> Plenty of incentives and options even with paper ballots, as many dictators have shown.
Is that so? If anything, they ignore the ballots or fake the numbers—see Venezuela. That is another problem entirely.
> Why wouldn’t this be possible with machines?
With paper ballot countings, even the most plain citizen, can witness the counting and ensure themselves nothing shady is going on: Watching as people create heaps from ballots with a checkmark in the same box, then call out the numbers is not beatable in terms of accessibility.
With machines, complexity is orders of magnitude higher: Watching the computer is pointless. You don't get to see how it adds up votes, you have to trust its software is working as advertised. There is just no way to verify no component of the stack has been manipulated.
> If you trust the people checking the machines have not been manipulated, this is not at all what would be happening.
But that is the point! Why would you ever do that? What power does that grant to these people? How dangerous would it be to be such a person? Who even would these non-government election experts be? Why limit the number of people able to verify an election is going on truthfully to a small amount of technically literate experts, when we have a perfectly working democratic system in place?
Tell me this; how can you check the machines have not been manipulated with a 100 percent certainty? How do you verify the screen displays what it is supposed to, the software has no backdoors, the hardware has no backdoors, there are no parts swapped out, there are no second-order effects of transistors flipping bits if the ambient temperature reaches a certain point, the cables have not been messed with, the data it sends is not intercepted or altered, the packages won't be dropped, or arrive in duplicates, or any of the other myriad of possible failure conditions? An election is an extremely rewarding target for both internal and foreign/rival state actors. Just think of Stuxnet if you think this is paranoid, and that was 2010! How would you ensure that even the most sophisticated attackers won’t come up with an exploit?
You cannot. You cannot do this reliably for every machine involved in the election; the combined knowledge and experience involved in each of these questions far exceeds even most IT professionals. So you immediately remove the ability to witness the election from almost every citizen, by making the process infinitely more complex and harder to understand, for no reason at all.
I know there are lots of interesting problems to solve here, but democracy is not the right place for complex solutions to interesting problems.
There is virtually no compelling reason to drop a working process in favour of machines here.
Edit: and one more thing with elections is that you cannot just try again if something seems suspicious. You’ve completely shattered trust at that point, and the victor would rightfully claim the election had been stolen from them. Again: Democracy is no place for technological solutions.