That difference may not be accurate, though. It's assuming there is a way of networking between the machines and that code will be portable enough to affect N number of machines. I don't think either is necessarily true.

Even for the paper side. Most paper ballets are counted by machine so they could easily be tampered with at the count or when they're printed having a similar scale.

>Most paper ballets are counted by machine

And recount-able by hand in case there is doubt.

> Most paper ballets are counted by machine so they could easily be tampered with at the count […]

Do you have a source for that statement? In Germany, ballots are counted by people making tally marks on paper.

Hmm, not off hand. I'll see what I can find. It was my understanding from the hanging chad situation in Florida years ago that, at least in the states, most are initially counted by machine.

Edit: Here you go. I couldn't find research or a study regarding it but this is probably good enough and was always my experience when I voted by paper: https://www.wired.com/2016/11/vote-counts-ballot-get-counted...

We count them twice by hand here in Sweden and we get the preliminary result just a few hours after the polls close, so I do not see why one has to use a machine.

> I do not see why one has to use a machine.

One reason is that Americans vote a lot more than Swedes. Here in California I have probably ~100 elections to vote in for each I had in Sweden.

> paper "hacking" scales as O(n)

It's quite common to drop handfuls of paper. And really, the grouping comes down to transportation issues as well; when you move ballots in boxes and trucks, 0(n) becomes less accurate.

There were numerous issues with the security, packing, and handfuls of paper in 2000.

Perhaps, but they have bulk insertion.

http://www.dailymail.co.uk/news/article-3796116/Putin-s-part...

And as you add more manpower to your attempt to defraud a paper-ballot election, you increase the number of co-conspirators. As you add co-conspirators, the likelihood of being undetected drops dramatically.

Yea but to his previous point, the auditability of a digital system could be much more sophisticated then paper

> while computer hacking scales as O(1).

Not necessarily. I'm not cryptography expert, but I suspect some kind of proof-of-work system can be used for ballots.