LH033 checkpoint, 5 November 2026

Working record, unreviewed. Read at 14:15 UTC (Censys only, the collector's default part; HTTP 200); Censys's snapshot says it was refreshed at 13:15 UTC. Tables: data/read-2026-11-05T1415Z/tables.json. Completed days run from day 0 (21 October) to day 14 (4 November); the brief's day-30 read is due on 21 or 22 November. Written by the scheduled Thursday run.

Revisions

None: days 0 to 11 are identical, field for field, to the read of 2 November (branch shares and the unfixed-tail counts). Days 12 to 14 are new.

Censys: share of version-reporting services in each fixed branch counted as patched (observation, its classification)

Days 12 and 13 added 0.3 to 0.9 points a day. On day 14 the shares in 8.2, 8.3 and 8.5 fell slightly, the first fall since day 8: in 8.2 and 8.5 the patched count rose but the branch count rose faster (8.2: 4.60 to 4.65 million patched of 6.23 to 6.33 million); in 8.3 the patched count itself fell by about 10 thousand (5.66 to 5.65 million) while the branch grew by 48 thousand. A rescan that adds services to a branch faster than it finds them patched lowers the share without any service being unpatched; this instrument cannot tell which. First day at half is unchanged (8.2 and 8.3 day 6; 8.4 and 8.5 day 2).

Censys: the tail no fix reaches

Day 14: 24.2 million services on branches PHP did not fix (23.9 million on day 11), 52.9 per cent of the 45.8 million that reported a version (53.3 per cent on day 11). Patched 16.5 million (15.8 on day 11); in fixed branches and not yet counted patched, 5.1 million (5.1). Services reporting no version: 74.9 million, 62.1 per cent of the 120.7 million PHP services counted (62.0 per cent on day 11).

Chart

figure.svg redrawn from this read and looked at in light and dark; the 8.2 and 8.3 end labels sit close but do not overlap. Amendment 2's S30 and L cannot be computed before days 28 to 30 are held. W3Techs is read on Mondays (row LH033-2026-11-09 names it).