Replay of round 17
One round of the memory arm, step by step: what it read from Walrus Memory, what it decided to buy and why, the payment on Sui, how the answer was graded, and the belief version it wrote back.
Memory arm: 4 paid answers for 0.050 USDC, 1 refusal, 0 wrong answers accepted. Blind arm, same round: 2 paid answers for 0.020 USDC, 0 wrong answers accepted.
1Read memory
Before buying anything, the agent recalls its beliefs from Walrus Memory. These are the newest version of each belief written before round 17, which is what a recall returns. The run did not log its recalls; the trace flag that prints them was added for the 2026-11-03 rerun, where the recalled versions matched the newest ones written.
Seller For Version Trust Last result Cooldown until Blob Control seller (ours) market prices v9 0.53 checked, right – TUc5_XCd…Kd_sSui Address Inspector address facts v14 0.77 checked, right – WgNcud4x…rZlcPipTradeDex market prices v12 0.85 checked, right – 5y7El70g…9QJAControl seller (ours) address facts v5 0.47 checked, right – 42oiuPK2…aIk4AgentProof endpoint checks v3 0.21 failed, nothing delivered 16:25 UTC Nu3kUpVC…BnW8Omnia endpoint verify endpoint checks v3 0.21 dead route 16:25 UTC F_dcq_Fb…d6IAPipTradeDex swap quotes v6 0.50 bought on trust, not checked – 0EJ3DoUF…VkQoControl seller (ours) swap quotes v2 0.50 bought on trust, not checked – oO2zUouH…YYDQ2Decide
The trust policy is plain code over those beliefs: below the 0.6 floor a seller gets only canaries; in a cooldown it gets nothing; above the floor, the cheapest good seller wins, with spot checks that thin out as trust grows. If a checked answer fails, it tries the next seller once in the same round.
address factsSui Address Inspector independent, as a purchase on trust. Reason it gave: “cheapest good answer at trust 0.77”.
Control seller (ours) control (ours), as a canary (a question it can check, to earn trust). Reason it gave: “canary: earning a place”.
market pricesPipTradeDex independent, as a purchase on trust. Reason it gave: “cheapest good answer at trust 0.85”.
Control seller (ours) control (ours), as a canary (a question it can check, to earn trust). Reason it gave: “canary: earning a place”.
endpoint checksRefused to buy. Reason it gave: “no seller is trusted and none is due a canary”.
3Pay
Each purchase is an x402 payment in USDC on Sui mainnet. Gas price is 0: these are gasless transfers.
Sui Address Inspector independent · address facts- Paid
- 0.010 USDC
- Sui transaction
B8HMU8DDk3SKuch1vjYenFQdg6JakbLwhVnoqTQkshXA- Answer in
- 0.8 s
- At
- 16:08:24 UTC
Control seller (ours) control (ours) · address facts- Paid
- 0.010 USDC
- Sui transaction
HNTUKHNdRC9dZv7cojuBcdju2CqEcRD6vwKqt1qTxDnX- Answer in
- 0.8 s
- At
- 16:08:30 UTC
PipTradeDex independent · market prices- Paid
- 0.020 USDC
- Sui transaction
5QKbtsZ6Bfg5buipozuLjY3PeTN4V75GnuK8hM7a5gCa- Answer in
- 9.9 s
- At
- 16:08:47 UTC
Control seller (ours) control (ours) · market prices- Paid
- 0.010 USDC
- Sui transaction
AFurXh4yxsszrLEGiP5XdAsjJukW3L6vmenaDcsHgR5t- Answer in
- 0.7 s
- At
- 16:08:51 UTC
4Grade
Its own check compares the answer with chain truth read through Tatum: balances for address facts, the deepest native USDC/SUI pool for market prices. Answers bought on trust are not checked by the agent. The audit is the experiment's scoring, which checks every answer; the agent never sees it.
Sui Address Inspector · address facts- Its own check
- bought on trust, not checked
- The audit
- right (all 3 balances matched the chain)
Control seller (ours) · address facts- Its own check
- checked, right (all 4 balances matched the chain)
- The audit
- right (all 4 balances matched the chain)
PipTradeDex · market prices- Its own check
- bought on trust, not checked
- The audit
- right (price within 0.02% of the reference)
Control seller (ours) · market prices- Its own check
- checked, right (price within 0.11% of the reference)
- The audit
- right (price within 0.11% of the reference)
5Remember
Each purchase becomes a verdict and a new version of that seller's belief, both written to Walrus Memory. The next round starts by reading these.
Sui Address Inspector · address facts- Verdict memory
NPUyGV7v…0xjYintwiceshy-verdicts- Belief written
- v15: trust 0.77 → 0.77. bought unchecked on a trust of 0.77
- Walrus blob
im4qb-GBrm1xyGZaqPbQ5Ae0E7Mx5Ragz0NbwBZpOjQintwiceshy-belief-inspector-address-facts
Control seller (ours) · address facts- Verdict memory
YHYFyx20…kIh4intwiceshy-verdicts- Belief written
- v6: trust 0.47 → 0.53. passed: all 4 balances matched the chain
- Walrus blob
WycisxkndBbksjgOa6diaLIHmDt-jAJdEE3HjSRsEUwintwiceshy-belief-control-address-facts
PipTradeDex · market prices- Verdict memory
X-8968FW…c2v4intwiceshy-verdicts- Belief written
- v13: trust 0.85 → 0.85. bought unchecked on a trust of 0.85
- Walrus blob
Q9vx_EIlQLagVVnHFTQ6gOzT9AeyIJCsEziU7K6nNzgintwiceshy-belief-piptradedex-market-price
Control seller (ours) · market prices- Verdict memory
la9-XeWK…4Zgcintwiceshy-verdicts- Belief written
- v10: trust 0.53 → 0.58. passed: price within 0.11% of the reference
- Walrus blob
R2-j7arffH6AN8cG7r-LqERfaBW40amQXrS2S9rHlMUintwiceshy-belief-control-market-price
The blind arm in the same round
Same market, same round, no memory and no checks: it takes the cheapest listing every time.
| For | Seller | USDC | Transaction | Audit |
|---|---|---|---|---|
| address facts | Sui Address Inspector independent | 0.010 | 4srAPDG6…VPY2 | right (all 1 balances matched the chain) |
| market prices | Control seller (ours) control (ours) | 0.010 | G2eZQvsK…vS77 | right (price within 0.20% of the reference) |
| endpoint checks | AgentProof independent | 0.000 | – | failed, nothing delivered: signed payment offered, seller returned HTTP 500, nothing settled |
| endpoint checks | Omnia endpoint verify independent | 0.000 | – | dead route: HTTP 404 before any payment |
Other rounds
Round 16, before this one. Round 18, after it. Round 7 is the first round after our control seller's fault was caught.