Choosing electronic voting machines

Share This Post

Share on facebook
Share on linkedin
Share on twitter
Share on email

Choosing electronic voting machines

In evaluating electronic voting machines we must look at two types of error. The first is system error, which is error induced by the machine. That is to say that you vote for choice A and choice B is recorded. System error can either be accidental or deliberately programmed. Since this editor spent much of his youth as a professional programmer, we can assure you that even the most trivial programs can have mysterious and unanticipated errors.

The second is user error, which is error that occurs because the voter does not understand how the device works and indicates the wrong choice. This might be caused by the design of the system, or by the voter simply making an honest mistake. To find out the potential scale of user error, we spoke with one of the original implementers of ATMs, and asked what user error rates were observed when ATMs were implemented. He noted that this was not entirely a fair question, because when ATMs were first introduced there was an attendant at each machine to help users understand how they were used. However, the rate of error in this somewhat artificial environment was roughly fifteen percent.

System error caused by mistakes in programming can either be random, giving roughly the same number of bad votes to the various candidates (this is the best case) or skewed so that one particular candidates gets more than their fair share of votes. System error that is deliberately programmed in – the goal of most efforts to hijack elections – would be designed to give the benefit to one candidate, or one party.

With system and user errors in mind, let us look at the three kinds of electronic voting machines.

The most desirable from the voters’ perspective are those where a person marks a paper ballot, after which the ballot is put into the machine which first checks to make sure the ballot is completely legible (i.e., there are no hanging chads). If the ballot is completely readable, it records the vote (this is the electronic part), and the paper ballot is kept for later verification if needed. If the ballot is not completely readable, the ballot is corrected if possible, or marked as invalid and another ballot issued. While user error may still occur (a person marks the wrong spot by mistake), system errors are likely to be of minimal concern for two reasons. First, the technology for readers has been around since the pre-electronic era. Second, the ballot is the original, not the electronic entry, and tracking errors can be found in a straightforward manner if the ballots are manually examined. The main concern here is the design of the form, as this will determine much of the user error.

Less desirable are voting machines that operate in the manner of a cash register, with the primary vote being electronic, and the paper trail being a backup. Since the paper trail is an artifact, there is no reason to expect that the vote recorded electronically is the vote cast. But at least there is a paper trail. The least desirable voting machines are completely electronic, where there is neither an actual paper trail, nor even the pretense of a paper trail. These machines have deservedly been the center of many jokes (“18-year-old Diebold programmer elected president of United States”), and the techniques for hacking these machines have been widely discussed.

An interesting question is how to nullify votes on completely electronic voting machines where it is anticipated that the vote will be unfavorable, but where either scrutiny will be so tight as to make rigging the machines difficult, or the expertise to do so is lacking. A solution we have heard discussed is to simply remove a particular machine from the voting pool after the votes have been cast. This would need to be done subtly, disabling a few critical machines in very close races, which likely means local races, rather than presidential races.

One approach we have heard discussed is use of an electromagnetic pulse (EMP) to destroy the circuitry. An EMP will fry most transistors, though it will not affect vacuum tubes. How do you produce an EMP sufficiently high to do what is needed, and no more? While detonation of an atomic device will generation an EMP sufficient to fry every electronic device around (thus explaining why MiGs used tube-based, not transistor-based, circuitry in their avionics), something more localized is needed. Some other form of energy discharge would need to be used.

The bottom line is that an election commission interested in fair and verifiable elections will choose electronic verification and tabulation of paper ballots.

More To Explore