Dutch govt to explore electronic voting — once more

UPDATE 2016-06-07: the Dutch House of Representatives (lower house) voted down the bill proposed by Taverne (VVD) that sought to blow new life in involving electronic means (i.e., computers) in the voting process. The bill itself did not specify what those means should look like, but specifically, an experiment was foreseen to use electronic ballot printers (specifications available here) that printed out a ballot with the voter’s electronic vote, where the voters would put these in a ballot box, and a scanner would electronically count the ballots (while still allowing manual verification afterwards). For the foreseeable future, the Netherlands will be voting using the traditional red pencil.

UPDATE 2015-06-16: it is reported that the Dutch Minister of the Interior is considering an experiment with electronic voting during regional municipal district elections in 2016 (in Dutch: “herindelingsverkiezingen”; not to be confused with the regular nation-wide municipal elections, which are scheduled for March 2018).

On September 17th 2014, the Dutch Minister of the Interior announced his intent to carry out security tests with internet-based voting services. On May 12th 2015, he further informed the parliament about this. The tests are to be carried out by the end of 2016. This particular activity is aimed at facilitating internet-based voting to Dutch voters abroad. The Netherlands is however also examining the possibility of re-introducing electronic voting inside the Netherlands,which — in the current design — will have a paper trail. The remainder of this post addresses that topic, with the aim of shedding some light on the current state of play.

In the Netherlands between 1970 and 2007, voting machines could be used during municipal elections, on a voluntary basis. Only a few municipalities chose to use pencil and paper. In 2006, a large debate took place in the Netherlands that shed light on risks associated with electronic voting (“We Don’t Trust Voting-Computers”), including issues of eavesdropping via EM emanations, issues with reliability, and issues with transparency/verifiability of the vote count. As a result, since 2009 all elections in the Netherlands are based on pencil and paper: national elections, provincial elections and municipal elections.

Two independent committees were established to investigate how it could happen that public trust in voting machines got lost, and how the electoral process should have new safeguards in the future. In 2013, as result of ongoing technological developments, yet another committee was established to re-assess safeguards in the electoral process and possibilities for the use of electronic devices.

In February 2015, the Dutch Minister of the Interior announced (in Dutch) it will examine re-introduction of electronic voting using a method that prints the voter’s choice on a piece of paper, which is then automatically scanned by a computer (for instance by a camera), but still allows manual verification.

Here is a fragment from the Minister’s letter (translated):

Security and costs

The answers that the Van Beek Committee provided to my questions confirm my view that the weighing of requirements for the vote printer and the vote counter is complex. That is especially the case for security. Specifically, the question is what risk profile the security should be based on.

In my opinion, the Van Beek Committee made a right choice by considering the paper voting process to be direction-giving for the electronic counting of (paper) ballots. By using that as a basis, errors (potentially as result of manipulation) in the vote printers and vote counters should not remain undetected. The voter can, after all, check whether the printed ballot contains the choice that he/she wanted to make. By checking the correctness of electronically counted votes, it is possible to detect incorrect counts. Of course it is a good thing that measures exist to detect errors, but if those errors come to light during the election day itself, nothing can be done to redress it. If many or all vote printer work improperly, then voting must be ceased. If it is detected that vote counters work improperly, the printed ballots can be counted manually. These are risks that, if they occur during an election day, can have significant impact on the progress of an election.

In more generally terms, there is the question, also addressed by the Van Beek Committee, whether it can be acceptable that persons and/or groups, outside the elections, can demonstrate that the vote printer of vote counter are not adequately secured. In 2006 that happened with the voting computers that were used back then. It was shown on TV how software on the voting computer could be manipulated, because no security measures had been taken to prevent that. It resulted in a debate on the reliability of the voting computers.

In my opinion, wide consensus about acceptable risks is necessary for a decision to introduce the vote printer and vote counter. Consensus must thus exist about the way in which these systems must be secured. Without wide support, the risk of the reliability of the vote printer and vote counter becomes and remains a topic of debate. That isn’t good for the trust that needs to exist in the systems.

The security level turns out to be of great impact on the costs of the vote printer and the vote counter. The Van Beek Committee provided an estimate of costs (150 to 120 million euro initial investment, then 6 to 10 million euro per election), but could not make it more precise. It has however been found that costs exist that have not been taking into account by the Committee. The Committee finds that no preciser estimates are possible at the moment, because of the large number of uncertain factors, among others because fundamental parts must first be further specified.

What’s next

I still believe that the introduction of the vote printer and vote counter can have benefits for the accessibility of voting and for the counting of ballots. On the other hand I find that the introduction of these IT systems has many complex issues, and is surrounded by uncertainties. That requires careful deliberation.

I agree with the Committee’s finding that, considering the potential benefits, it is worth the effort to take the next step by examining whether it is possible to eliminate uncertainties and reduce complexity. The assumption is that it will then be possible to make (much) more precise estimations of the costs. The Committee recommends that the Ministry of the Interior further develops the specifications for the vote printer and the vote counter. That is a useful proposal. I do think it is essential that during that development, it is constantly monitored whether there is wide support for the specifications. I intend to, as recommended by the Committee, establish a group of (external) experts that have knowledge of the relevant IT topics and of public administration. I promise the Parliament to inform you about the progress at the end of May 2015.

We currently wait for the Minister to fulfill the latter promise. Meanwhile, he did inform the parliament on May 12th 2015 (today) about his intent to examine the security and costs of internet-based voting, meant to facilitate Dutch voters located abroad. Although he did not reference the developments described above, a system suitable to allow internet-based voting to voters abroad may also be suitable to allow internet-based voting to voters located inside the Netherlands.