Which tells me that with voting and democracy we don't quite have the same rationality and that there is something else going on (celebration, history, rituals etc.).
Remember that election results should be verifiable for everyone. "Pen on paper in a ballot" is very easy to reason about compared to "huge software stack somewhere running complicated cryptography". For elections, "easy to verify" is a top priority feature or we can stop having elections in the first place. Not to mention that the theory behind such systems is still in its infancy and many details are not worked out yet.
For an electronic version of this you could do something similar: Allowing all parties to audit the voting systems insuring that traces of who voted what were destroyed and that votes were counted probably.
The idea that paper ballots are somehow ungameable or easily verified is a myth. Our history is littered with examples of elections being stolen in the paper ballots era. Many even believe Bush v Gore was such a scenario.
Or this: https://youtu.be/PKQEQ7qHvgM
(I fraudulently voted as someone else a long time back and he voted as me, as part of a bet on whether this was possible. Nothing stopped us, there was no ID check, no security, no one found out.)
No one has ever shown it's a problem, people have only claimed or insinuated that it's a problem from a place of "truthiness"; i.e. they claim to know intuitively "from the gut" or because it "feels right" that in-person voter fraud is a problem.
That is essentially the system we currently have for electoral fraud.
I.e., if p(evidence|presence) ~= 0, absence of evidence is not evidence of absence.
Of course, it is pretty strange how so many people are strongly against the collection of evidence.
Just for the record, I'm not a US citizen and don't live in the US and have therefore obviously never voted there. I can only talk about how we do it here in Germany.
You can ensure that no dead or non-citizens vote by having a well-kept population registry cross-referenced with other databases. If many public officials work with the data chances of fraud in the registry is extremely low. In any case, you would have the same kind of problem with any electronic system.
We tick off any voter from the list in the moment they throw their ballot into the box. After the station closed, the first thing we do is to count the ballots and cross-reference that with the number of ticks. We also count the voter cards (which gets mailed to every voter) and have a tally sheet of people voting without that card (by showing their ID). I'm not exactly sure how I would "lose" some votes. These numbers are phoned to the election office and written down in the record and signed by everyone working there. After the election is over I verify that the votes we counted correspond to the published number for that station.
Any questionable or invalid ballot needs be inspected by anyone working at the polling station and everyone needs to agree on how to proceed with that ballot. Those ballots are marked and the decision is recorded for every single one individually. We then package them up into a special sealed envelope. These questionable votes are verified by other people on a random basis and always if the election is a close call or any signs of fraud come up.
You could probably vote for someone else around here if you can get access to their voter's card as we usually don't check ID. We can see gender and birth date of the voter so they should roughly match. You also can't do that more than once or twice in the station or we start to remember your face. Additionally, you should hope that they haven't already voted themselves and haven't registered for absentee voting. It would be a logistical nightmare to effectively change the outcome of the election by that way and the risk is high.
The biggest risk I see is absentee voting. For example, there is evidence of people "helping" elderly to fill out "their" vote. But again, if you allow people to electronically vote from their home the same kind of fraud is possible.