adv-accept report: the 00:00 BST reading drafted at 23:27 (153,158 programs; count and pool state refreshed at the hour)

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

docs/analysis/cryptanalysis/report-acceptance-rule.md

@@ -813,18 +813,20 @@ among adv waiters is not strictly by arrival), so every shard of mine on box 2 w
 (s08 at 346 rows, frontier 800349) until gap-deep4 held its 8 cores, then the five shards re-queued from
 their frontiers (s08 from 800349, s06c from 609791, s04b from 400603, s02e from 221880, s00g from 40822).
 
-## Where the pass stands (for the 00:00 BST reading; numbers refreshed as rows land)
+## Where the pass stands: the 00:00 BST reading (drafted 23:27 BST; the program count and the pool state are refreshed once more at the hour)
 
-| Item | Reading (20:50 BST; rows that land later are added above this table) |
+| Item | Reading |
 |---|---|
-| Box-hours spent (running time, both boxes, nice 10) | about 3.3: the two shards at 88 threads for 1.9 h of wall under contention (20:27 to 21:21 BST and the re-launch), two adv-live confirmation chains, the 10^6-nonce census (51 programs), the gap tool (61 programs), the row-90 plant, the class v5 check, and shard 01b at 32 cores since 20:40; pod-hours 0 (no GPU, no pod). Runs killed at 20:21 BST by main's order and re-queued through the lease are counted to their kill |
-| Q1 bound reached | 29,032 accepted programs drawn through the chain path, of the 10^6 planned (shards 00b, 02 to 09 queued in the lease pool, 0 free cores at 20:48 BST). Live at 2^24 nonces: 12 lowest stand-in-ratio seeds, 5 beyond the f8 1.2x gate and 7 within; 17 random, 1 beyond; the hot-set test X_f >= f fired on 1 of 29 (seed 100767, X_0.1% +0.155%, 2.05x), and that hot set is 1 MB of items holding 0.31 percent of reads: gain 1.002x, no row of chip-model-v3.md moves. It survives the class v5 dataset unchanged (pre-floor draw; the frozen v5 floor (c''') refuses it at attempt 2) |
-| Q2 bound reached | 54 programs: no stand-in gap above the 2^20 sampling noise (max 0.0004 per site), 0 verdict disagreements; the zero-dataset plant fires |
-| Q5 | lossy last write in 83.6 percent of 29,032 accepted programs with no live effect found beyond Q1's; register-set collapse 0; the 0.98 floor is live in the tail (0.9832 at 2^20 seen); the 256-unit proxy is a weak selector (7 of 12 false positives against the f8 gate), the 2^20 ratio the real one |
-| Row 90 (owner adv-accept-3, unspawned; claimed) | plant fired (a rule that rejects everything reaches the cap and hands the last resort, printed and re-checked: accepted as drawn, no lossy op, distinct mean 127.9 to 128.0); the 20k-seed census of real attempts is queued on build-1 (killed at 20:21 before any summary, re-submitted 20:42) |
-| BLOCKED | per-card hash-rate confirmation of any gain (no GPU tonight); priced analytically against chip-model-v3.md instead |
-| Partial, named as partial | 29,032 of 10^6 seeds; 12 of 25 lowest-ratio live rows; 17 of 20 random live rows; row 90 census not run; shards 02 to 09 not run beyond 2,474 rows of 02 |
-| What a longer pass adds | the other 97 percent of the 10^6 seeds (the tail by the 2^20 ratio the rule computes, not the 256-unit proxy), the live row for every seed under 0.985 at 2^20, and the 20k-seed attempts census: more rows of the same shape, not a different answer unless a seed reads a hot set over 1 percent of reads, which 29,032 draws and 29 live rows did not produce |
+| Box-hours spent (running time, both boxes, nice 10; pod-hours 0, no GPU, no pod) | about 5.0: the two 88-thread shards for 1.9 h of wall under contention until the 20:21 kill, then shards at 32 or fewer cores through the lease pool with repeated releases to the class v5 work (every release by pgid file, rows kept, re-queued from the recorded frontier); two adv-live confirmation chains, the 10^6-nonce census (51 programs), 34 live rows at 2^24 of the stand-in tail plus 20 random, the gap tool on 61 programs plus 17 decisive 2^20 reads, row 90 on 10,000 seeds, the class v5 exemplar check |
+| Q1, the bypass: bound reached | 153,158 accepted programs drawn through the chain path at 23:27 BST (of the 10^6 planned; shards 00 to 09 all running or queued from their frontiers). Eight pass every part of the rule and flag the f8 hot-set test on the LIVE dataset at 2^24 nonces (100767, 4346, 5245, 1738, 170, 106924, 107022, 122960: X_0.1% +0.102 to +0.221 percent, 1.54x to 2.24x over the window model), all eight in the lowest 34 stand-in-ratio seeds; 20 random accepted programs: 1 beyond the f8 gate, 0 hot sets. Every hot set is the top 0.1 percent of items (about 1 MB) holding 0.26 to 0.41 percent of reads against 0.15 expected: an on-die copy saves under 0.3 percent of DRAM reads, gain 1.002x at the largest; no row of chip-model-v3.md moves. Mechanism: a load site whose source reaches zero or all-ones (or one bit off) in hundredths of a percent of evaluations, under the (c') 1 percent limit, mapped by the era stride to one fixed item (item 0x000000 the hottest in four); two further shapes seen (a low-entropy band without saturation; one constant image from a non-saturated source, seed 29307). The exemplar reads the same hot set under the class v5 dataset (the pre-floor draw) |
+| Q1 against the class v5 per-site floor at 0.995 | all 8 live hot sets refused (minimum sites 0.9821 to 0.9919 at the acceptance's own 2^20 sample); 3 mild concentrations missed (Devnet 3 epoch 0 site 0 at 0.9992, adv-cache-2's era-fixed-20 at 0.9997 and era-drawn-2 at 0.9992: about 1.0004x to a chip, adv-cache-2's rotation class); 6 clean-live programs refused and 1 clean passed among the 9 deepest 256-unit seeds (the floor's 2.4 percent clean rejection) |
+| The selector | the 256-unit stand-in ratio enriches the live gate 13x (17 beyond of the lowest 25 against 1 of 20 random) and finds the hot sets (8 of 34 against 0 of 20), with false positives 8 of 25 on the gate and 26 of 34 on the hot-set test; below 0.995 it is not sampling noise but a different fact: a low ratio is a thin spread of repeated indices, not a hot set. The 2^20 ratio the rule computes is the real selector; one chain draw costs about 3 s of one core and a seed under 0.998 costs about 1,000 draws |
+| Q2, the stand-in gap: bound reached | 54 programs plus 17 decisive reads: closed-form and live per-site ratios agree to 0.0004 at 2^20 (the sampling noise), 0 verdict disagreements, (c) metrics agree to 0.019 of 128; the zero-dataset plant fires. No steering through the gap |
+| Q5, distinguishers | a lossy last write to an output register in 83.5 percent of 153,158 accepted programs, no live effect beyond Q1's; register-set collapse 0; the 0.98 floor is live in the tail (0.9811 at 256 units, 0.9814 at 2^20 seen) |
+| Row 90 (owner adv-accept-3, unspawned; claimed) | 10,000 seeds: rejection by part (a') 83.3, (a) 11.7, (b) 3.1, (c'') 1.1, the rest under 1 percent; per-candidate rejection 0.6787 flat across attempts; P(exhaust) 8e-44; 0 exhaustions; the last resort reachable only by a rule that rejects everything (plant fired) and then accepted as drawn; the second 10,000 queued |
+| BLOCKED | per-card hash-rate confirmation of any gain (no GPU tonight); priced analytically against chip-model-v3.md |
+| Partial, named as partial | 153,158 of 10^6 seeds; 34 of the stand-in tail and 20 random measured live; row 90 at 10,000 of 20,000; the 2^20 read of seed 3664 and the live attribution of the clean deep seeds not taken further |
+| What a longer pass adds | the other 85 percent of the 10^6 seeds, the 2^20 ratio (not the 256-unit proxy) as the ranking, and a live row for every seed under 0.985 at 2^20: more rows of the same three shapes, not a different answer unless a seed reads a hot set over 1 percent of reads, which 153,158 draws and 54 live rows did not produce. The open design question is adv-cache-2's: the stride rotation that lets a product's low bits into the item index, a load_index change rather than a floor |
 
 ## Running notes