Remote voting in the UK House of Commons – Remote Divisions become reality

On May 12, 2020 the UK House of Commons conducted its first remote Division (remote vote).

UK Parliamentary Business – News – MPs cast first ever remote votes in Commons Chamber
The vote was conducted through MemberHub, the UK Parliament’s member website, which has Microsoft authentication.  Multi-factor authentication (MFA) was used to protect the authentication for the remote voting (the Internet voting).

There is some background on the development of the system in a Wired UK article by Chris Stokel-Walker: Inside the troubled, glitchy birth of parliament’s online voting app

Messaging about the voting system, which piggybacks on existing parliamentary IT systems, through the MPs MemberHub application, hasn’t been enormously clear. …

“We were asked to start looking into it just before Easter weekend,” says Matt Stutely, of Parliament Digital Services, who has been developing the voting service. Stutley dug out what he calls “a dusty chest of war plans we have in case we were ever asked to implement [online voting]”.

UPDATE 2020-05-14: Matt Stutely, the Head of Business Systems Development for the Parliamentary Digital Service, has written a blog post about the process of developing this service in the incredibly tight timeline of four weeks.

MPs make history with remote voting – the story of how it happened

In early April 2020, we were asked by the House of Commons to build a remote voting application for Members in just four weeks.

He indicates that making a service for remote voting (Internet voting) for the House of Lords will be next.

END UPDATE

UPDATE 2020-05-13: On May 6, 2020 the Procedure Committee wrote to the Speaker about the remote voting system.  The correspondence system has the full letter (PDF).

Members who by their actions facilitate a non-Member to cast a vote in a division of the House are very likely to be found to have committed a contempt of the House and to have breached the Code of Conduct, and can expect to be punished accordingly.

Call for Evidence

The Procedure Committee is conducting a Call for Evidence about all aspects of changed procedures during Coronavirus restrictions.  The call ends 3 June 2020.

Full Report

On May 8, 2020 the Procedure Committee issued a full report regarding remote voting in divisions.

This report notes:

The integrity of the system depends on Members. The remote voting system is not as secure as a system where a Member must vote in a division lobby in person.

and the Rt Hon Karen Bradley MP, Chair of the Procedure Committee, said

The present remote voting system was developed at high speed as a temporary measure for use during the pandemic.

For more information:

There is some technical detail in the full report, although at a very high level.  See Technical aspects of the remote voting system on pages 11-16 of the PDF above (items 23 through 51).

24. System security is delivered by the use of MemberHub, which uses single sign-on and multifactor authentication. All data is encrypted and sent over a secure connection, and voting records are stored in both MemberHub and the existing electronic divisions system. The bicameral Information Authority has issued a decision statement confirming it is content with the information security of the remote voting system, taking account of advice it received from the National Cyber Security Centre. The Speaker has been informed of the Information Authority’s decision.

28. The existing arrangements for divisions in person through the lobbies have particularly secure authentication arrangements which may be evident but are worth repeating here. To gain access to a voting lobby a Member must first gain access to a secure area of the estate using a security pass with a photo, and must pass a number of security staff and doorkeepers. In order to vote successfully, a Member who has taken his or her seat in the House25 must pass through a lobby containing several other Members and typically actively patrolled by party whips, and must then give a name to a division clerk and pass out of the lobby between two tellers.

29. This high level of authentication is not replicated in the remote voting system over MemberHub. …

30. The Committee’s opinion on the suitability of the remote voting system over MemberHub is given on the basis that the system is designed for temporary use during the COVID-19 pandemic and has not been designed for permanent use to replace the existing arrangements for physical divisions.

END UPDATE

Remote Division

Before the remote Division, the Speaker made a Statement, including:

I ask all Members to pay careful attention to what the Procedure Committee says about the integrity of the system. As the Committee states, any attempt to allow anyone who is not a Member to vote is likely to be a serious breach of privilege.

The UK House of Commons and UK Parliament Twitter feeds shared images:

Remote Division was called.

The results are in Hansard and can be viewed in detail at https://votes.parliament.uk/Votes/Commons/Division/783

More detail about the system is expected to be forthcoming in a blog post by the UK Parliamentary Digital Service this week.

Parliamentary votes are different from votes in a general election in at least three major ways:

  1. Votes can be coerced (in fact the role of the Whip is basically to enforce party direction on how to vote)
  2. Votes are not anonymous
  3. Votes are not secret

That being said, there are still lots of considerations for remote voting and technology voting, including concerns about the chain-of-custody, as multiple systems are most likely involved with the transmission and counting of the vote, concerns about auditability and concerns about security.

Auditability is a really challenging one.  Basically either each individual MP would have to check that their vote has been counted based on their intention, and even then, they’re no longer all standing in a room where they can see how other members voted (unlike the Canadian system where members stand one-by-one to be counted, in the UK MPs literally go to gather together by Aye and No votes in two physically separate locations, as described in the Voting section of MP’s Guide to Procedure).  Unlike counting people in a room, online it’s hard if not impossible to get a good sense of whether the vote count reflects the votes cast.

Security is also a challenging one given that computers can lie, with customized malware capable of showing one result (e.g. an Aye vote) on screen and sending another (e.g. a No vote) to the voting software.  In that light, it’s worth mentioning that the vote took place over the web on Patch Tuesday, with both Microsoft and Adobe releasing patches for vulnerabilities (“A remote attacker could exploit some of these vulnerabilities to take control of an affected system.”)

It will be interesting to learn what risks were identified and how they were mitigated.

There is also a larger question, deeply related to human intentionality, about the physical and psychological differences between literally standing to be counted or literally voting with your feet by moving to one room or another, versus tapping a square on a screen.

Remote voting (Internet voting) in a Parliamentary context is different from electronic voting in the chamber itself.  I covered some of the considerations for in-chamber voting in the Canadian context in my blog post Electronic voting in the Canadian House of Commons.

The First Incorrect Votes

In a remote Division on 13 May 2020, the Deputy Speaker reported

I have been informed that a small number of Members have inadvertently cast their votes, by electronic means, in the opposite way to the one in which they intended to vote. I am informed that their use of technology was not quite as good as they felt it ought to be and that a few Members have made a mistake. There is no provision under the current temporary system by which a Member can change their vote once it has been cast, but I am satisfied that even if a small number of votes had been cast in a different way it does not affect the result of the Division.

When such a situation is detected and affects the result of the Division, the Speaker has the authority to call a revote:

If problems in the conduct of a remote division which might have affected the result are reported after the result is announced, the Speaker may declare the division to be null and void and make arrangements for it to be re-run.

Auditability in a Whipped Parliamentary System

This also gets to a point about voting in a whipped Parliamentary system, which is that in the absence of a free vote, Whips are expecting votes along party lines, which makes it pretty easy to detect potential voting errors.  So there are definitely different auditability concerns than in a totally free vote; even if an individual member doesn’t notice they have voted opposite from their intent, their party is likely to notice very quickly.

SIDEBAR: This is another example of how Internet voting in a Parliamentary context differs from Internet voting in a general election.  In a general election, in order to preserve the secret ballot and to limit coercion, it must not be possible for anyone, including the elector, to show how they voted, or to verify how they voted.  Which makes one wonder e.g. how many Ontario and Nova Scotia municipal Internet votes might have been incorrectly cast, with no way to verify the intended result.  END SIDEBAR

News Story

In a story that I think is probably from PA Newswire, with headline including “amid remote voting errors”, it was reported

The division list showed 22 Conservative MPs supported the amendment, and in theory rebelling, although they included Chancellor Rishi Sunak – who made a mistake in the voting process rather than staging a shock bid to depart the Government.

A source close to Mr Sunak blamed “online teething problems with the system”, adding: “The Chancellor did not intentionally vote against the Government. He called the chief whip straight away to explain.”

As dozens of newspapers and news sites carried the wire story, you can pick your source, the first one that comes up in Google for me is the Express and Star.

Background

Remote voting (Internet voting) was authorised by the UK House of Commons Speaker on May 6, 2020 and was extended to May 20, 2020 by agreement of MPs.

The system was developed by the UK Parliamentary Digital Service.  Thanks to the Parliamentary Digital Service and Head of Business Systems Development Matt Stutely for responding to my questions on Twitter.  Thanks to the Procedure Committee, on Twitter @CommonsProcCom, for sharing links to its detailed report.

Elections Quebec consultation on Internet voting

Elections Quebec (Élections Québec) is consulting consulted on Internet voting (vote par internet).

https://www.electionsquebec.qc.ca/internetvoting/

The consultation closed November 3, 2019

You could answer a survey in English or en français.

You could also submit a position paper in English or French by email to

consultation_vpi@dgeq.qc.ca

Information en français https://www.electionsquebec.qc.ca/voteparinternet/

There is also a short URL which will land you on the French page http://voteparinternet.quebec/

Moratorium on electronic voting

I think it’s important to remember that in its 2005 elections, Quebec had severe problems with electronic voting.

“We all remember the events that marked the municipal elections of November 6, 2005,” recalled the Chief Electoral Officer. “Not only did the systems fail, but the corrective measure proposed were insufficient, poorly adapted and often came too late. … “

« Nous nous souvenons tous des événements qui ont marqué les scrutins municipaux du 6 novembre dernier », a rappelé le directeur général des élections. « Non seulement des systèmes ont fait défaut, mais les correctifs proposés étaient insuffisants, mal adaptés et souvent tardifs. … »

The problems were so severe that the Directeur général des élections du Québec (DGEQ) declared a moratorium on electronic voting in Quebec.

Many Canadian Internet Voting Studies

It is also worth noting that Ontario, British Columbia, New Brunswick and the Government of Canada all did various types of extensive studies or consultations on online voting at the provincial or national level, and all of them rejected it.  Nova Scotia and PEI have also examined online voting at the provincial level and rejected it.

See Canadian reports recommending against Internet voting.

Many National Internet Voting Studies

In addition, Finland studied and recommended against national Internet voting, Norway actually did trials of Internet voting and then rejected it, and Switzerland (which never had very much Internet voting) now has none because of security issues.

See Internet voting at the national level.

No Internet Voting in Canadian Federal Elections

For an overview of why Canada doesn’t have national Internet voting, including questions about turnout and security, see Why is there no online voting in Canadian federal elections?

Internet voting in Switzerland

There is currently no Internet voting in Switzerland, primarily due to security issues.

It’s complicated to write about Internet voting in Switzerland for several reasons:

  • Switzerland has a political structure of cantons, voting is done by canton with different systems in each canton
  • Switzerland does not have a history of voting privacy; historically and in a few locations even today voting is done by a public show of hands
  • Switzerland has many votes throughout the year on what are basically referenda
  • Switzerland has a good, but quite complex, set of regulations around Internet voting

Internet voting has been an option in some cantons.  I believe testing began in 2004.  Because of the Swiss Internet voting regulations, as best I understand the maximum percentage that can vote online is 30%.  (More than 30% voting online triggers additional requirements.)

The 30% figure is a bit misleading however.  Because only some cantons participated in online voting trials, it was open to just 3.8% of the overall electorate in September 2018 (and now is not available at all).1

1 Source – Slides “Trust in e-voting” (PDF, 1 MB, 07.02.2019), from Federal Chancellery FCh > E-Voting

As indicated above, the absolute number of voters was always relatively small.  In my own analysis of reports available online, I find that under 5% of the eligible voters vote online, representing 200,000 or fewer votes per voting period.  (My understanding is that voters have to register in advance to vote online; it’s not clear to me whether the numbers in these reports are just the number of registrations, or the actual number of ballots cast online.)

The map below summarizes the online voting testing that has been done by cantons, as well as making it clear that there is currently no online voting at all (in French « Pour l’instant, il n’est pas possible de voter par voie électronique en Suisse », roughly translated “For the moment, it is not possible to vote online in Switzerland”).

La Suisse - Essais de vote électronique dan le cadre de scrutins fédéraux
La Suisse – Essais de vote électronique dans le cadre de scrutins fédéraux

Above map from Chancellerie fédérale ChF > Vote électronique.

Turnout

Research indicates that turnout did not increase, specifically youth turnout didn’t increase.2

2 Internet voting and turnout: Evidence from Switzerland, by Micha Germann and Uwe Serdült in Electoral Studies, Volume 47, June 2017, Pages 1-12. https://doi.org/10.1016/j.electstud.2017.03.001

Background on Geneva and Swiss Post

Geneva developed two systems, CHvote 1 and CHvote 2.  As best I can understand CHvote 1 has been suspended, and there’s no money to further develop CHvote 2 to the level it would need to reach.

Swiss Post developed two systems, including a new one with a third-party for-profit private vendor.  The old system is being discontinued.  As required by Swiss law, the new system was put to a public intrusion test (with restrictive conditions) and the source code was made available (with restrictive conditions).

Swiss Post makes a remarkable claim about the new system.

The new system with universal verifiability was subject to a public intrusion test (PIT) in spring 2019. During the test, it withstood attacks from over 3,000 international hackers.

This is at best misleading.

The conditions on both the general testing and the availability of source code were restrictive.

There was not in any sense either unrestricted public testing nor unrestricted publicly available open source code.

And, through access to the source code outside of the restrictive agreement, three serious flaws in the system were found.

You can read e.g. Researchers Find Critical Backdoor in Swiss Online Voting System by Kim Zetter.

Three reports are available about the Swiss Post system from the Swiss government site, two in English and one in German.

    • Final report Locher, Haenni and Koenig (English) – (PDF, 1 MB, 29.07.2019) – Members of the e-voting research group at the Bern University of Applied Sciences BFH (Philipp Locher, Rolf Haenni, Reto E. Koenig): analysis of the cryptographic implementation of the Swiss Post voting protocol
    • Final report Teague and Pereira (English) – (PDF, 731 kB, 29.07.2019) – Vanessa Teague (The University of Melbourne, Parkville, Australia) and Olivier Pereira (Université catholique de Louvain, Belgium): analysis of the cryptographic protocol and its implementation according to the system specification
    • Final report Oneconsult (German) – (PDF, 303 kB, 29.07.2019) – Oneconsult: Review of Swiss Post’s operational security measures

Why is there no online voting in Canadian federal elections?

The short answer is that a Parliamentary Special Committee on Electoral Reform studied the issue and recommended against online voting in federal elections in their 2016 report

Recommendation 4
The Committee recommends that online voting not be implemented at this time.

and in 2017 the Government accepted the recommendation in its response to the report

We will not implement online voting at this time.

Here are some Frequently Asked Questions:

  1. Q: Don’t many countries offer online voting in national elections?
  2. Q: Won’t online voting increase turnout, including increasing youth turnout?
  3. Q: Aren’t there studies showing smartphone mobile voting increases turnout?
  4. Q: What standards does Canada have for online voting and computer vote counting?
  5. Q: Don’t we already have the security needed to conduct online transactions?
  6. Q: Aren’t these kinds of attacks on online voting theoretical?
  7. Q: Maybe we need more study of Internet voting in Canada?
  8. Q: What do computer scientists say about online voting?

Please feel free to ask additional questions.

Q: Don’t many countries offer online voting in national elections?

A: No, the only country in the entire world that offers online voting for all citizens in national elections is Estonia, population 1.3 million. The majority of Estonians still choose to cast their votes on paper; only 247,232 votes were cast online in the 2019 Estonian Parliamentary election.

Estonia’s turnout in its 2015 national election was 64.2%.  That’s turnout lower than Canada in its own 2015 national election; Canada’s turnout in 2015 was 68.3%.

Estonia’s 37.6% turnout in the 2019 European Parliamentary elections ranked it 20th of 27 countries for turnout.

For more information see:

In terms of other countries and national online voting:

Q: Won’t online voting increase turnout, including increasing youth turnout?

A: No.  Over and over again, at different levels of government in different countries, no.

Estonia’s turnout is generally declining.

Switzerland saw no effect on turnout.

For many other examples see Online voting doesn’t increase turnout.

I understand that it seems intuitive that voting is about convenience, so online would increase turnout, or that voting is about familiarity, so making it online would increase voting by young people who are already used to doing things online.  This mental model of voting convenience increasing turnout is simply not supported by the evidence.  People choose to vote or not vote for complex reasons.

Q: Aren’t there studies showing smartphone mobile voting increases turnout?

A: No, there was a single study. With a sample size of 144 (one hundred and forty-four) votes.

The paper was presented at the 2019 Election Sciences, Reform, & Administration Conference (ESRA 2019), in the section Public Trust in Elections.  Promises and Perils of Mobile Voting – Anthony Fowler (Univ. of Chicago).

You can see the sample size on page 7: “In total, 144 votes were cast with a mobile device in the November election.”  You can see the total number of votes cast at https://sos.wv.gov/news/Pages/11-16-2018-A.aspx (or archived version in the Internet Archive).

Here’s some advice on citing scientific studies: be very careful when citing a single study that has a small sample size.

It’s also important to note that in the exact same paper, on page 2, Fowler states:

Several European countries abandoned internet voting after seeing that the increases in turnout were not as large as expected…

Q: What standards does Canada have for online voting and computer vote counting?

A: Canada has no standards whatsoever for online voting and computer vote counting (vote tabulation).  Really.

Sometimes systems for computer vote counting at the Canadian municipal level are certified to voluntary US standards, often extremely dated (2005) voluntary standards.

Ontario has finally recognized this as an issue, with Elections Ontario recommending establishing standards and certification for elections technology.

Q: Don’t we already have the security needed to conduct online transactions?

A: Voting is a unique interaction due to the need to both verify that an individual is eligible to vote AND ensure that each individual’s vote is completely anonymous.  This is a very different problem from e.g. online banking, where transactions are not anonymous, in fact they are known to both the bank and the individual.  Banking online is actually not completely secure, but the shared knowledge of transactions enables false or erroneous transactions to be reversed.

The US National Academies of Sciences, Engineering and Medicine (NASEM) has a comprehensive study process, used to arrive at a scientific consensus.  In their report Securing the Vote, they find “There is no realistic mechanism to fully secure vote casting and tabulation computer systems from cyber threats.” and “the risks currently associated with Internet voting are more significant than the benefits”.  For more information see Securing the Vote – US National Academies 2018 consensus report.

The requirement for anonymity is based on the history of voting, where voting was heavily coerced.  People used to have to vote by public declaration, and would be paid or rewarded for voting a certain way (or punished if they didn’t vote for the desired candidate).  For an overview of some of that history from a US perspective, see Andrew Appel’s video Internet Voting? Really?.  For an extensive account from an Australian perspective, including the history of the creation of the secret ballot (also known as the Australian ballot) see Judith Brett’s book From Secret Ballot to Democracy Sausage.

If the idea of someone trying to compromise a Canadian election seems improbable to you, please check out Canadian reports on election security and misinformation, including two reports from Canada’s government cybersecurity agency.

If the idea of someone attacking Canadian government computers seems theoretical to you, it’s important to know that key Canadian government departments including the Finance Department, Treasury Board Secretariat, Defense Research and Development Canada and the National Research Council have all been successfully hacked.

Q: Aren’t these kinds of attacks on online voting theoretical?

A: No, see Practical Attacks on Real-world E-voting (including Internet voting in section 7.3) by J. Alex Halderman and e.g. Researchers Find Critical Backdoor in Swiss Online Voting System by Kim Zetter.

Q: Maybe we need more study of Internet voting in Canada?

A: Online voting has been studied and rejected at the provincial level by New Brunswick, PEI, British Columbia, Ontario, and Nova Scotia.  It has also been studied and rejected at the municipal level by Toronto and Waterloo, amongst other municipalities.

See Canadian reports recommending against Internet voting.

Q: What do computer scientists say about online voting?

Computer scientists, particularly computer security experts who specialise in election security, are overwhelmingly opposed to online voting.

For more information see Internet voting and computer security expertise.