LH033 interim checkpoint, 30 October 2026: measures 1 and 2 to day 7
Interim, unreviewed. The brief allows an interim checkpoint after the read of 29 October, with day 7 complete, reporting measures 1 and 2 to day 7. This one uses that read only (data/read-2026-10-29T0635Z/tables.json, Censys refreshed 29 October 06:15 UTC). The 30 October read's day 8 is in the daily checkpoint and is not used here. No figure in days 0 to 7 was revised between any two of our reads. Measures 3 to 5 wait for the read holding day 30. Written by the scheduled daily run of 30 October.
Measure 1: share of version-reporting services in each fixed branch counted patched by Censys (a stock of services as last scanned)
| Branch | Day 1 | Day 2 | Day 7 | First day at half |
|---|---|---|---|---|
| 8.2 | 22.1% | 41.2% | 53.0% | 6 |
| 8.3 | 23.8% | 42.2% | 52.4% | 6 |
| 8.4 | 35.3% | 58.5% | 67.8% | 2 |
| 8.5 | 36.1% | 70.5% | 78.8% | 2 |
Measure 2: the unfixed tail
Services on branches PHP did not fix were 23.3 million on days 0 to 2 and 24.4 million on day 7, as a share of version-reporting services 57.2 per cent on day 0, 55.3 on day 2 and 54.3 on day 7. Services reporting no version were 28.4 per cent of all PHP services counted on day 0 and 59.5 per cent on day 7; the jump came on day 2 (23 October), when Censys's total count of PHP services rose from 61.4 to 95.8 million in a day. The total kept growing, to 110.9 million on day 7.
Beside the Python comparators (named in the brief; a different unit)
The Python figures are a day's at-or-newer share of a library's downloads from the Python Package Index (data/python_side.json, from LH011 and LH008): a flow of fetches, not a stock of services.
| Day | PHP Censys patched share, 8.2 to 8.5 | Four Python security fixes, download share | Median of 37 projects' medians |
|---|---|---|---|
| 1 | 22.1 to 36.1% | 53.8 to 60.8% | 57.6% |
| 2 | 41.2 to 70.5% | 52.7 to 63.7% | 57.2% |
| 7 | 52.4 to 78.8% | 55.6 to 59.5% | 57.9% |
As descriptions only: on day 1 every PHP branch's Censys share was below every Python fix's download share; on day 2 the 8.5 share was above the four Python fixes, 8.4 within their range and 8.2 and 8.3 below it; on day 7 8.2 and 8.3 were below the Python range and 8.4 and 8.5 above it. On pins, 18 of LH008's 102 repository pairs on the four Python fixes had moved within 7 days. The units differ (services as last rescanned against requests in a day), Censys's patched share is a floor where distributions backport fixes without changing the version, and its population changed on day 2. No statement here says which ecosystem is faster or why.
From studies/LH033/checkpoints/interim-2026-10-30.md in the repository, last changed 30 October 2026.