Re:ElectionGuard – A free open-source voting SDK
by mises on Monday May 06, 2019 @01:22PM
(#19841626)
"It is very important that you use a No. 2 pencil. It is very important that you fill in the entire circle darkly and completely. If you change your response, erase as completely as possible. Incomplete marks or erasures may affect your score."
And here I was, thinking I was done with scan-trons when I finished high school.
But seriously, couldn't you modify a scan-tron's firmware to falsely report results? It's much harder to tamper with every person in an election across wide geographical areas.
Re:ElectionGuard – A free open-source voting SDK
by 24gttghh on Monday May 06, 2019 @01:25PM
(#19841668)
A hand recount of the physical ballots would hopefully expose the correct count if a modified scan-tron was miscounting.
Re:ElectionGuard – A free open-source voting SDK
by mises on Monday May 06, 2019 @01:43PM
(#19841851)
...which takes us back to hand-counting. In that case, what is the purpose of scan-trons in the first place? They may be easier for machines to grade, but are harder for humans to grade (harder to follow question number to line). So we will lose time gained with scan-trons to slower human re-counts.
Re:ElectionGuard – A free open-source voting SDK
by northernjames on Monday May 06, 2019 @02:18PM
(#19842203)
When margins are wide, you could count 10% of the ballots by hand as a sanity check on what the machines are reporting.
When margins are close, you could count them all by hand.
Pretty sure we could develop ballots that are easy for both humans and scanners to read. By using computer vision rather than whatever they had in the 1980s I'm pretty sure we could find a solution here.
Even so, given the value of our democracy, I'd happily support returning to counting all ballots by hand every election.
If automation is needed, use scantron technology (where you fill in bubbles on actual paper), which can be manually audited if necessary.
No firewalls, phishing, spectres or meltdowns.