Thirty-five years ago or so Fortune Magazine had a cover story about the "First Computer Fraud." An insurance company in California (Equity Funding, I think) was accused of computer fraud. What was really happening was several of the executives of the company were having weekend meetings in which they would manually create false applications for insurance and then shepherd them through the policy issue process and then steal commission advances. The computer was simply an innocent bystander. I think a similar thing has happened with voting machines. The makers of such machines have done such terrible work, perhaps even criminal work, that they have given the machines a bad name -- when the actual problem is people.
So if we were to design a voting machine now to have ready for the 2008 elections what features would it need?
(1) It needs to be hack-proof.
(2) It needs to provide a paper trail.
(3) It needs to be affordable.
(4) It needs to be dependable.
But those are easy things to say. What is meant by hack-proof and how would we make it so?
How would the paper trail work? Should the voter be handed a copy of his electronic ballot on paper? What would he do with it? Take it home, or put it in an audit ballot box?
How will the ballots be collected for tabulation, both electronic and paper?
How can we get the major computer and network vendors to work together to build these machines as a public service?
How should the voting machine work from the voter's viewpoint?
On voting day, how will we know through the day whether a particular machine is working properly?
Should we periodically be able to check a paper ballot against has been recorded electronically, and against what has been transmitted to the tabulation center?