Electronic voting. Yea, even internet voting. Really shouldn't be impossible.
Tom Scott says this is a terrible idea, but I don't think it's so unsolvable. The ways to cheat are:
- stuffing the ballot box with bogus votes
- counting or recording the votes bogusly
- voting more than once or voting for someone else
- voting more than once or voting for someone else
Voter confidentiality must be preserved.
Here's my solution.
- every voter must authenticate with some non-government system that 1) ensures user ID uniqueness 2) contains a method for contacting the voter (can be a form obscuring contact details) 3) creates a random code which is not retained by the system. This is easily done by Google, Facebook, or any tiny NGO. They would need to register and be subject to audit.
- when a user votes, the data is logged in two public registers. 1) a voters register showing the person's user ID (or a unique variant from the authenticator) 2) a vote register showing the random code and how they voted.
With this system, anyone can
- re-tally the election
- check for duplicate voters
- check for bogus (e.g. dead) voters
- check that their vote was counted accurately (or anyone for whom they have the random code)
- perform a retrospective study, selecting random people and asking them to indicate whether their vote was correct or not.
Note: The random code doesn't need to be unique per se, since the risk of overlap isn't such a big deal, but it should be complex enough that this is a statistical unlikelihood.
Some kinks would still need to be worked out, but I think with these elements this is doable:
- public register of voters (already the case more or less with people seen at voting places)
- public register of votes (not already the case)
- separate authentication (could be done in the same way as voter registration. Each voter signs up with their authenticator and then registers in the government system with ID and SS# indicating which authenticator they're using and unique ID.)
- polling booths can still exist just with the same electronic systems there.
- check for duplicate voters
- check for bogus (e.g. dead) voters
- check that their vote was counted accurately (or anyone for whom they have the random code)
- perform a retrospective study, selecting random people and asking them to indicate whether their vote was correct or not.
Note: The random code doesn't need to be unique per se, since the risk of overlap isn't such a big deal, but it should be complex enough that this is a statistical unlikelihood.
Some kinks would still need to be worked out, but I think with these elements this is doable:
- public register of voters (already the case more or less with people seen at voting places)
- public register of votes (not already the case)
- separate authentication (could be done in the same way as voter registration. Each voter signs up with their authenticator and then registers in the government system with ID and SS# indicating which authenticator they're using and unique ID.)
- polling booths can still exist just with the same electronic systems there.
Comments