Thread by @rad_atl

1/ The argument seems to be that untrained, unmotivated, possibly stressed-out, tired, mildly cognitively impaired (if over 50) voters can be encouraged to verify machine-marked paper ballots effectively enough to make them useful and trusted paper trails for Risk-Limiting Audits

2/If a fraction of machine-cast votes are unverified, then RLAs become a meaningless exercise bc an attacker of a close election can always arrange for a complete hand-recount which necessarily sweeps these poisoned ballots into the final audited tally.

3/So let’s test this with a similar task, say proof-reading, another error-detection skill involving searching for errors in a document. I am willing to admit that I am at present a terrible proof-reader. But I believe that with training and experience I can improve.

4/I write books so I’m motivated to improve. How much better can I get? Here’s what the British Proofreading Standards organization says: “A good copyeditor picks up 80% of errors; a good proofreader picks up 80% of what's left. “
sfep.org.uk/standards/stan…

5/Yet, the proposition is that untrained average voters can be encouraged, without training, to improve their error-detecting skills to the levels needed for effective ballot verification, certainly above 80% (bc attacks are easy to mount if 20% of ballots are poisoned)

6/ In other words by some cognitive trick & without being subjected to rigor pedagogy the average human brain can improve error detection from the currently observed levels of 25% or less to what would be professional levels by simply being encouraged to do it.

7/How would this miracle occur? Not by Anders Ericksson’s 10,000 hours of deliberate practice (because that’s ridiculous)
projects.ict.usc.edu/itw/gel/Ericss…

8/And not through Bloom’s Mastery Learning which moves everyone’s skills by two standard deviations (bc that’s prohibitively expensive)
web.mit.edu/5.95/www/readi…

9/But it would have to involve the dopamine-fueled reward-stimulus cycle that rewires neurons and that’s a repetitive process involving redirection of attention and repetion (both anathema in the polling booth) princeton.edu/~ndaw/ds08.pdf

10/And so I am mystified by how voters who are currently not good at all at verifying actual ballots under realistic conditions become experts in large enough numbers to matter.

11/I’m also concerned that #BallotMarkingDevices whose security depends on voter error detection rates exceeding professional proof readers are being used in public elections before this question is answered.

12/ #unroll @threadreaderapp