The debate over electronic votes --- as noisy and as intellectually impoverished as it has been --- has missed the central point, and the central problem. Many of the technologies being pushed --- Diebold's touch-screens come to mind --- make the fatal error of conflating two distinct acts: the casting of votes, and the counting of the votes cast. The focus has been on counting, not casting: with accurate counting as the priority, the “new and improved” process is optimized for the convenience of the county clerk, whose (tedious) job it is to tally ballots.
This focus on counting misses the point of voting in the first place: if the first priority isn't accurately recording the voter's intent, accurate counting doesn't count for much. If we were serious about solving (or at least mitigating) the problems that have made electronic balloting famous (or infamous) of late, we might try examining it as a system --- with discrete component parts --- rather than as a function. The essential components are, of course:
Record the votes: The voter's choice is accurately recorded, in fixed and tangible form. This means that a paper trail is required: simply displaying the choice on-screen won't do.
Count the votes: The recorded ballots are tallied, and (if necessary) tallied again. For the sake of simplicity, the paper ballots might be scanned, separated, and tallied; for the sake of accuracy, it might be worthwhile to employ multiple methods.
Much has been made of the need for paper trails, but even so the processes of casting and counting, of recording and tallying, coalesce into the single process of “voting.” A paper trail actually serves two distinct purposes: it verifies, to the voter, that the vote recorded is the vote he cast (see 1, above); and it provides a means of counting tangible, physical ballots (refer to 2, supra).
[Update: an editorial in The New York Times stresses the need for a paper trail and calls for Congress to act, after yet another election debacle in Florida.] The fundamental difficulty, as noted above, is that the purchasing decisions are effectively made by the people who specify the equipment: the county clerks; and those decisions are heavily influenced by the equipment vendors, who often hire retiring clerks as “consultants” who can credibly lobby their former peers. Every interest, in short, is served save one: that of the voter.
What's wrong with this picture?