[jasonlotito]

For starters, cost. Now you have to buy two machines for every single machine you buy. It also means taking up more room. Now, instead of fitting 10 machines in a single space, you can only fit 5, creating even longer lines. This also increase the chances of failures. A single machine failing means both machines in the pair fail. So, you've gone from 10 machines to 5 pairs of machines for twice the cost. You guy from a single machine putting you down to 90% original capacity (assuming 10 machines) where as a single machine brings you down to 80% capacity (40% of original capacity).

Not to mention the additional level of confusion and potential for errors. All of this so you can buy two machines from two different vendors who might, in the end, still have ties via investors.

[agumonkey]

All good points, maybe there could be one testing machine per area or zipcode instead of one per room. People wouldn't know which voting booth would be tested. If the test pass potential false positive are still there, but you at least a quantifiable bit of error catching and restricted scoping. Cost still up for sure, but maybe people wouldn't mind paying a bit more instead of 100% for trust in the system. Would probably even motivate some to vote.

[hvidgaard]

I don't think we can prevent fraud, the best we can do is source voting machines according to a spec from multiple vendors, and source software from multiple vendors as well, and do independent inspections of both by multiple 3rd parties. It would be a significant undertaking to compromise everything.

With a paper vote, you place trust in those who collect, store and, count, and you can always recount if needed. With electronic voting you place that trust in those that produce the hardware and software.

[agumonkey]

Trust it is. But the system has to try to reduce it to a minimum.