ArchiveSCV

Corrections

What the archive got wrong, and what it says now.

This log covers the archive’s own work — attributions, dates, counts, recovery verdicts. It does not cover what SCV Talk published between 2006 and 2015. Those posts stand as written, errors included, because correcting them would alter the record rather than describe it.

  1. We said the 2007-2008 comments never existed. They did

    The What is missing page stated that the v2a-era pages carried no comment section, that there were no comments to recover and none were missing. That was wrong, and it understated a real loss. Every surviving v2a post page prints its own comment count in its header - one of them reads 140 comments - and no capture of that era contains the comment section itself. The discussions happened; the crawler brought back the articles without them. The page now counts those comments as missing, where they belong. Two counting faults were fixed at the same time: a claimed count was being read from month archives and category listings, which repeat a post's own line, and both the claimed and recovered columns were counted per captured file rather than per post - the same double-counting corrected on 14 August, still present in this one table.

  2. Images marked unrecoverable were never actually tried

    53 images carried a verdict of unrecoverable. Re-running the recovery found that verdict was written by a process failing on a full disk, not by any archive returning nothing - and because the old manifest recorded no reason, a disk error and a genuine absence looked identical afterwards. Every verdict was discarded and re-tested from scratch. So far more than 85% of them have been recovered. The recovery protocol now records the evidence behind every verdict, including the failures.

  3. The comment total was overstated by more than double

    The archive reported 68,690 comments. The true figure is 31,616. The larger number was an artifact of counting files rather than comments: 1,765 pages had been captured more than once, and every duplicate carried its whole comment thread again. Nothing was lost in the correction - the comments were always there once each. The count was wrong, not the archive.

Support