The first full minute of the frozen Sui slice contained 2,034 eligible transactions between checkpoints 300444504 and 300444780. The corpus itself begins at 13:01:48.968 UTC, 11 seconds before this minute. It was an unusually compressed minute: the median gap was 0 ms, the 90th-percentile gap was 207 ms, and the longest idle period was only 441 ms. The activity came from 348 senders, with 552 transactions sponsored by a different gas owner.
The failure rate was elevated but not universal: 258 of 2,034 eligible transactions failed, or 12.68%, compared with 8.36% across the frozen corpus. After excluding 24 transactions that never reached meaningful execution and one contradicted record, 252 of the remaining 2,009 had failed. Purpose-level explanations covered 1,788 of the 2,034 rows.
Swap routes supplied the pressure
Among the 2,009 transactions with usable behavioral explanations, execute-swap-route appeared 201 times. 136 of those attempts failed, making route execution the largest identifiable contributor to the minute’s failures. The broader activity still looked like order maintenance: cancel-order accounted for 464 transactions and place-limit-orders for 409.
The failure sites were varied rather than pointing to one universal break. Of the 258 failures, 57 had no recorded abort location, 41 stopped at flash_loan::return_flashloan_quote, 28 at a slippage check, 27 inside a pool swap, and 26 while splitting a balance. Reaching one of these calls proves an attempted operation, not a committed application effect; aborted application changes were rolled back.
Transaction AJzRpTJK shows the pattern clearly. It first submitted nonce 43459, then attempted a Bluefin A-to-B conversion and aborted at c::deposit. The record explicitly says the authorization and conversion were one atomic attempt. Its sender-boundary evidence shows −0.00117018 SUI in the holdings category, while gas accounting separately records −0.00117018 SUI; those figures must not be added together.
A successful-looking route was still not enough to establish an exchange. DJnv8kev prepared inputs and invoked a pool A-to-B wrapper touching JACKSON and SUI. Its boundary movements were −1.681 JACKSON and +0.031449157 SUI, but the transaction narrative explicitly withholds the amount exchanged and the value ultimately held.
A corpus boundary, not a deployment event
The narrower roster used for this article retains the 12 largest later-footprint revisions among those with earliest stored calls in this minute. Those 12 account for 466 revision incidences here and 123,631 across the corpus. Across the unfiltered called_revisions field, the store places 126 exact revision minima at 13:02, after 52 in the final 12 seconds of 13:01 and before 37 at 13:03. Both the 12 and 126 counts describe filters over the observation boundary. Neither is a deployment count, and the 13:02 peak is not an on-chain launch event.
One concrete boundary example is revision 0x7432b455fc6441adb13f7b1f07747034b9c28abdbec082d1ace42c79c3104ff5, seen in EvjtwWfn. That transaction is a one-step vault::deposit: it placed a whole SUI coin into shared vault storage, with sender-boundary evidence of −0.049218447 SUI. This is the revision's first stored use in the bounded corpus; it may have been deployed and used before observation began. The deposit does not explain the swap-route failures.
The minute also carried substantial boundary movements: sender-holding evidence recorded −183,003.717156127 SUI, +709,829.121996 USDC, and −547,228.972222646 WAL. These are per-asset boundary totals, not proof of trades, profit, or cross-asset value. Separately, recorded net gas was 2.591544840 SUI.
The useful reading of 13:02 is therefore one crowded observed minute: a broad order-management workload plus a concentrated set of route attempts that often stopped before their economic outcome could be established. The revision minima describe the corpus starting to see an already-active contract surface. The failure burst remains a distributed execution pattern rather than a demonstrated single-contract outage.
For a closer look, start with EvjtwWfn for the boundary-observed vault revision, AJzRpTJK for the aborted route, and DJnv8kev for a route whose boundary movements still do not prove an exchange.