ours is closer to the real vote records for most nominees, but each board has its own blind spot (◕‿◕) i used a third, independent index that counts the actual vote records in people's repos (Constellation): ┌─────────────────┬──────────────┬──────────┬──────┐ │ nominee │ real records │ official │ ours │ ├─────────────────┼──────────────┼──────────┼──────┤ │ moara │ 1786 │ 1787 │ 1781 │ ├─────────────────┼──────────────┼──────────┼──────┤ │ jamellebouie │ 1772 │ 1768 │ 1771 │ ├─────────────────┼──────────────┼──────────┼──────┤ │ atrupar │ 1719 │ 1710 │ 1718 │ ├─────────────────┼──────────────┼──────────┼──────┤ │ dieworkwear │ 1501 │ 1482 │ 1499 │ ├─────────────────┼──────────────┼──────────┼──────┤ │ katie.bzky.team │ 1441 │ 1444 │ 1425 │ ├─────────────────┼──────────────┼──────────┼──────┤ │ mayor.nyc.gov │ 1404 │ 1390 │ 1400 │ ├─────────────────┼──────────────┼──────────┼──────┤ │ gonzo │ 518 │ 500 │ 518 │ ├─────────────────┼──────────────┼──────────┼──────┤ │ catsofyore │ 449 │ 429 │ 448 │ ├─────────────────┼──────────────┼──────────┼──────┤ │ antlervel.vet │ 563 │ 562 │ 554 │ ├─────────────────┼──────────────┼──────────┼──────┤ │ michaelhobbes │ 421 │ 418 │ 420 │ └─────────────────┴──────────────┴──────────┴──────┘ the record count slightly overstates what should count, because it includes self-votes and over-limit votes. - where bsky38.com is wrong: - for catsofyore, i checked every voter we count against their API. 21 of our votes are missing from the official index. - all 21 are real records in those voters' repos, cast sep 18–20. - the official indexer seems to have dropped votes in that window and never backfilled. that matches victoria saying their site was behind. - where we're wrong: - for katie, 16 voters have a real vote for her that we don't count. the official board counts all 16. - for each of those voters we show no ballot at all, and those katie votes were all cast sep 4–5, right at launch. - either our initial import missed them, or a repo sync event wiped them. our store deletes votes on sync events, and i suspect that path, but i haven't proven it. - antlervel (-9) and moara (-5) fit the same pattern. so we're more right about recent votes, and they're more right about some launch-day voters. near the #38 cutoff, all three sources agree within about 3 votes. action items: - say the word and i'll chase the launch-day voters before the 23:00 UTC close, or after it, since standings freeze then. - it may be worth telling the organizers their index is missing sep 18–20 votes. i can write up the evidence.