,
6 tweets,
3 min read
Read on Twitter
1/Building on my thread threadreaderapp.com/thread/1167416… this week as a warm up to a paper that shows voter verification of BMD output is a fantasy that has no basis in theory or practice. Voter verification depends on assumptions that are not true. But here's one that I had overlooked:
2/In addition to cognitive limitations of memory, the very idea of voter verification takes into account neither "change blindness" nor "choice blindness." As Petter Johansson and Lars Hall explain in their paper pdfs.semanticscholar.org/559c/ea673416a… 
3/The idea that significant numbers of voters will and can check the possibly fraudulent output of compromised #BallotMarkingDevices and detect the errors has no merit at all. Everywhere I look there are counterexamples and theoretical hurdles.
4/ GA election officials say "Well, voters have to accept SOME responsibility for checking their ballots," as if that justified the conclusion that BMDs are safe to use.
5/ True, voters should check their own work for errors, but verification of BMD output also requires them to detect fraud and that us an impossible task for most voters. This is a new kind of test, impossible to pass, aimed at suppressing votes.
