13 ms·
At some point the data hits a computer system, and that will be a vulnerability. You can always go back the the scans, but what if the scanners themselves are c
by ineedasername 7y ago
At some point the data hits a computer system, and that will be a vulnerability. You can always go back the the scans, but what if the scanners themselves are compromised? You could require a certain sample size to be manually counted and cross-referenced with the scans to be increase confidence, but I'm not sure there's a clear sweet spot, or maybe it's that I think your sweet spot needs some elaboration: Filled out on paper & scanned, sure, but then every digital link from scanning to tabulation and data storage should be designed to specs similar a similar level of security used in other very high security computing environments.
- gruez 7y ago>You can always go back the the scans, but what if the scanners themselves are compromised? [...] I think what the parent poster meant by "scan" was "tabulate". ie. you fill out a paper ballot, you put it into a scanner which scans it and records the result, and the ballot itself drops into a ballot box. It's still possible to hack the scanner to give a fake result, but it's easily auditable/detected since you still have the original ballots. You can even have the tabulators be entirely offline (since all they do is spit out a number), and their results can be easily cross-checked by random sampling.
- smitty1e 7y agoJust so. There can always be tiny nits in the numbers, but the delta between the pollbook, the unused ballots, and the counts will be vanishingly small. With enough people involved, and small enough pollbooks (local county maxes out ~4k per polling place) there is a good tradeoff between: - maximizing participation - keeping the integrity of the process - minimizing fraud As with encrypting data, there really isn't any way to make the process completely bullet proof; the task is to make real fraud infeasible.