Vote count marred by computer woes:

Lebanon [Indiana] -- Boone County officials are searching for an answer to the computer glitch that spewed out impossible numbers and interrupted an otherwise uneventful election process Tuesday. ...A lengthy collaboration between the county's information technology director and advisers from the MicroVote software producer fixed the problem. But before that, computer readings of stored voting machine data showed far more votes than registered voters. "It was like 144,000 votes cast," said Garofolo, whose corrected accounting showed just 5,352 ballots from a pool of fewer than 19,000 registered voters.

I believe electronic voting can work. But not if the systems are privately designed by companies which operate strictly on profit-motive. There is a rich tradition of delivering buggy products in the electronics and software industry. There is a common "foot in the door" mentality: we establish our relationship with the client in such a way that it would be infeasible for them to withdraw late in the game. As long as we continue to support the product, and iron out the bugs eventually, it doesn't matter if our initial deliveries are less than perfect. This may be a great model for, say, hospital products. But it is absolutely no good for voting systems. The design of these machines and software should be totally transparent. They should be reviewed by multiple private, government, and university groups. If the design is not open and there are flaws, then it is in the interest of the private manufacturer to conceal those flaws (as with Diebold). In the case of voting machines, such knowledge can be very lucrative. A black-market for information may arise surrounding such systems.