The twitter thread is referenced in this excellent article by the same journalist, Alan MacLeod, which I posted here yesterday, but I think it deserves its own post.

Thread reader: https://threadreaderapp.com/thread/1818050593468072023.html

If you don't feel like clicking a link, here are the tweets transcribed, with some links added:
        • iie [they/them, he/him]@hexbear.netOP
          link
          fedilink
          English
          arrow-up
          3
          ·
          edit-2
          3 months ago

          To talk about this argument we need to know where the numbers are from.

          This author appears to have gotten them from a screenshot someone sent him

          P.S. I haven’t checked any of these numbers myself. I’m assuming the screenshots that Zambrano sent are the real thing. If he’s scamming me here I’m gonna be really angry!

          We need to see that screenshot and investigate where its numbers are from

          In the meantime, there is a possible benign explanation, but it’s only speculation:

          P.P.S. Commenter Ryan points out that you could also explain this data pattern as a result of sloppy post-processing, if votes were counted correctly, then reported to the nearest percentage point, and then some intermediary mistakenly multiplied the (rounded) percentages by the total vote and reported that. I have no idea; you’d want to know where those particular numbers were coming from.

          I’m suspicious because I haven’t seen many people talking about this. You’d think there would be a feeding frenzy… or maybe there is and I’m just out of the loop