igneum-network/igneum
1
0
Fork 0
Code Issues Pull requests Releases Activity

adv-accept report: base rate over 143,734 accepted programs; four deep-tail seeds queued

Browse source 
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
This commit is contained in:
igneum-labs 2026-10-07 22:11:06 +00:00
parent 1b658814ef
commit 14f7a6cd7d
1 changed files with 12 additions and 12 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

24
docs/analysis/cryptanalysis/report-acceptance-rule.md
View file

@ -193,28 +193,28 @@ programs gave X_0.1% = 0.155%. If the tail scales as the extreme of the populati
times that, still under 1% of reads. The census over consecutive seeds (random accepted programs) says how
often the hot-set test fires in the population; that number lands below.
### The selector's base rate (sweep rows read at 21:55 BST: 88,051 accepted programs, every shard file de-duplicated by seed)
### The selector's base rate (sweep rows read at 23:10 BST: 143,734 accepted programs, every shard file de-duplicated by seed)
The stand-in per-site ratio at 256 units, over every accepted program so far (one row per seed; every seed
of the F8 label space yields an accepted program through the chain draw):
| Quantile | min | 0.1% | 1% | median | max |
|---|---|---|---|---|---|
| ratio | 0.9885 | 0.9980 | 0.9993 | 0.9999 | 1.0000 |
| ratio | 0.9811 | 0.9981 | 0.9993 | 0.9999 | 1.0000 |
| Threshold | Programs under it | Tries per hit |
|---|---|---|
| ratio < 0.9990 | 539 of 88,051 | 1 in 163 |
| ratio < 0.9986 | 198 | 1 in 445 |
| ratio < 0.9980 (the lowest 25 measured live sat at or under this) | 82 | 1 in 1,074 |
| ratio < 0.9970 | 35 | 1 in 2,516 |
| ratio < 0.9960 | 22 | 1 in 4,002 |
| ratio < 0.9990 | 850 of 143,734 | 1 in 169 |
| ratio < 0.9980 (the lowest 25 measured live sat at or under this) | 133 | 1 in 1,081 |
| ratio < 0.9970 | 55 | 1 in 2,613 |
| ratio < 0.9960 | 33 | 1 in 4,356 |
| ratio < 0.9950 | 10 | 1 in 14,373 |
Over the 88,051: accepted attempt 0..11 = 28,107, 19,319, 12,854, 8,758, 6,107, 4,063, 2,837, 1,933, 1,302,
874, 606, 392; max 31; mean 2.134; last-resort programs 0; a lossy last write to an output register in
73,548 (83.5 percent); register-set collapse 0. The five lowest seeds of the 88,051 (122960 at 0.9885,
130488 at 0.9890, 19921 at 0.9914, 600738 at 0.9930, 602822 at 0.9942) sit below every seed measured live so
far and are queued for the live hot-set read at 2^24 and the 2^20 ratio read (rows land below).
Over the 143,734: accepted attempt 0..11 = 46,120, 31,486, 21,002, 14,347, 9,948, 6,648, 4,515, 3,098,
2,124, 1,410, 955, 638; max 31; mean 2.122; last-resort programs 0; a lossy last write to an output
register in 120,145 (83.6 percent); register-set collapse 0. The twelve lowest: 148927 (0.9811), 122960,
130488, 150347 (0.9897), 19921, 29307 (0.9914), 34501 (0.9929), 600738, 602822, 3664, 117832, 210492; the
four not yet measured live (148927, 150347, 29307, 34501) are queued at 2^24 and at 2^20 on box 2.
Cost of the selector: one chain draw per try (about 3 s of one core: the accepted attempt's 2^20 ratio pass
dominates) plus a 256-unit ratio read (milliseconds). A seed-steering attacker who can choose among epoch