Re:Clinton E. Curtis, programmer, testifies to Congress abou...
by brlewis on Thursday December 13, 2007 @11:07AM
(#89181)
Random 1% error is easily dealt with. When the election is close they do a supervised recount.
Re:Clinton E. Curtis, programmer, testifies to Congress abou...
by angryprofessor on Thursday December 13, 2007 @11:33AM
(#89197)
There are certainly no problems with supervised recounts. Look how well they go:
http://en.wikipedia.org/wiki/Washington_gubernatorial_electi...
http://en.wikipedia.org/wiki/United_States_presidential_elec...
http://en.wikipedia.org/wiki/U.S._presidential_election,_200...
I'm not saying electronic is definately the way to go. I'm just saying that paper ballots also have big problems, and that one can protect reasonably well against pure software hacks.
Better electronic systems are possible.
In NJ, we have an electronic machine which is just a bunch of buttons and lights: push a button, a light in row 1 goes on, push the commit button, and "Row a, column b" is recorded. A translucent paper ballot indicating the meaning of "Row a, columnb b" is placed on top of the lights/buttons, so the machine doesn't know which row/column corresponds to which candidate.
Using a machine like this, and physically randomizing the rows/columns machine by machine on election day (which isn't done in NJ), software fraud would be difficult.
Of course, none of this will stop fraudulent voting (which is very easy), zombie voting and other old fashioned trickery (google the washington governor's election, seattle district in particular).
I voted under a phony name in 2006, due to bad handwriting. After I discovered this, I called up my old roommate, who voted under my name (rather than his) in my old town. Vote totals were unnaffected, but only because we were honest.