At 13:04 UTC, Sui processed 1,680 eligible transactions in one of the corpus’s busiest bursts: the median gap was 0 ms, the 90th-percentile gap was 210 ms, and the longest idle period was only 655 ms. The minute recorded 76 failures, or 4.5% of eligible transactions—below the frozen corpus rate of 8.36%—but its activity was unusually concentrated.
The largest sender, 0xcde6dbe01902be1f200ff03dbbd149e586847be8cee15235f82750d9b06c0e04, submitted 453 of the 1,680 transactions. Among the 1,589 transactions with an asserted semantic reading, the leading activities were 357 limit-order placement attempts, 313 cancellation attempts, and 226 price-update workflows. This repeated cadence is consistent with automated execution, but the address itself does not establish a person, company, or bot.
The money-sized exception
The minute’s standout movement was not a single order. The sender-boundary record for ArDRFoAn7DyDNfHK4xUYvYUdwJaawHY6zMDfsjKW2XKu shows 0x6728607fe261c0bbf219721c655f50567812d9419401d3d8ee8c5c05a90318d3 receiving 516,708.482979825 SUI and 125,230.10514 USDC after a reward-collection attempt followed by a liquidity-position closure.
A nearby ledger row for the same sender shows an earlier open-liquidity-position movement of 516,708.237748041 SUI out. Together, those rows resemble an open-and-close cycle with a small SUI difference, but they do not establish profit or explain the complete economic history of the position. The transaction detail is explicit about the sequence: “First, it ran a claim call. Then, it ran a position-closing call.”
Across all senders, sender-holding movements recorded 1,385,517.818183969 SUI in and 1,540,062.430292781 SUI out during the minute. That is a boundary accounting result, not a price-based valuation. Gas must be kept separate: the minute’s aggregate gas field reports 2,096,481,834 MIST.
A repetitive loop with a noisy failure tail
The same top sender accounted for 61 of the 76 failures. Fifty-nine failure records had no named abort site, and many repeated the short structure associated with DeepBook owner-proof activity. The successful cancellation transaction 3a2Ykd5jSQ9qZ24rihK58HuzoLQ2dZrswwrrrvpuuBDM shows the concrete two-step form: generate an owner proof, then call pool::cancel_order. Its detail binds a resting NS/USDC price of 0.01222 USDC per NS and 7,019 NS of resting size removed from the book. That is cancellation evidence, not proof that an order traded.
The minute also contains a different failure mode. In HFZYTgtZrcpVwdsc1AEnmwhcvPG6Ths5vN34hmVqE2ga, a 12-step partner flash route reached a minimum-output check and aborted at validation::check_min_amount. The record says the minimum-output guard was 121,242,533; it does not say that this amount was received. Application effects were rolled back, leaving only the attempted route and gas movement.
Of the 1,589 transactions with an asserted semantic reading, 17 failed. The other 59 eligible failures were outside that denominator, which is why the failure rate for understood activity is not the same as the rate for the minute as a whole. Overall, 1,477 of 1,680 transactions received a complete structural classification, below the corpus-wide 92.2% rate but above the previous minute’s 89.0%.
What changed in the minute
The repertoire shifted visibly within the 60 seconds. Placement and cancellation dominated the opening five-second buckets; price updates briefly led around 13:04:10; then a group of 24 of 118 transactions at 13:04:15 had only call-level composition rather than a whole-transaction interpretation. By 13:04:25, placement and cancellation were again the two leading activities, with 59 and 50 transactions respectively.
Thirteen exact revisions have their earliest stored calls in this minute, only a little more than two minutes after the frozen corpus began at 13:01:48.968 UTC. That boundary-adjacent count does not establish that code arrived or was deployed at 13:04. The more concrete story is the combination of a highly repetitive order-maintenance stream, a cluster of unnamed failures, and one unusually large liquidity-position boundary movement.
Useful exits: ArDRFoAn7DyDNfHK4xUYvYUdwJaawHY6zMDfsjKW2XKu, 3a2Ykd5jSQ9qZ24rihK58HuzoLQ2dZrswwrrrvpuuBDM, and sender 0x6728607fe261c0bbf219721c655f50567812d9419401d3d8ee8c5c05a90318d3.