Does electronic voting undermine democracy?
If you ask a random person in the world of computing, there are good chances they will not have a good opinion on electronic voting1. Indeed, this population, more informed than average, knows that there is no information system without flaws, whether they come from bugs, design errors, or intensional insertion of "backdoors". Moreover, the majority of the electronic voting systems in use today require to give the keys of the voting boxes to a "trusted third party", without any possible check on the correct conduct of the ballot, from which a legitimate distrust arises, all the more important when the stakes of an election are great.
But it would be absolutely naive to be content with rejecting electronic voting on the whole because it is already in use everywhere: from professional elections, to associations... and even in certain political elections. Moreover, many politicians are now advocating for a general implementation of electronic voting, in particular with the aim of improving the rate of participation. We therefore shouldn't refuse to look at and analyze the risks and benefits induced by this type of ballot. Or rather, we should say, these types of ballots. Electronic voting can be of two very different kinds: voting machines and Internet voting. We will deal with both but take care to distinguish between their issues.
The voting machine is a sort of specialized computer, installed in the polling center, and which serves to replace purely-paper ballots. Some of these machines do not completely replace paper because they print the voter's ballot which they will drop into a ballot box, allowing a double-count of the votes, supposedly making the results more reliable2. The machines can be connected to the Internet, to a local network, or be completely isolated, knowing that isolating machines make external attacks harder but not necessarily impossible, for example if the attacker comes and connects to them directly. Finally, each vote takes place in two stages (registration of the voter and vote casting). The machines can be designed for the second stage only or the combined process. Voting machines are already widespread, notably in the United States, but also in Brazil, in Venezuela, in India, in Belgium and they were much used in the Netherlands until their abandonment in 2008 (due to a lack of trustworthiness of the machines reported by researchers and several institutions).
Voting by Internet introduces a radical change in that it allows every voter to vote from wherever they want, from any computer. In this second case, the electronic voting system (generally a web interface coupled with a server) necessarily integrates the two processes of registering the voter and counting the vote. Certain advantages are clear. It is easier to vote because it is not necessary to go out to do it. It is also easier to organize the election because it is not necessary to set up a polling station (unless you want to preserve the possibility of voting in person for the electorate). Certain drawbacks are also obvious: it is easier to hack into an individual's computer than a voting machine, the servers tallying the votes are necessarily connected to the Internet and so vulnerable to attack, and finally one can never be completely sure that it is really the elector who is voting and that they are not voting under duress, or under the scrutiny of a family member...
Voting in political elections by Internet is allowed in a few places in the world (in Estonia since 2005 but also in some Swiss cantons and for French people living outside France) but it is already extremely widespread for internal elections of associations, for administrative councils and staff representatives in businesses, etc. The importance of these types of election should not be under-estimated, especially the economic stakes that they represent. That is why we should be just as careful about the procedures and software employed.
Before dematerializing ballot papers
One thing is sure, voting did not arrive with digital computing, so non-technological solutions are possible. To be able to evaluate electronic voting, there has to be something to compare it to.
In the United States, in order to reduce the cost of elections and increase turnout, among other reasons, it is traditional that all the polls are held on the same day of the year, called "Election day". Moreover, the electors are consulted on a very large number of questions: the choice of their political representatives of course, but also that of judges and sheriffs, as well as local referenda. The consequence of this is that the ballot papers are generally very long. To make the counting easier, most jurisdictions already use optical scanners which do the work in place of humans. We can already see these machines as voting machines, and replacing them with a machine on which the voter directly registers their choices can be seen as a good thing, especially because it avoids ambiguity, badly filled ovals3... In France, however, since the electoral process is so simple and well tested (a printed ballot for each candidate or list, transparent ballot box, citizens assessors, exit polls to get results as quickly as possible...), one can seriously question the basis on which some city councils are now investing in voting machines4.
The criteria for a successful election
The good functioning of representative democracy rests on elections. For those elected to have the necessary legitimacy, the population must think they were elected fairly. Therefore all citizens, and in particular representatives of the various candidates, have some control over the election process, from beginning to end (the ballot box is transparent and public, anyone can ensure that no ballot is added irregularly nor withdrawn and that voter anonymity is preserved — thanks to voting booths), and more particularly on the counting (anyone can attend). It is important that electronic voting respects the same criteria, especially of transparency. It seems that if, for the most part, citizens place their confidence in the reliability of voting machines5, a citizen who is less trusting than the others is not in a position to verify the electoral process, unlike the case of paper voting. This need to trust is an abnormal novelty accompanying electronic voting6.
It is on the base of this argument that a recent article (in French) opposes Internet voting: "There are of course technical assurances provided by the providers of electronic voting and the administration which entrusts them with the founding act of democracy, but while traditional voting is a matter of transparency, Internet voting becomes matter of trust. You have to trust the provider's sincerity and the reliability of the entire process, which must be guarded against abuse or hacking. And yet democracy is founded on transparency, not on trust."
Now there exist electronic voting systems (machines or Internet voting) allowing a certain degree of transparency to be re-established, but unfortunately they are too little used. These are the voting protocols called "end-to-end verifiable", of which even free and open source implementations exist7. These protocols are not perfect however. In particular, understanding the "verifiability" mechanism presupposes a good level of maths or else trust in mathematicians and computer scientists who will have taken the trouble to verify it. However, it is easier to trust the very many people who will have taken this trouble, than the small number having to validate the opaque systems in use today. Moreover, some researchers are working at the moment to improve the experience of electors wanting to verify that their ballot has really been taken into account.
Is electronic voting worth it?
We have seen that electronic voting, both on a machine or through the Internet, introduces a significant number of complications, which could discredit the results of the election8. Technological solutions allowing a certain form of verifiability exist, and it is therefore necessary to use them when the choice of electronic voting is made. However, it is equally necessary to very carefully weigh the advantages and the risks before making this choice. The most common arguments of an increase in turnout with the introduction of Internet voting or lower cost when using voting machines are not enough to justify taking that risk, especially when the existing electoral processes are functioning well.
The argument of turnout in particular is to be compared with the importance of the election. In a poll of lesser importance (like the one to designate an association's board of directors), one can imagine that voters are not willing to make the effort of displacement. It is thus reasonable to rely on a vote by Internet or by mail. However, for political elections in which the stake is enormous, citizens should be prepared to spend (at most) half an hour to vote. In the case of France, where the polling stations are numerous and well organized, that's all it costs. In countries that organize their very first elections as Afghanistan, people are willing to wait for hours in order to vote. The turnout problem can thus be solved in other ways than the chimeras of technology or compulsory voting. It can be solved by a political environment that puts greater emphasis on the issues at stake. The only acceptable exception is where voters living abroad are concerned. For them, enough polling stations cannot be set so as to make voting easy.
Nevertheless, other reasons could impose the use of technology and citizens should judge whether the good they bring is worth to bear the risks and challenges of electronic voting. Greater participatory democracy is a prime example. When the City of Paris asked the Parisians to vote on the allocation of 5% of the city investment budget, no guarantees were taken to prevent ballot stuffing. Anyone could go in any borough hall and vote without control or registration. Anyone could create one or more accounts on the website of the city to vote. If participatory democracy is to grow and be legitimate, there will be a need for more secure practices. In this case, the high frequency of public consultation would justify engaging in electronic voting.
Retractable votes are another example of a miracle that only technology would make possible. Yet, no one seems to be defending Internet voting in that aim. In this unusual democratic setting, there would not necessarily be regular elections. Every citizen could change their mind at any time and reassign their vote from one candidate to another. Elected representatives would thus become permanently accountable to the people...
Of course one can find computer scientists who support electronic voting, in particular among those working to develop such solutions. It is interesting, however, to note that there are opponents of Internet voting for the most important elections even among the creators of Internet voting systems. For example see: https://vote.heliosvoting.org/faq ↩
If one day you have had the opportunity of filling out a questionnaire intended for optical scanning (for example, in standardized tests), you must have been confronted with those little ovals that have to be completely filled. ↩
According to France Élections, one of the three companies supplying voting machines approved for the political elections in France, the criteria of choice are cheapness, simplicity and perceived reliability. Yet the latter is far from satisfactory if one believes http://archive.francesoir.fr/actualite/societe/elections-machines-voter-moins-fiables-que-papier-28041.html. ↩
A survey commissioned by France Élections and carried out by OpinionWay in 2012 asked to 1021 French people the question, "Do you trust the French electoral institutions (state, city councils, civil servants) to put in place and supervise the use of voting machines in France?" and 68% responded 'yes'. This result appears in section 3 of the inquiry report entitled "Trust in voting machines". ↩
According to this article http://www.agoravox.fr/actualites/citoyennete/article/ordinateurs-de-vote-la-confiance-23198, the only case when we should rely on trust is for proxy voting. It follows that the use of voting machines would be equivalent to a gigantic system of forced proxies. ↩
For example, for Internet voting, the Hélios and Bélénios systems offer such guarantees. If that does not make these systems invulnerable and therefore not necessarily suited to political elections, it is very unfortunate that most associations and companies which allow Internet voting hire service providers relying on obscure and unsafe technology, rather than going for these open solutions promoted by researchers working on electronic voting. The eVACS system which is developed and exploited by the Australian company Software Improvements claims to offer similar guarantees (and as in the case of Helios and Bélénios, it is free software). ↩
The controversy that followed the close election of George W. Bush against Al Gore in 2000 was also partly due to the use of voting machines. It is to avoid such difficulties in the future that the Voting Technology Project was launched by Caltech and MIT. ↩