Give everyone a piece of paper with a list of candidates' names on it. Person circles (or places 'X' next to, or something equally obvious) name of preferred candidate. Drops ballot in a box on the way out. Ballots in box are tabulated by people, using their eyes, at the end of the day, in some fair way (i.e. by more than one person, representing more than one party, etc).
From everything I have heard about all our other voting technologies, I don't believe there are any valid complaints about or improvements to be made upon such a system. Efficiency? Simplicity? This has it. Look at all the procedures that this guy describes, by contrast; cards and sleeves, papers and books, people putting cards on a piano... I mean WTF is the point. And if you look at 2000 when suddenly everyone decided that by default (unless the Supreme Court "steals the election") it's ultra important for all ballots to be hand counted ANYWAY, *in addition to* the machine tallies, and you start to realize, what the hell is the point of the initial machine tallies? And what about cost, you're telling me Diebold isn't thrilled to have this contract? How on earth could it be a cost savings over papers n pencils?
The only real advantage seems to be a savings in time, since all other things being equal computers can count faster than humans. But 1. that time savings is partially eaten away by all the extra procedures and 2. so what? we can't wait a day or two to learn the outcome?
This method has worked before, but it really is prone to fraud. Example: forge some ballots (or steal them before the election). Someone distracts the judges for about half a second, and whlie they're looking the other way, you slip an extra hundred votes for your guy in the box.
We really can do better. It's just that the current method isn't enough better.