Pretty much everyone (i.e. by a margin that would be considered a landslide in a normal election) who works near tech agrees that electronic voting is fine as long as it produces a paper record (or a uses a paper ballot) per vote which provides a mechanism to verify the integrity of the system after the fact and makes tampering hard to scale.

There. Done.

Edit: changed wording, receipt vs record.

The use of the word receipt is problematic since people believe that it means they can take it home with them and somehow use it later. It can be a "voter verified paper trail", I suppose. But, what's better is just having paper ballots that are optically scanned. If someone needs assistive software to fill out the paper ballot, so be it.

Even with paper ballots (or a vvpt), you still need regular procedures to audit the record to ensure that the counting and down-stream handling of the counts are correct.

A receipt confirming a particular vote could be useful to those that want to pay citizens for their vote.

An electronic machine that only exists to print a paper ballot makes the most sense IMO. You get the convenience and consistency of a touch screen (no hanging chads), but you get to see the paper that gets printed out and see that it matches what you put in, and the paper is the actual system of record that matters.

Do you see a role for the electronic machine to provide a count as well? My sense is that the switch to electronic machines (and the switch to mechanical booths prior) had to have been driven at least in part to make the preliminary count easier to obtain.

Counting paper votes is a solved problem, and nations that use entirely paper ballots don't have slow preliminary counts. Australia, for example, has a somewhat complicated preferential voting system and relies entirely on paper ballots, and yet they are generally able to provide a conclusive result the night of an election (with televised tracking of counts throughout the country).

This may not make sense intuitively, but counting paper scales quite well by being distributed to the polling locations and physically sorting the paper ballots. The Australian Electoral Commission documents the counting process quite thoroughly [0].

The only time counting paper takes a long time to get a win/lose result is when results are close enough that every vote needs to be counted. This is surprisingly rare in practice.

[0]: https://www.aec.gov.au/Voting/counting/index.htm

For those of us who want to switch to alternate electoral methods like RCV, approval voting, multimember districting, etc, using machine counting is a big deal. If you are paper only, the increased labor to do repeated tabulations (as is done to an extent in RCV) is significant. It is the reason a bill to implement RCV in a state I used to live failed, after passing the state house and being discussed for some time in the state senate.

Whatever counting method you are using and whatever reporting method you are using, it needs to be verifiable. I still can't believe we had ballots from 2016 that were destroyed after brin subpoenaed, without any consequence for those who had charge of their possession and safety.

I beg to differ. I live in Minneapolis, where we do RCV on paper. It has not caused any of the problems you're talking about.

I think I'd rather just have the paper ballots counted by a separate machine/person. I know it's redundant but it's worth it to have the systems totally independent of one another (and independently upgradable). It's the IRL equivalent of sacrificing some DRYness to achieve loose coupling.

I don't feel we have a track record of responding well to elections where irregularities are suspected. Especially after a winner has been declared, and the loser has conceded.

I mean, the 2000 US presidential election had a recount... that was cancelled. The Brexit referendum had a bunch of spending irregularities... but the result stands. The 2016 US presidential election had allegations of Russian interference... but the result stands.

It's pointless having an early electronic result and a late paper result if the early result always stands - the late paper result gives you nothing except a false sense of security.

>I mean, the 2000 US presidential election had a recount... that was cancelled. The Brexit referendum had a bunch of spending irregularities... but the result stands. The 2016 US presidential election had allegations of Russian interference... but the result stands.

In none of those cases was there proof that actual vote counts were altered.

There are still ongoing investigations into Russia's interference, but the side that the interference helped is the on that's currently in power, and there's enough ambiguity for them to avoid taking any direct action. If we'd had a paper ballots that proved that Russia hacked voting machines and actually changed 100k or so votes--I'm 100% confident we'd have had a completely different outcome.

  In none of those cases was there proof
  that actual vote counts were altered.

I was thinking more of the recounts that were cancelled. If you can't recount to check if vote counts were altered without first having proof that vote counts were altered, what's the point?

More broadly, I hoped to illustrate the fact winners oppose fishing expeditions to double-check election integrity.

I simply don't believe we have the option of getting the result wrong on the first announcement. It has to be right the first time. And because it has to be right first time, we should apply our most trusted counting method the first time; the addition of an early-but-untrustworthy count is a move in the wrong direction.

>I was thinking more of the recounts that were cancelled. If you can't recount to check if vote counts were altered without first having proof that vote counts were altered, what's the point?

But you can. Many states already have automatic recount laws when vote totals are close. In many cases candidates can pay for recounts even if they aren't close.

Recounts are common in American politics, and sometimes they even change the result--Al Franken won in 2008 after a recount flipped the result.

The Florida recount would most certainly have been finished if any of the counties had found obvious cases of voting machine hacking. Every country completed at least 1 recount, and most completed manual recounts.

Sure .. As long as everybody who voted takes their receipt, and is willing to go to the government and demand proof that their vote was counted. Let's see, in the USA you get 54% voter turn-out .. a re-count requiring those receipts should get what, 15, maybe 20% of those voters to participate?

There. No.

> As long as everybody who voted takes their receipt

Voters strictly can't take a record of their receipt; this opens up a whole other level of voter intimidation, where, if you voted for the wrong person you lose your job, lose benefits, get beat up, get killed.

It's been a long time, but somewhere I voted, maybe California or Michigan, but somewhere I voted gave me a real receipt. It had a bar-code that was tied not to my identity, but to my ballot itself.

The "receipt" was a perforated piece of paper with a barcode. The votes were cast by connecting lines, and the right half of that line crossed the perforation lines, and was mine to keep. This seemed like a reasonable system, and I could then verify that my vote was counted.

As for voter intimidation, if we really believe that our democracy is worth killing and dying for in foreign wars, then it would logically follow that we must accept the risk of demanding accountability domestically.

Couldn't the receipt of votes be an encrypted result derived from the voter's key for which only the government/district has the private key ? It would not show the vote but only guarantee that your vote has not been modified. You could go to the government to get the vote confirmed privately.

Ideally, assuming you would be voting with a key linked to your identity (SIN), both the IDs of voters and their votes could be entirely public and on your receipt.

The act of counting votes would be for a set of parties to be work with only the public list and given the private key(s) to generate results. The trust there should be similar to counting paper ballots, especially if independent parties combine subsets and arrive at the same result.

Yes, that would be possible; but is significantly beyond what people typically mean when they say "just give me receipt"

> Ideally, assuming you would be voting with a key linked to your identity (SIN), both the IDs of voters and their votes could be entirely public and on your receipt.

What piece of your identity does your employer not have access to? In the U.S. they've typically seen your id, your passport, your pay stubs, your health insurance.

There's also the question of how the printed ballots get counted.

If you put them all into a black box machine (or set of machines) that spits out a total at the end, you've recreated all the same issues.

No, you haven't. The issue with touchscreens isn't the machine total, but rather the lack of a mechanism for validating the machine total. Machine-read paper ballots can be and are sometimes hand-counted to validate the machines.

Paper ballots are worthless if they're counted electronically.

They're not worthless if you mandate that they're kept around for verification.

You can automatically verify some random percentage of them and do a full recount if needed.

You can recount paper ballots on different electronic counters to verify integrity. However, you cannot vote on multiple electronic machines to ensure consistency.

Hell, you can easily recount paper ballots by hand. No need to bring any machines into it. We did this for hundreds of years, and it still works.

The ambiguities we had in the 2000 election in Florida were from people improperly marking their ballots (hanging chads, etc.). If the paper ballots are generated by machines, or at least have already been verified as being readable by a machine at the time the ballot was cast, then you will have a much lower margin of error during the manual recount.

What's the old argument? This article refers to a current legal case and testimony that happened last week, and a judge is expected to reach a decision today about whether Georgia must change its election system just weeks before voting in 2018 takes place.

The debate is more than just paper vs. digital. Georgia is interesting not just because it has gone completely paperless (4 other states have done so), but because it is also the only state in which the system is centralized. Delaware, one of the other states to have gone paperless, have systems too old to connect to the Internet: https://www.delawareonline.com/story/news/politics/2018/08/1...

While you're quite right, if you wanted to see what the old argument was, please look at your sibling comments.

Voting is one of those problems that is, on its face, very simple: write a name down, count how many people wrote each name down, person with most names wins.

This seeming simplicity, combined with the fact that no one thus far has seemingly gotten it right, lends itself strongly towards all kinds of armchair opinions from anyone at all related to any of the fields involved.

I'm a software engineer, for example. "Design and build me a system that collects votes and produces results." is in the format of the requests I get at my job on a daily/weekly/monthly basis, so I can take it and run with it, as I am trained to do. I can come up with all kinds of additional requirements in my head, tons of edge cases, different scenarios I need to watch out for, etc etc. Basically, as can anyone who does my job, I can turn it from a (immensely vague in this case) business need into a set of technical requirements.

However, I am wholly unqualified to design a ballot device, despite the seeming simplicity of the request, and how it seems to fit nicely into my professional intake. Will that stop me or anyone else in my position from offering an opinion? Probably not!

Thus, you get the constant "re"hashing of "old" arguments, because people see these problems as "simpler" than they actually are, and "re"voice the same ideas and concepts, because it seems like some obvious problems can be solved with "just a little" modification to the existing systems.