LH033 checkpoint, 9 November 2026
In short. By Censys's count, 18 days after PHP's emergency fix, between about 72 and 90 per cent of internet-facing services in each fixed branch that report a version are counted patched, rising slowly. About 23 million services, 52 per cent of those reporting a version, still run branches PHP did not fix. W3Techs' survey, read today for the first time since 2 November, now puts roughly twice as large a share of each branch's websites at or above the fixed release as it did a week ago. Working record, unreviewed.
Read at 14:27 UTC (Censys and W3Techs, --parts censys,w3techs; every request HTTP 200 and allowed by robots.txt, the first read under the keeper's rule of 8 November 2026, via scripts/lh_fetch.py). Censys's snapshot says it was refreshed at 14:15 UTC. Tables: data/read-2026-11-09T1427Z/tables.json. Completed days run from day 0 (21 October) to day 18 (8 November); the brief's day-30 read is due on 21 or 22 November. Written by the scheduled Monday run.
Revisions
None: days 0 to 14 are identical, field for field, to the read of 5 November (branch shares and the unfixed-tail counts). Days 15 to 18 are new.
Measure 1: share counted patched, by fixed branch (Censys)
| Branch | Day 14 (4 Nov) | Day 18 (8 Nov) | Services in branch, day 18 |
|---|---|---|---|
| 8.2 | 73.4% | 75.0% | 6,169,837 |
| 8.3 | 70.4% | 71.7% | 7,867,558 |
| 8.4 | 81.0% | 82.4% | 3,397,643 |
| 8.5 | 89.7% | 90.4% | 3,762,089 |
Each branch's share rose a little on each of days 15 to 18 (8.5 dipped by 0.02 points on day 17) while each branch's count of services fell, so part of the rise is in the denominator: the patched counts themselves fell slightly in 8.2, 8.3 and 8.4 over the four days. First day at half: unchanged (8.4 and 8.5 day 2; 8.2 and 8.3 day 6).
Measure 2: the unfixed tail (Censys)
Day 18: 23,365,120 services run a branch PHP did not fix, 52.4 per cent of the 44,563,559 that report a version (52.8 per cent on day 15); 62.3 per cent of all 118.2 million services report no version.
Measure 4: W3Techs, share of each branch's sites at or above the fixed release
| Branch | 2 November read | 9 November read |
|---|---|---|
| 8.2 (8.2.35) | 10.7% | 20.0% |
| 8.3 (8.3.36) | 12.1% | 21.4% |
| 8.4 (8.4.27) | 16.0% | 27.2% |
| 8.5 (8.5.12) | 23.8% | 38.9% |
W3Techs dates both pages "9 November 2026"; "less than 0.1%" rows are counted as 0, as before. These are shares of websites in W3Techs' surveyed sample by the version they reveal, a different unit from Censys's services, and a week's jump this size is W3Techs' figure, not checked against anything else. Description only.
Figure
Redrawn (figure.svg) and looked at rendered, light and dark: legible, with one blemish: the label "PHP 8.2 75.0%" now overlaps the label "Four Python fixes: range" at the right edge. Left for the figure's next revision; it is a working figure.
Not flagged for review: nothing here changes a public account.
From studies/LH033/checkpoints/2026-11-09.md in the repository, last changed 9 November 2026.