Question of the opposite spirit: politics.stackexchange.com/questions/18443/…

I would just point to any of the many security breaches of online data to show that online activity isn't secure enough for voting.

@JoeW AFAIK the Estonian system is designed in a way where everything can leak to the public and you'd still not able to prove who voted how. It's the magic of modern encryption techniques.

Security isn’t just about leaking data but the information on the server being altered.

@JoeW they've also got a system in place where voters can validate that their vote was properly counted and that the voter registry wasn't change during the election.

@JoeW in any case, Estonia's been using it with great success. You could say they're a small country so its easier for them, but there are lots of other small democracies out there.

That doesn't change the fact that online security is a major concern for most of the world.

Most small countries have big aspirations. So they're more likely to emulate big countries than other small countries.

There are many other countries besides Estonia, that do not vote online and have also a flawless track record. That's not a unique point for Estonia. I guess online voting would need to to have other advantages over offline voting to be considered the better option.

It’s a bit odd to have the question mention and then immediately ignore its own answer in the form of the Russia case. An approach working well in one case and miserably in others isn’t much of a strong point.

@MisterMiyagi Russia had to be mentioned to pre-empt "actually, Russia will do it soon!" answers. I wanted to point out that its irrelevant due to Russia being an autocracy at this point.

@JonathanReez I don't see how that makes it irrelevant, quite the opposite actually (hence me wondering if an answer along these lines would match the question). An approach being useful to enable or support autocracy isn't exactly good PR for adopting it in states that should stay democracies.

The Estonian e-vote system is perfectly secure if it works exactly as intended. But that is true of any software. That the general software architecture looks safe is merely proof that they didn't deliberately screwed up, it is in no way proof that the system is actually secure. It is certainly not. It may be close enough to secure that it is sufficient for the purpose but that is very hard to judge, even from the inside.

And Estonia is a single case and small country and 20 years is a relatively short term. Maybe Estonia was just lucky to have a flawless record so far. Maybe next year their online voting system gets hacked.

What are the proposed advantages? Probably ease of access. However, the turnout (around 63% at the last parliamentary election in Estonia) is not particularly convincing and way below some other European nations.

@Muschkopp Mail-in ballot is probably similarly easy to access. And I'm sure that Estonia offers multiple ways to vote, not only online. Maybe online voting would be considerably cheaper. But it's not sure that money is really a convincing argument for a less tangible process.

Not sure if this has been an actual consideration by any real governments, but compare the slow pace (with good reason) of major changes like voting systems to the rapid evolution of cybercrime threats.

Presumably, internet voting is acceptable in Estonia because enough Estonian people trust it. I doubt that the same would be true in most countries where it might be proposed. In the US in particular, whichever party proposed internet voting, the other party would surely be against it because they don't trust their opponents to implement it (or oversee its implementation) securely and fairly. And sure, plenty of Americans don't trust their current election system either, but that is more because the GOP has told them not to trust it; the GOP would tell them not to trust internet voting, too.

@kaya3 the U.S. system is actually uniquely vulnerable compared to most EU countries. Not having to show one’s ID card (only issued to citizens, not like a drivers license in the U.S.) to vote would be considered absolutely nuts in most EU states. Estonias is better because you need to show ID to get your private key and there’s overall easier ways to check voter validity thanks to digitization. I.e. dual voting is definitely impossible there.

@JonathanReez The UK hasn't required voter ID until very recently, and only 52% of the public supported it when it was implemented. UK elections were not considered "particularly vulnerable" before 2021, because in-person voter impersonation is extraordinarily rare; the best case for an impersonator is adding one vote to the tally with a very low chance of swinging the result, and the worst (and reasonably likely case) is going to prison, so the incentives just don't make it make sense.

Likewise, I think your claim that most people in the EU would consider it "absolutely nuts" is very speculative; do you have any evidence for this? Voter ID might have popular support in a lot of countries but that doesn't imply that people would think it crazy not to require it. I would guess that more people would consider online voting "absolutely nuts", and you can talk about cryptography and proofs all you like but this won't sway a lot of ordinary people.

Besides, there is a strong bias towards the status quo in such important matters as how elections should be conducted; the public will tend to think the current system is fine (absent a strong reason to believe otherwise) and be skeptical of proposed changes. Consider for example the 2011 UK referendum on the Alternative Vote which would move to a system which nonpartisan experts unanimously agree is better than FPTP (and which works in other countries without problems), but the public rejected it by 68% to 32%.

Obligatory XKCD: xkcd.com/2030

@bta XKCD is often funny but also often wrong.

As a professional software engineer for decades with a graduate degree in Computer Science, I can tell you this: When experts in the field of information security are trying to determine the most secure way to design a system that would be a very high value target, the answer they never, ever arrive at is, "Let's put it on the Internet!"

And, on that same note, I'm currently still receiving identity protection services paid for by the U.S. federal government after a bunch of my personal information was stolen from their system storing the personal information of people with security clearances for accessing classified information. If they can't protect that reliably, do you really they can reliably protect an Internet voting system?

@reirab There are experts and "experts". "Experts" build a career by finding a few exploits and then spending years talking about them despite the issues being long resolved or even not that serious in the first place once you remove all the window dressing. The work by Springall et al. (linked in the post) is a great example: 10 years have gone by but none of the authors have talked about it ever again (I've checked). They got their paper, got their sensationalist news headline, but now refuse to admit that most of their issues have been resolved because this is a boring headline.

@JonathanReez The issue, though, is that defenders must be right 100% of the time, while attackers only need to be right once. Sure, an issue that's found will most likely get fixed before the next election, but that doesn't provide much consolation when your current election has been compromised. Most holes that attackers use do eventually get closed, but it's very often only after they've been exploited (usually because they were only discovered in the aftermath of the attack.) An important election in a large country is the kind of target that APTs will save up their zero-days for.

@reirab no the Estonian system is specifically designed in a way where such an attack would be impossible to pull off. The attackers could take down the infrastructure servers forcing a fallback to paper ballots but it’s impossible to fake the results. The explanation is inside the technical papers, I can’t summarize it in a 400 character comment. But zero days do not matter for election counting or results, at all.

@JonathanReez That sounds... exceptionally unlikely. Though I am interested to read the paper at some point when I have some more time.