Online voting is conceptually (nearly) the same as mail-in voting. Some countries reject the latter.
Perhaps more discouraging to basic privacy is the fact that remote
voting systems (both paper and electronic) inherently allow voters
to eschew confidentiality. Because remote systems enable voters to
fill out their ballots outside a controlled environment, anyone can
watch over the voter’s shoulder while she fills out her ballot. [...]
The Estonian system
uses public key cryptography to provide a digital analog of the
“double envelope” ballots often used for absentee voting.
Voters can verify [to some degree] that their votes were correctly received using a smartphone app, but
the tallying process is only protected by procedural controls .
The voting system does not provide evidence of a correct tally, nor
does it provide evidence that the vote was correctly recorded if
the client [app] is dishonest. [...]
If voters
make their selections on their own devices, then there is an even
greater risk that these devices could be infected with malware that
records (and perhaps even alters) their selections (see, for instance,
the Estonian system [97]). [...]
Denial-of-Service (DoS) is an ever-present threat to elections
which can be mitigated but never fully eliminated. A simple service
outage can disenfranchise voters, and the threat of attack from
foreign state-level adversaries is a pressing concern. Indeed, one of
the countries that regularly uses Internet voting, Estonia, has been
subject to malicious outages. [104]
So, a somewhat more secure channel than paper mail (due to PKI etc.) if you discount the endpoint attack possibility, which for a [big] state-level actor is not negligible. So even if "large portions of the online voting system were published as open-source for the sake of transparency" that doesn't help much as long as the underlying phone OS code is whatever the average Joe spent their money on.
As for the DoS aspect, the incident cited (as [104]) was not specifically about that voting system, AFAICT but a broader attack on gov't infrastructure.
BTW, the US [army] and one US state did experiment with something not so dissimilar -- "Voatz". (It's not covered in that 2017 paper though that I got that summary table from, because it happened mostly later. Also, code transparency wasn't exactly a selling point of Voatz. But even they had done that, that phone app used like 20 libraries that one apparently could not easily audit etc.)
Russia plans to implement it nationwide by 2026, although critics argue the primary motivation is electoral fraud rather than an effort to increase voter participation.
TBH even if Russia implemented the pinnacle of research in this area like including end-to-end (vote) verification etc., it doesn't make much of a difference as long as they control who can register as a candidate. Same thing if Iran used something like this for their elections.
Given Putin's regime track record for harassing opposition supporters, my guess is that another, if not the main reason to move to a system like this is to find out who exactly is voting for any opposition, or not at least not voting for him.
(Anyhow, from the WaPo story you linked to, the Russia system was more like Voatz, with no external auditing allowed for their server-side blockchain. Decreasing transparency was apparently a general goal as they no longer published per-precinct results either for the e-votes.)
BTW, in Estonia, it's the Russian minority there who (say they) trust I-voting less:
the best predictor of online voting was
trust in i-voting. On average, trust in i-voting in Estonia is about as high as in
Switzerland: on a scale of 0–10, over 60 percent of Estonian voters pick values
between 6 and 10. However, 17 percent of voters do not trust i-voting at all (a
value of 0), indicating a high level
of polarization. But what predicts
trust in i-voting? In the Estonian
context, we would expect ethnic
Russians, who make up about 30
percent of the population and about
20 percent of the eligible voters, to
trust i-voting less than ethnic Estonians, in part because the voting
application was at first only available in Estonian. The initial lack
of a Russian-language option gave
rise to suspicions that the Estonian
government had found another way
to disenfranchise ethnic Russians, many of whom still do not have Estonian
citizenship and resent Estonian language laws, which make Estonian the only
official language. This is precisely what we find: even controlling for higher
education and acknowledging the feasibility of verifying one’s vote, ethnic Russians are significantly less trustful of i-voting than ethnic Estonians.
Aside, Estonia at one point implemented cast-as-intended verification by allowing the user to check the vote was sent properly using a second device. (This isn't E2E-V by the way, but does at least address some endpoint concerns.) This feature is only briefly mentioned at the end of the current EE gov I-voting FAQ.
Estonia also uses a fairly centralized ID card.
Most importantly,
the government required all citizens to hold a digital ID card, which replaced
a plethora of other ID cards, from bank cards to health insurance cards, and
can be used to authenticate a person’s identity and to sign documents online.
As a result of these policies, 79 percent of Estonians are frequent users of the
internet, more than in any other East European country. In this context, i-voting seemed like a natural extension of electronic government.
In the US, many would balk at such a centralization of services, and some even at such forms of voter ID. This is cultural/political in part. E.g. the UK gov't scrapped a national ID, after an alternation in power.
One mitigating (or enabling) factor for Estonia is that they use a PR system, so a small boost on one side of the political spectrum (that i-votes give) isn't too concerning. However, in other (Eastern) European countries this isn't the case, i.e. the system is first-past-the-post or mixed (e.g. in Lithuania it's mixed, for instance). Thus in some of these other countries a small change in voting difference can result in a large[r] change in seat outcomes. So, there's more opposition to a (voting method) change that could possibly enable such an event.
E.g. Lithuania's PM emphasized in 2024 that
E-voting cannot be introduced in Lithuania until 100 percent security is guaranteed, Prime Minister Ingrida Šimonytė has said following a feasibility study.
“The main question is security, and as long as it’s impossible to answer that security question 100 percent, then I think experiments in this area would be simply dangerous for democracy,” she told reporters at the Seimas on Thursday.
“I believe this is a question for future generations, not for future governments,” Šimonytė said, adding that any e-voting system must ensure both data security and voting secrecy.
“To be honest, I’m not the best expert on IT security, but I have not seen any experts who assure me that it is possible to meet both requirements simultaneously. That is probably the main problem,” she noted.
Given that all democracies are now targeted by regimes that use information systems to wreak havoc, as well as steal or falsify data, the issue of security in the current geopolitical context is crucial, Šimonytė pointed out.
It's also mentioned in a EU round-up that
Norway meanwhile trialled internet voting in both 2011 and 2013 but decided against continuing because of public perception about the security of the vote.
Also in Finland
Internet voting was trialled in 2008.
A review in 2016-17 concluded that
the risks associated with internet
voting outweighed the limited
benefits that it brought.
And France introduced (around 2012) and maintains remote electronic voting for expats only, as an alternative to mail-in voting (which is also for expats only), but this has been controversial enough that it's limited to some elections like legislative elections (and not presidential ones, for instance). Interestingly however, at least based on opinion poll (that I've not checked out well-conducted it was) a majority of the French reportedly want such a system, despite its issues
The experience of many French citizens [...] exposes some of the challenges
of e-voting. For instance, authorities struggle to find a balance between cybersecurity,
including voter identity, and the simplicity of the process. As explained by many French
citizens residing abroad that tried to vote online for the June 2022 election, a system error
made it impossible for them to cast their vote, which generated frustration. Once of the main
problems was that voters that had registered with a Yahoo email address never received the
username needed to sign into the voting platform.
Beyond the [2022] elections, some French citizens have persistent concerns about e-voting more
generally. It is worth noting that in a post-election poll done by the People 2022 project (led by
the ESPOL Lille), 60% of respondents were in favour of introducing e-voting in the presidential
elections, while 30.7% were against it. These results coincide almost exactly with respondents
who claim that they would use e-voting to cast their vote in a presidential election (60.4%
would, while 30.9% would not).
Even though a majority of respondents declared they are in favour of and would use e-voting,
amongst those who wouldn’t the main reason is security (60.7%), followed at a very
considerable distance by the lack of previous experience with the system (20.7%). For those
who would, the most cited reason is that it would take less time than voting in a polling station,
closely followed by their daily use of the internet (45.9%). France has tried to mitigate security
concerns by increasing transparency around the testing methods used to validate and select
e-voting systems.
One practical issue that is discussed in the sequent is that France, unlike Estonia lacks a digital ID card (and various email/SMS ad-hoc solutions being rather problematic), however they are taking steps in that (online) direction with "France Identité".
Also, an INRIA report about the software used in that election claims that
We have shown that some attacks allow the attacker to cheat, depending on the attacker model. We cannot say whether such attacks have been exploited during the 2022 election. We have shown that these attacks wouldn’t have left any detectable trace, not even under later investigation.
Alas, I know a lot less what the situation is like on other continents.
