Finland had communal elections a few weeks back. Some communities tested electronic voting machines, and naturally it failed miserably. Now the communities are working on having new elections. The central ministry that should be taking care of these things is silent – except for the justice minister who said the electronic voting thing was decided before her term and she opposes it. Here’s an article in English by Helsingin Sanomat.
The mechanism of the flaw was this: People would have an electronic card where the machine in the booth would write the vote, the people would take the card and drop it off to register the vote. But they pulled the card out too at a time when the vote had not yet been written on the card, and the machine did not indicate anything was wrong.
Machines and extra complexity are needless when the process is extremely simple: choose a single two or three digit number. The vote card can be human readable all the time.
A paper trail (a human readable receipt) would have helped somewhat, but still, it is conceivable that elections could be rigged by manipulating it so that the machine at random times prints different vote cards from the selection made by the user. It does not need to work all the time to still effect the outcome of the election.
Electronic voting machines are extremely hard to inspect or troubleshoot on the spot. The mechanism is always black box like and is easily tampered with. They are also very failure prone compared to paper and pencil. Never mind expensive.
People with the most experience with computers are the least trusting of them on matters like voting, where the process is so extremely simple that computers bring little added value. Also because the user base is extremely varied, it is hard to design a foolproof system, or a system as foolproof as pencil and paper.
What Are Electronic Machines Good For Then At All?
Computers are good for some very reliability critical things like hospital monitors or other equipment where they bring some added benefits with their calculations. Also often such critical systems are used by professionals and are designed for narrow uses, which increases their reliability a lot.
One application that resembles voting quite a lot is lottery. The user base is vast, and the data is extremely simple (actually, with voting, the data is even simpler). Electronic lottery has worked over here for decades, though I don’t know of any critical analysis of “lost tickets” or anything. In any case, everyone gets a human readable receipt for their submitted numbers, and they include all relevant information so that you can get your money in case the numbers match. I don’t know if there has ever been a case of a person possessing a winning receipt but the numbers not being submitted in the system because of some glitch. Technically it could be possible to forge the simple paper receipt after checking what the lottery numbers were and then claim the prize money. I don’t know if any encryption or unique id:s are included in the receipt to make this harder. You could always calculate such a check number with a key from the numbers submitted so that receipt forging would be impossible.
Electronic lottery has sped up the process hugely and reduced the costs as well. No more do the coupons have to be mailed over to one location and then read optically in a failure prone process. Veikkaus has the state controlled monopoly on lottery in Finland, and the winnings left over are used for the public good (though the inefficient Veikkaus company’s overhead is huge). Same with slot and poker machines, which are operated by RAY or Raha-automaattiyhdistys. State owned monopolies are one sensible model for activities that are in essence licenses to print money.
Nice analogy. Note that even in Finnish lottery, where much of the process is computerized, the critical part where the winning numbers are actually generated is mechanical.