Paper ballot state votor here. Each precinct is issued enough ballots to account for registered voters -early voters -absentee ballots + a few spares.
Prior to voting, each voter is ID’d, and checked against the precint registration rolls, and must sign a master list. They are then issued a “voting ticket” with thier precinct registration number.
This ticket is then exchanged for a ballot. Each ballot has a serial number on a numbered tear off stub. There is no serial number on the ballot itself. HOWEVER, the serrations between the stubs and the ballots are not uniform. The stubs are kept bound into the pad of ballots. (we’ll get to the reason for this later) The ballot number is recorded on the voting ticket which is kept for audit purposes.
Marked ballots are fed (by the voter, supervised by poll worker) into a counting machine which records the votes, and counts the ballots, Integral with that machine is a sealed bin that collects the paper ballots.
I see the electronic scanner as the weak link, because it represents a convienient place to alter lots of votes. However the paper ballots are available for manual recounting or audit purposes.
Given this system the following safeguards exist:
-The only way to get a paper ballot into the collection box is via the electronic scanner. The only way to get it out is via a sealed door.
-If the electronic scanner were to fail, or be suspect, all the ballots that went through that scanner are in one box, and can be manually counted. THIS is the main advantage of paper ballots, IMO. Pretty hard to lose the data.
-The number of ballots in the box must equal the number of ballots signed for by the voters. Makes it hard to stuff the ballot boxes unless the precinct workers/observers are all in cahoots.
-All ballots issued to each precinct must be accounted for…the recording machine total plus the leftovers must match the issued number. So if I stuffed in extra ballots, I’d have to get them from the stack issued to the precinct. That would create a discrepancy between the number of signatures/ tickets vs. the number of ballots.
-If a pad of ballots was short or long, the ballot stubs are available to track down the problem.
-If serious monkey buisiness were suspected, there is process where a particular voter can be linked to a particular ballot. This is all paper, and not a computer data base. It could be done by forensic comparison of the tear pattern between serrations, then searching through the ballot tickets, then looking up the precinct voter number, and finally arriving at a name. It is only practical for a small random sample, and pretty well addresses voter secrecy concerns.