Idea: How to design a nearly perfect electronic voting system

by Alan Cohen

This idea’s been bouncing around in my head since not long after the 2000 Florida election fiasco. A simple solution to allow fair, non-manipulable vote counting, easy to use, and easy to audit.

The problem: Existing voting systems have problems, as witnessed in Florida. All paper systems have problems with unclear choices, legibility, and so forth. Most existing or proposed electronic systems (at least those I know of) raise fears of tampering by those who design or program the machines. Also, it takes a long time for votes to get counted, there are often discrepancies and a need for recounts, which can be laborious, or, in the case of machines, impossible…

The idea: Use two different electronic machines made by different manufacturers, and print the results to create a paper trail. The voter goes to Machine 1, chooses her candidates and issues, sees a summary screen telling her who she’s chosen, prints two copies of her vote out (including an identifying bar code), deposits one in a ballot box, and takes the other home with her to keep or destroy as she sees fit. The votes in the ballot box then get scanned into a second computer, and the two tallies from the two computers are compared at the end of the day (neither machine should be connected to the internet or any centralized network). They should match exactly; if they don’t, it should be straightforward to identify which ballots are causing the discrepancy and find the paper copy. The results can be posted online by barcode number (no individually identifying info), allowing the voter to check and make sure her vote got properly counted. If it didn’t, she can show up in person with her hard copy.

Some additional specifics: There are many variations on this idea that could work, and the key elements are above, but many additional elements can be added to improve functioning. The base assumption is that the system must work easily and reliably for a little old lady who’s a bit hard of hearing, doesn’t have the best eyesight, and may not know anything about computers. These are ideas to make sure that can happen, and to facilitate overall functioning.

(1)  Include pictures of the candidates on the touchscreen system.

(2)  Have the system check after each element of the vote: “You have chosen Barack Obama, Democrat, as your choice for President of the United States. Is this correct?”

(3)  Have the system show the full ballot at the end before it’s printed, and allow voters to go back and make modifications.

(4)  Have a practice system where poll workers can show voters how to use the system. In order to avoid biases introduced by poll workers suggesting certain parties or candidates, have the practice system use hypothetical candidates and parties, such as the excellent animal parties in the blog videos of CGP Grey.

(5)  Keep the paper ballots numbered and ordered in ways that correspond to the way they are scanned in, in small bundles. Thus, if there is ever a discrepancy between the two computers, the original paper ballots can be easily located.

Why it works: The primary problems with paper voting are the possibility of unclear choices (think: hanging chads), confusion of voters, and lengthy and laborious recounts. This system solves them in the same way any electronic system would (as long as care is taken to make the electronic interface very simple to use). The primary problem with electronic voting is the lack of a paper trail, and thus the inability to do recounts and the risk of rigging. This system solves that by having printouts and an additional scanning step. But more than this, by simply and easily creating four independent records of each voting event, the system facilitates recounts and analysis of discrepancies. As my friend John Tillinghast mentioned to me, naval vessels used to carry three chronometers below deck so that if there were ever a discrepancy, the one discrepant chronometer could be identified: it is thus essential to have more than two sets of records.

As an additional benefit of this system, the posting of votes online allows instant auditing by anyone, allows for analysis of voting data by political scientists (for example, ticket splitting, which cannot be fully deduced from the totals for each candidate), and, most importantly, allows individual voters to confirm that their vote has been counted, increasing confidence in the system and a sense of ownership in it.