Is there a way to avoid software fraud?

[Onlooker]

Do the Democrats have a plan of action to ensure that the software on the voting machines is honest? After all, if the Republicans can sneak in a little code that reduces the Obama votes by, say, 50 votes per machine, that could change the outcome of the election. Also, from what I've read, it's possible to write code that vanishes after it's run. Does anyone here know anything about this? Is there any way to prevent this sort of corruption? Are there any laws or procedures in place to inspect and test the code and then to verify (via checksums, perhaps?) the inspected code is the one installed on the computers?

[The Traveler]

1. These voting systems are so insecure

and vulnerable to so many forms of manipulation I cannot vouch for them. It is not a software implementation issue ... the design/architecture of these systems makes them exploitable. Not at all like lottery systems, where fraud is difficult to purpetrate and easy to detect. Strange that states provide much higher security for lottery sales and cashing functions than for voting.

[RichardRay]

2. From a software engineering perspective there are about

a brazilian ways to do it well. From a social engineering perspective it's much harder. From a practical perspective it's scary.

The ways to do it well require way too much sophistication from everyone from the software designers to the voters. I don't think we're there yet.

[TahitiNut]

3. It's too bad they don't require "MilSpec" for voting systems.

Open source, extensive stress testing and 'edge conditions," EPROMS, etc. HighRel software CAN be done.

[ladjf]

4. Yes, there are some ways to ensure the integrity of the voting

machines. But, I haven't seen any signs that politicians on either side of the aisle are making any real efforts to incorporate adequate safeguards.

If we lose in November, the two main reasons will be (1) the bought and paid for media and (2) voting fraud.

[MarjorieG]

5. No. That's why we need a return to mechanical, static systems.

The vote is too important, and hand-counted paper ballots, while transparent, are too much work for election administrators. They're unwilling.

No demand for mechanical, doesn't seem cutting edge, but where the vote is important, I wish we'd stop playing with systems that can be stolen by plugging in the wall.

[RichardRay]

6. Paper is no more of less intrinsically secure than digital.

More people can successfully alter a paper ballot than can alter a digital one, although one person who can alter a digital one can easily alter a lot more of them. OTOH, one person who can detect an altered digital ballot can detect a whole lot more of them than one person trying to detect altered paper ones.

My idea?

Vote on paper and electronically in one pass. Sample both counts and compare. If the samples match then take the electronic count - cheaper, faster, more accurate, more flexible for alternatives like 'instant runoff' voting. If the sample don't match then start counting the paper. That also provides proof at the end as to how the two votes failed to match and whether there was malicious activity or just mistakes.