Freegle / Iznik / 23783

72%

push

circleci

fix(stats): exclude rippled-in copies from per-group activity stats

The daily stats generator counted rippling-out copies
(messages_groups.rippled_in = 1, collection 'Approved') as native activity.
Because the dashboard SUMs the per-group rows for the systemwide figure, one
post was counted once per group it reached (avg fan-out ~7 in production),
inflating ApprovedMessageCount / Replies / Outcomes / Weight and the derived
Activity. On the heavy-rippling days of late June 2026 ~40-50% of the daily
national Activity total was duplicate rows, not genuine new activity (verified
against live prod: e.g. Replies on 2026-06-27 = 2888 = 1358 distinct + 1530
rippled join-rows). Native distinct posting/replying was flat over the same
window, so the apparent surge was a measurement artifact of the rollout.

Add rippled_in = 0 to every messages_groups-joining count/breakdown in
StatsGenerationService (Outcomes, ApprovedMessageCount, Feedback, both
Breakdowns, Replies, Weight) and the regenerateWeightForRange subquery. Same
class of fix already applied to IP-abuse (4b6d7b3c3) and the ModTools edit
queue. Per-group and systemwide figures now reflect native posts only.

A post-deploy one-off backfill is needed to correct the historically inflated
rows (documented on GenerateDailyStatsCommand):
  php artisan stats:generate-daily --from=2026-05-14 --to=<yesterday>

V1 iznik-server Stats.php twin intentionally left untouched (retired).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013zGET5BRtdWbRc2nsf1Lss

11303 of 15029 branches covered (75.21%)

Branch coverage included in aggregate %.

10 of 10 new or added lines in 1 file covered. (100.0%)

124009 of 174466 relevant lines covered (71.08%)

37.28 hits per line