The debacle caused by audit-proof electronic voting machines has yet to be resolved in
North Carolina, but one thing is clear:
The paperless touchscreen voting machines do not work.
Their task is relatively simple; count votes. They may be able to perform this task, but it can never be proven. Indeed if the standards of proof required for using the F-word (not fuck) were applied to proving election results correct, these machines would fail catastrophically. The reason? It can simply not be shown that the votes reported match the votes cast. It is basically a faith-based election technology which asks us to trust the skills and scruples of a handful of programmers, and their corporate paymasters.
Please join me below the fold to find out why paperless EVMs (electronic voting machines) are worthless.
I feel like the NYT editorial linked above provides an opportunity for me to collect and clarify some points about EVM technology, since there is a tendency for many voting rights advocates (like ACLU of CA) to see the obvious benefits of touch screen voting, while overlooking or misunderstanding some fundamental flaws. The benefits include increased ease and speed of voting, decreased spoilage, very fast counting, language and ballot layout flexibility, and enhanced accessibility for some impaired voters, among others.
"But wait", you say, "I thought EVMs had been vindicated. The vendors and MSM have all pointed to recent elections as proof that their systems work." Unfortunately lack of proof of problems is NOT the same as proof of lack of problems. Indeed, it is in the lack of proof area which is the heart of this technology's fatal flaw.
The flaw is simple. The voter has no way to know that her vote was recorded correctly. Indeed noone can prove that the votes were tallied correctly. It is possible to run a test by entering a known set of votes, and see if they come out as expected, but this guarantees nothing. The machine could work differently on election day intentionally, or through bugs start counting backward once it reaches 32767, like in FL, or stop counting at 3000 like in NC. These 2 bugs might not be caught by testing (in fact they weren't), and there is no way to know what other, more obscure problems might await.
The solution is also relatively simple, but requires political will and true respect for democracy, which is alas, a rather rare combination these days.
The solution is voter-verified touchscreen-generated paper ballots. This mouthful has several important facets. The most critical is that the voting machine is understood as a ballot printer, not a collector of digital ballots. This is an important legal distinction. It is similarly essential that the ballot is verified by the voter, assuring that the touchscreen input translated correctly into the intended vote.
Such a system may very well look like this:
- Voter enters choices on a touchscreen terminal.
- The printer creates the official legal ballot, displayed behind glass, the voter voter verifies the ballot is correct, or rejects it and tries again.
- The accepted ballot drops into the bin and is electronically tallied.
- The machine's initial canvass can be immediately reported, without manual inspection of the ballots that are sealed in the hopper and protected as any other ballots should be.
- The machines are all secured, while random fraction of machines undergo a rigorous manual audit of their electronic tallies and their ballots, to verify the system. The tallies should match exactly. A larger fraction of the machines should undergo an electronic audit, where another machine reads the ballots and matches the EVM's results.
Now to pre-emptively rebut some arguments against this system:
1. Adding paper is too complex/unreliable, it slows voting down too much, it costs oo much
It does cost more to have a printer, but the cost is not prohibitive. Reliability should match those in ATMs and cash registers, which see some heavy action. If there is not enough time to vote, the problem may better be solved by extending the voting period beyond 12 hours, or simply supplying enough machines so that even the poor can vote.
If this is still too inconvenient or expensive, then I suggest that we simply cannot afford democracy and should consider some of the cheaper/easier forms of governance.
2. It negates the access benefits of touchscreens for impaired voters
While some voters may be able to manage the touchscreen but not be able to verify their ballot, that is no reason for noone to be able to verify their ballot. The fact that most ballots will be verified means that the benefits of cheating with these machines are reduced substantially anyway. Other schemes to help the small number of special needs voters can also be implemented that are not revlevant to the integrity of the vote for the overwhelming majority of voters.
3. But <insert previous technology> had a much higher rate of <insert problem here> than the paperless touchscreens in <insert state/election here>.
Other voting technologies' problems are irrelevant. When we have no idea whether the count is accurate from auditproof machines, we simply can't say what their error rates were. Moreover, computers make things faster and easier, and this is particularly true with tampering. With paper-based technologies or lever machines, one would have to tamper with large numbers of machines to really impact the results. With EVMs, only a tiny code change could tamper with ALL the machines.
With increased power comes increased responsibility, just like in Spiderman. Trading the integrity of the process for the illusion of improvement is NOT a good deal.
Have I missed some counter arguments? Am i a lunatic and/or an idiot? I wanna know, so in the words of resident Bush, Bring it on!