I read recently that in the primaries, and during the absentee balloting that preceded the elections, touchscreen voting machines proved to have serious problems, frequently failing to properly record and tally votes. Over the past six years or so I've read all manner of experts weigh in on this subject, and haven't yet read one who has pointed out what I think is obvious to readers of this blog (and anyone competent to read any blog): We're talking about an awesomely simple programming task. Here's what it takes:
* A table listing the candidates, the office for which they are candidates, and their party affiliations, which can be downloaded into a server located in each polling place.
* A cheap web server in each polling place that stores the candidate table and renders it into a ballot.
* Voting machines equipped with browsers, touchscreens, and printers.
* A program that runs in the server that records the results for each voter into a votes table that can be uploaded when the polls close, and that prints out a "receipt" for each voter, to verify their vote and to be used in a manual recount should one be needed.
* A dial-up modem, attached to the web server, to provide a temporary connection to the central system before the polls open (to download the setup) and when they close (to upload the results).
* A central server to accept the votes tables, consolidate them, and run a simple SQL query that tabulates the results.
* Pollwatchers alert for voters who covertly approach the site server with their own mouse and keyboard to try to hack the system without anyone noticing.
There is, to be repetitious, nothing remotely complicated about building a voting machine system. There are those who would claim that calibrating the touchscreens and maintaining the calibration is a challenge. I certainly sympathize. As anyone knows who uses a Palm PDA knows, you have to calibrate your screen ... oh, wait. You have to calibrate a Palm PDA touchscreen exactly once, when you take it out of the box for the first time. After that, it holds its calibration approximately forever.
Conspiracy theories abound about why touchscreen voting machine systems are so unreliable, most swirling around the sordid connection between the president of Diebold's voting machine division and the Republican party. I have a different theory:
The companies that sell these puppies have persuaded the agencies that buy them that the technology involved is difficult and complex, so as to raise the price beyond the US$350 per voting machine and $1,500 per site server that should be the maximum for hardware and software that is, in every respect, less complicated than the market survey systems provided commercially and inexpensively by any number of providers for use on the World Wide Web.