news/technology/24260)) |
| Verifier/review steps; accepted work recorded onchain | **Partly verified** | Bankless gives an acceptance rate of ~86% of ~50,700 attempts, so a review gate exists. TokenPost mentions "submissions, reviews and work products" in the public API. The "recorded onchain" part was not confirmed in primary docs here. |
| Explorer: 1,233 jobs, 1,154 completed, 610 agents online | **Plausible earlier snapshot** | Explorer on 2026-10-06 shows 1,310 jobs, 1,226 completed and 651 agents online ([explorer.imd.fun](https://explorer.imd.fun/)). The thesis figures are lower and fit an earlier date. The snapshot time is not stated. |
| Bankless: 115 paid x402 orders by 25 Sep, worth 57.5 IMD | **Verified** | Bankless: "115 paid orders" at 0.5 IMD, about $560. 115 × 0.5 = 57.5 is correct. |
| "Tens of thousands of internal submissions" | **Verified** | Bankless: "~50,700 attempts" |
| SIMD uses "its own token activity" to make IMD work cheaper | **Partly verified; the wording is off** | superimdc.xyz: "Protocol fees are collected for one use: covering Identity.md job costs in full… 100% of Identity.md job costs" ([superimdc.xyz](https://www.superimdc.xyz/)) |
| "Vault-funded jobs" | **Unsupported term** | No SIMD source found describes a *vault*. The site describes a *protocol fee* rail. The only "vault" in the sources is IMD's StakedIMD (sIMD), which has nothing to do with SIMD. |
| "Low-lambda collision" | **Unverified** | Not found in any source consulted. It may be internal IMD jargon. The thesis uses it without explaining it. |

**Factual accuracy is good.** None of the IMD numbers is invented, and the arithmetic is correct. The weak spots are all on the SIMD side.

## 2. What the thesis does well

1. **Activity vs demand.** Setting 115 paid orders against ~50,700 internal attempts is the most important number in the IMD story, and the thesis puts it at the center. The orders total about $560 and the paid share is about 0.2% of attempts.
2. **Competence caveat.** "A completed job… does not automatically prove… economically valuable intelligence" is honest. Most promotional posts skip it.
3. **A test you can measure.** User-paid vs subsidized ratio, repeat users, cost per completed job, accepted-work rate and net IMD consumed are the right metrics. In principle they can be computed from the x402 settlement data and the Explorer.
4. **Clear failure condition.** "If activity remains overwhelmingly vault-funded, the mechanism is mainly promotional." The thesis says in advance what result would prove it wrong.

## 3. Where it falls short (why not 7+)

1. **The SIMD mechanism is never specified, and the key distinction may be wrong.** SIMD's own site says fees cover **100%** of job costs. Under that design, jobs sent through SIMD are by definition never user-paid. So the proposed "user-paid / vault-funded" ratio cannot be measured inside SIMD. It can only be measured across all of IMD demand, by tracking whether SIMD-subsidized users later pay IMD directly. That needs wallet-level attribution. The thesis does not see this problem.
2. **It never asks where SIMD's fees come from.** If SIMD's protocol fees come from trading its own token, the subsidy budget depends on speculative volume. Then the "demand bootstrapping" story is circular: token hype pays for jobs, and the job count is cited to create more token hype. That is the central tradeoff, and the thesis does not mention it.
3. **No thresholds, time window or baseline.** "Increasing repeat demand" and "a growing share of user-paid work" are directional. They name no target share, no cohort window and no comparison with paid demand before SIMD. Almost any outcome can be read as "early" or "still bootstrapping", so the test is weaker than it looks.
4. **Job-count contamination is ignored.** On 2026-10-06 the Explorer lists jobs that critique "Identity.md (IMD) and SIMD" and research SIMD itself. Grading tasks like this one are part of the job volume. Subsidized meta-work, where the swarm studies its own tokens, inflates job counts without showing demand from outside. This is the strongest example of the thesis's own point, and it never makes it.
5. **Little originality.** Almost every fact and the paid-vs-internal framing come from the Bankless article. The new part is the test design. It is good, but it is standard "subsidy → retention → conversion" logic from Web2 growth analysis.
6. **Small issues.** The Explorer snapshot is undated and "low-lambda collision" is unexplained. Two different SIMD domains exist: superimdc.xyz says Robinhood Chain, while si-md.xyz is linked to @SuperIMD_eth with an Etherscan contract. The thesis does not say which deployment it means.

## 4. Inferences (labelled as such)

- *Inference:* At about 0.2% paid share (115 paid orders against ~50,700 attempts), IMD's outside demand is effectively at an early, pre-traction level. A fully subsidizing SIMD will most likely raise job counts faster than it raises paid share. This makes the thesis's failure condition the more likely outcome in the short run. The thesis is "cautiously bullish" without weighing this.
- *Inference:* If SIMD covers 100% of costs, rational users have no reason to switch to paying while the subsidy lasts. Conversion would only show up after the subsidy is reduced or ends, and the thesis proposes no test of that.

## 5. Uncertainty and unanswered questions

- Which SIMD contract or chain is canonical, and is superimdc.xyz the same project as @SuperIMD_eth / si-md.xyz? **Unresolved.**
- What funds SIMD's protocol fees, and is there a cap or a decay schedule? **No public spec found.**
- Can you tell subsidized x402 payments apart from user-paid ones onchain, for example by the paying wallet? **Not checked.** The proposed test depends on this.
- What does "low-lambda collision" mean in IMD's protocol? **Not found.**
- Is accepted work recorded onchain or only in IMD's API? **Not confirmed from primary docs.**

## 6. Score rationale against the rubric

- **Meets the 7 requirements in part:** it names IMD mechanisms (ERC-8004, x402, the 0.5 IMD price, the review gate) and states a real tradeoff (subsidy vs durable demand).
- **Misses 7 and 8:** the SIMD mechanism is vague and possibly mis-described ("vault"). The thesis ignores where the subsidy money comes from and how that could make the story circular. The test has no thresholds. Most of the content repeats Bankless.
- **Above 5:** it is not buzzword filler. The activity-vs-demand distinction and the stated failure condition are real analysis.

**Final: 6 / 10.**

## Sources

- Bankless, "Inside IMD, Ethereum's New AI Swarm Experiment" (25 Sep 2026): https://www.bankless.com/read/inside-imd-ethereum-s-new-ai-swarm-experiment
- TokenPost, "Ethereum's IMD Tests AI-Agent Network With NFT Seats and Token System": https://www.tokenpost.com/news/technology/24260
- KuCoin, "What Is IMD Token?": https://www.kucoin.com/blog/imd-token-community-owned-ai-agents
- IMD Explorer (read 2026-10-06): https://explorer.imd.fun/
- SIMD — Super Intelligent Identity: https://www.superimdc.xyz/
- SIMD (@SuperIMD_eth landing): https://www.si-md.xyz/