# Thesis score: "$SIMD protocol fees could fund useful agent work" (@unlistd1)

- **Tweet:** https://x.com/unlistd1/status/2107085689548427267. I couldn't fetch it: X returned HTTP 402. This review scores the thesis text exactly as it was given in the task.
- **Author:** @unlistd1. Follower count was not scored, as instructed.
- **Review date:** 2026-10-05

## 1. What the thesis claims

1. @SuperIMD_eth could use "$SIMD protocol fees" to pay for agent jobs.
2. Subsidized jobs make it cheaper for developers to try the network. If the results are useful, they come back, build applications and bring in users.
3. If that growth raises fee income, there is money for more jobs, which creates a flywheel.
4. "Collision bounties" and an experimental build-and-review pipeline make the agents' output independently checkable.
5. A self-imposed condition: the loop is only sustainable if repeat demand and fee income grow together. Subsidized job volume alone does not prove it.

## 2. Evidence (facts, with sources)

| # | Fact | Source |
|---|------|--------|
| F1 | Paid requests cost a fixed 0.5 IMD per action, paid via x402 + Permit2 on Ethereum mainnet. | [mtezy/imd-writeup](https://github.com/mtezy/imd-writeup), [TokenPost](https://www.tokenpost.com/news/technology/24260) |
| F2 | sIMD is an ERC-4626 share from the StakedIMD vault. It is not a separate protocol. Its yield comes from the POOL4 hook: of the IMD trimmed when sells push the pool over its cap, 85% is burned, 6% goes to orchestrator compute, 4.5% to sIMD stakers and 4.5% to NFT seats. | [KuCoin blog](https://www.kucoin.com/blog/id-imd-token-community-owned-ai-agents), [Bankless](https://www.bankless.com/read/inside-imd-ethereum-s-new-ai-swarm-experiment) |
| F3 | About 32% of supply (≈2.3M IMD) was staked as sIMD in late September 2026. Supply is ≈7.1M, down from 10M because of burns. | Bankless (data as of 2026-09-25) |
| F4 | Demand is small: only 115 paid public orders (about $560 in IMD), against about 43.8k accepted internal submissions and about 17.3B inference tokens used. | Bankless |
| F5 | There is a verification pipeline. A verifier rebuilds each submission in a sealed container and checks which files changed. Other seats review the work adversarially, and accepted work is recorded on-chain as reputation. Acceptance rate is about 86%. | Bankless, KuCoin, [explorer.imd.fun](https://explorer.imd.fun/agents) |
| F6 | The 6% compute reserve already pays for orchestrator compute from trading activity. Seat operators provide their own model subscriptions. | KuCoin, TokenPost |

## 3. Inferences (my reasoning, not sourced facts)

- **I1 – The mechanism is misnamed.** "$SIMD protocol fees" doesn't match the public design (F2). sIMD receives a 4.5% share of trims caused by sell pressure. It does not collect fees. Funding jobs "from sIMD" therefore means one of two things: redirecting staker yield, which is a governance and holder-incentive trade-off the thesis doesn't mention, or changing the POOL4 split. The thesis doesn't say which, or who would decide.
- **I2 – The funding is tied to the wrong thing.** Trim income depends on sell volume above a cap that steps down over time (F2). It does not depend on job demand. A flywheel needs job usage to raise the fees that fund jobs. In the current design that link is indirect at best: usage → token demand → trading → trims. The thesis says "if that growth also increases the fees", which is the right conditional, but it doesn't notice that the current design doesn't provide this link.
- **I3 – The idea fits the data.** With 115 paid orders (F4), adoption is clearly limited by demand, not supply. Subsidizing trial jobs to find builders is a reasonable response. The thesis's own test (repeat demand plus fee growth, not raw subsidized volume) is the right metric and avoids vanity-metric thinking.
- **I4 – The scale is unknown.** At 0.5 IMD per job (F1), even small fee flows could pay for many jobs. The thesis gives no numbers, so we can't tell whether the idea matters at real scale.
- **I5 – "Verifiable" is partly supported.** Independent checking via the sealed-rebuild verifier and adversarial review is backed by public sources (F5). "Collision bounties" are not (see U2).

## 4. Uncertainty and unanswered questions

- **U1:** I couldn't verify @SuperIMD_eth's role: whether it's official, a community account or a proposal account. I found no public source for it.
- **U2:** I found no public primary source on "collision bounties". They may be a recent or internal feature, so this claim is unverified rather than false.
- **U3:** Data from 2026-09-25 may be out of date by 2026-10-05, since the project moves fast.
- **U4:** Open questions the thesis leaves unanswered:
  - What share of the fee or trim flow would be redirected?
  - How would sybil or self-dealing "experiments" be prevented? That is the classic failure of subsidized usage.
  - What repeat-usage threshold would count as success?

## 5. Assessment

**Strengths**
- It is coherent and well hedged. It names its own falsification condition and explicitly rejects "more subsidized jobs" as proof.
- It responds to the real bottleneck: low paid demand (F4).
- It correctly treats checkable output as a selling point (F5).
- It contains no price talk and no shilling.

**Weaknesses**
- It mislabels the mechanism: sIMD is a staking share, not a fee protocol.
- It ignores the trade-off with staker yield, and that trim income depends on sells rather than on usage.
- It gives no numbers.
- The "flywheel" framing is generic.
- It relies on an unsourced feature (collision bounties).

**Score:** 6/10. This is above-average discourse: honest, conditional and relevant. Its originality and technical precision are limited.

```json
{"quality":6,"impactNote":"Shifts IMD/SIMD discussion from token mechanics toward the real bottleneck (paid demand: ~115 public orders) and proposes a falsifiable success test — repeat demand and fee income growing together, not subsidized job counts.","notes":"Strengths: coherent, self-hedged, names its own sustainability condition, aligns with verifiable-output pipeline (sealed verifier + adversarial review). Weaknesses: calls sIMD a fee-generating protocol when it is an ERC-4626 staking share receiving 4.5% of POOL4 sell-side trims; ignores staker-yield tradeoff and that trim income tracks sells not usage; no numbers; generic flywheel framing; 'collision bounties' not verifiable in public sources; @SuperIMD_eth role unverified.","flags":["constructive","hedged","mechanism-imprecise","no-quantification"]}
```