Consultation Response: Electoral Pilots at the C.E.P. and Local Elections Tim Storer, Ishbel Duncan, Alan Ruddle, Matthew Kreeger and Anastasia Gemona 18th November 2003 Introduction and Pilot Assumptions The Government recently announced it was preparing for a further stage of electoral reform pilots, to be held in 2004 at the scheduled European and local authority elections. The intention is to combine these two elections, and conduct all postal voting pilots in three European regions and a further e-voting pilot in one of those chosen regions. The pilots are a significant change from those conducted over the last three years, since they will not be restricted to local authority elections. Under the terms of the Representation of the People Act [1], pilots may only be conducted at local authority level. Since the government is keen to test the reforms on a larger scale, primary legislation has been introduced into the House of Commons in order to allow pilots at European Elections [2]. In 1999 turnout fell to just 24% at the European elections, so the pilots are considered a test of whether reforming voting mechanisms are capable of improving turnout[3]. During debates in Parliament, it became apparent that the pilots have raised a number of concerns, particularly amongst the opposition parties, who have noted the security considerations of conducting remote elections (electronic or otherwise) and the worth of addressing voting mechanisms as a means of improving. This document has been prepared by the Digital Privacy, Security and Trust Group at the University of St. Andrews* in response to the consultation paper published by the Office of The Deputy Prime-Minister (ODPM) [4]. We respond to specific questions raised by the consultation, laying particular emphasis upon the e-voting component of the pilot, where our particular experience is strongest. *Digital Privacy, Security and Trust at St. Andrews The Digital Privacy, Security and Trust Group was formed in the autumn of 2001 at the University of St. Andrews. The group is based within the School of Computer Science, but has members from the schools of Sociology and Management, as well as the University s IT services department and links with outside organisations. The group has active research projects developing requirements engineering techniques for incorporating security into software and is exploring the topic of online criminality and network security. In addition, the group is considering the prospects for a practical remote electronic election in the UK. 1
Q1: Timetable Practicality As the consultation notes, the timetable for the proposed pilots is extremely tight. Legislation to allow the pilots to be conducted has not yet received Royal Assent, and from the anticipated timetable, will not do so until March 2004. In order to compensate for this restriction on time, the intention from the consultation document appears to be to conduct preparation work in parallel with the progress of the bill, selecting regions to conduct the pilot, choosing vendors etc. We cannot comment too closely on the general timetable for the pilots, although we would assume that the Government would require chosen regions to prepare fallback plans, should the legislation not receive Royal Assent on schedule and conventional voting mechanisms are required. In particular, we note the confusion arising in California during the recent recall and election of the state Governor, amid a major transition to new voting technology as a result of the recent Federal legislation[5]. Our experience raises considerable concerns on the feasibility of implementing e-voting systems within the stated time period. Software engineering projects commonly suffer financial over-runs and time delays because of the complexity of the product. As such, we would make the following recommendations regarding the e-voting pilots: The pilots should utilise technology from a vendor that has already participated in an earlier pilot in the United Kingdom. This would have the effect of increasing the timetable for implementation to several years, since the experienced vendor would have had time to develop a more reliable and tested product. The pilot should take place in a region where the Returning Officer, staff and political parties have a stated enthusiasm for an e-voting pilot in 2004. Although we do not doubt the professionalism of election officials in complying with a pilot order, the tight timetable suggests a higher likelihood of success where participants have expressed a prior interest in conducting a successful pilot. The pilot should take place where e-voting has been tested before. This provision would exploit the experience of election officials in the region, anticipating and remedying difficulties. In addition, the pilot would be able to test whether sustained use of e- voting has any influence on turnout. It is not suggested that the short timetable for implementation will inevitably lead to failure of the pilot, although, we would recommend that the scope of the e-voting pilot takes this restriction into account. We would note that the controversy regarding electronic voting in the United States is partly due to the rushed implementations following the debacle of the 2000 Presidential Elections. Limited pilots in 2004 do not rule out further pilots at a future election, and indeed, a successful pilot this time round would gather support for expanded future pilots. Q2: Funding No response. 2
Q3: e-voting A recent debate on the Bill passing through Parliament to facilitate the 2004 pilots raised some serious concerns for us regarding the integrity of the e-voting technology to be employed[6]. The responses given to concerns raised about verifiability were, in our opinion unsatisfactory. Although we appreciate that the Government intends to leave technological recommendations to the Electoral Commission, we are concerned that the Electoral Commission will in turn devolve responsibility to individual vendors. As strong proponents of remote electronic voting (in contrast to many other computer scientists who have conducted research, or commented in this area), we are concerned that the consequences of badly implemented electronic voting system (as occurred in the United States) will arise here. Part of the difficulty experienced by United States has been the need for a voting system that can manage the large number of races that are placed on each ballot. Previously, this has led to an accumulation of paper ballots which electoral administrators have found difficult to manage. Mechanical voting mechanisms were introduced in the United States in order to reduce some of the paper management, a motivation very different from the UK Government s modernisation agenda. The later move to electronic voting machines has been to facilitate the use of voting mechanisms for a wider electorate (different languages, coping with disabilities etc) without re-introducing the logistical difficulties of paper ballots. A consequence of this need for mechanisation is the perceived lack of verifiability of ballots for individual voters. Various anecdotal evidence suggests means for tampering with all the mechanical or electronic voting machines in use in the US[7]. The absence of verifiable mechanisms makes such claims hard to disprove, such that elections can be disrupted merely be spreading uncertainty[8]. As such, much of the debate there has been focused upon the need for a voter verifiable paper trail, in which an electronic voting machine is used to produce a printed paper ballot, displayed to a voter from behind a glass screen. Once the voter has verified that the paper ballot has recorded their vote as they cast it on the machine, it is dropped mechanically into a sealed ballot box. Under ordinary circumstances, the electronic ballots recorded in the voting machine are used to calculate the ballots - in the event of a dispute the paper ballots can be examined. Of course, for remote elections, the provision of such a paper ballot is more difficult to achieve. A fact missed by most commentators is that elections that utilise a paper ballot are not necessarily more verifiable for the voter than elections that use purely electronic ballots. We believe that the electoral process is verifiable when it is robustly scrutinised by both independent and partisan observers. The traditional UK polling station system is successful because the ballots and the voting process may be scrutinised, both in the polling station and during the count. The obscurity of many electronic systems in the U.S. has led observers to suggest the electronic medium itself is to blame, which we do not believe to be the case. Paper based elections are just as vulnerable to manipulation when scrutiny breaks down, through systematic ballot stuffing, or theft of ballot boxes, for example. We believe that whereas in the past, scrutiny of the electoral process has been conducted by cooperation between the Returning Officer and staff and representatives of the political parties, the remote electronic medium provides an opportunity for the public at large to scrutinise the election process and ensure that their own ballot was counted correctly. The advantage of this approach would be to re-engage the electorate with the process by which a result is tabulated, without overly taxing each individual elector s time. Finally, we would like to comment upon the motivation for implementing electronic voting 3
for elections in the UK. Commentators have noted that the last three years of pilots have shown that postal voting has had a significant effect on turnout, whilst the results from electronic pilots have at best been indifferent. Additionally, technical problems during electronic pilots have received far more coverage than administrative difficulties of postal pilots (such as strikes by Royal Mail workers). We would agree with such commentators that the end usefulness of electronic voting is not yet clear, in contrast to the immediate increases in turnout resulting from the postal voting pilots. However, we would argue that the electronic pilots so far conducted have been on a more limited scale than postal pilots. The electorate perhaps do not yet associate the new technologies with the electoral process, or as comfortable with them as they are with posting a paper ballot. Without wishing to recommend endless pilots in order to perceive the benefits of electronic voting, it may be that initial expectations should be relatively modest, with a greater expectation of benefits in turnout for the long term. We would recommend that the goal for e-voting pilots conducted over the next few years should be to establish public confidence in the new technologies and ensure that implemented casting mechanisms are robustly scrutinised by a wider population as possible. Should these objectives be achieved, electronic voting would be available as a convenient mechanism for future mainstream elections and would likely see its take-up increase in later years. Q4: Publicity Evidence from the recent Parliamentary debate suggests that where pilots have been conducted, innovative means have been employed in order to raise public awareness of the new mechanisms of voting. Contributors described a range of techniques for advertising the nature of pilots being conducted in particular regions. The comments made by Lindsay Hoyle MP were particularly notable, describing a previous postal voting pilot in Chorley, we got the newspapers involved, because the media have a part to play...the local newspapers were helpful and urged their readers to get behind this new idea...peoples imaginations were caught. They were asking each other have you voted yet? Not wanting to trust their votes to Royal Mail, they were arriving at the town hall to deliver them in person. There was a ballot box for those who wanted to use it, which had to be emptied four or five times a day. [6, c585] The advertising campaign began as a conventional cooperation between the local council and newspapers, but ended with the electorate effectively advertising the pilot to each other, far more effectively than officials could have achieved by themselves. Such partnership campaigns that engage all the political parties, the election officials and local media are most likely to be successful in raising awareness of the pilots. The emergence of websites serving specific geographic regions could also facilitate this distribution of information. If implemented with care, email campaigns could also be conducted that notified relevant mailing lists of the details of electoral pilots in a region, or direct them to websites that contain the relevant information. A further simple example would be to encourage the active political parties to include details of the pilot in their election literature, paying particular attention to describing what to do if something goes wrong. For example, the literature could include the final date by which postal votes should have arrived, together with the telephone number to ring if it hasn t. Additionally, party workers who are participating in door-to-door work such as canvassing 4
or walk-abouts should be fully briefed with the details of the pilot so that they can inform electors who have queries. In this way, political parties can be involved in the electoral process without disrupting their campaigns. The parties are aware that the electors who they are targeting with leaflets and/or canvassing are the ones they are likely to need to ensure have voted, so their partisan interest is employed in the greater effort of increasing awareness of the pilot mechanisms, and hence increasing turnout. Q5: Evaluation Techniques We regard the evaluation of previous pilots by the Electoral Commission as thorough in terms of reporting upon the efficacy of different mechanisms in improving turnout, and also in general terms regarding the ease of use of the remote electronic mechanisms[9, 10]. We note that these reports consider the statistical evidence from the pilots as well as public opinion through post election surveys. We feel that such surveys are extremely valuable for deciding upon the use of different voting mechanisms in future pilots, and indeed in mainstream elections. However, we would also recommend greater analysis of verifiability of the new mechanisms. For example, rather than considering confidence in a particular voting mechanism, we would recommend experiments with receipts for remote ballots, or at least asking electors if they feel they would prefer to hold some proof of their vote after having cast it remotely. We note the concerns of some commentators that the issuing of receipts in remote elections is liable to encourage vote buying or coercion. However, in practice, all remote elections are coercible, since they are conducted out with the security of the polling station. We believe that if the polling station is dispensed with, then the issuing of an explicit receipt to electors should at least be considered. In addition to surveying the voting mechanisms, we feel it would be of use to consider the implications of changes to the electoral system upon election campaigns conducted by political parties. Although it is not intended to suggest that reform should be conducted on behalf of the parties, they have traditionally been involved in the administration of the election and the scrutiny of the counting process. Further, how political campaigns are fought do have an influence upon the wider public, with the presentation of policy ideas coming to the fore during manifesto publication for public consideration, and other events. Although such research is probably not within the statutory remit of the Electoral Commission, other bodies may be capable of taking on this role. After the 2001 General Election, Stephen Coleman (now Professor of E-Democracy at the Oxford Internet Institute) studied the influence of the Internet on the election campaign and public discussion[11]. The government might consider employing an organisation such as this in evaluating the effect of remote electronic elections on the European and local authority campaigns. Q6: Non-financial Support and Assistance We have described a number of activities that we believe could facilitate successful electoral pilots in 2004. A key theme of these recommendations has been the need for partnership between various organisations in order to achieve a successful outcome, from implementation, to raising awareness and participation. In order to foster such partnerships we believe the government should employ the Electoral Commission as a depository where the experiences of the different pilots (both past and present) may be compiled and discussed. In a sense, the 5
Commission is already moving towards this role by producing reports on the previous pilots and consulting with regional returning officers on the appropriate locations for the 2004 round of pilots. However, we feel this role could be expanded to provide more information for the political parties, as well as a greater amount of technical information for both the postal and e-voting pilots. If such information was collected centrally, returning officers and staff could quickly access the latest information about the pilots, including final changes to procedures, as well as providing a source of advice (backed up by evidence) of how to proceed with pilots they have not attempted previously in their area. Finally, we would recommend that the Electoral Commission establishes a database of election fraud (both past and present) for the UK). As no official database currently exists, this could provide a tool for returning officers and other election administrators when they are checking for indications of electoral fraud within their own boundaries. The database should contain both allegations and cases of conviction for fraud, given the difficulty of prosecuting fraud when it occurs. A database of electoral fraud would also go some way to satisfying opponents of election reform by at least providing evidence for levels of fraud (proven and allegations) under different voting mechanisms. We do not suggest such a database would dispel fears of widespread fraud, but it would at least move the debate from anecdotal evidence and allow proper evaluation of the risks of different voting mechanisms. Summary of Recommendations In summary, we make the following recommendations for the electoral pilots in 2004: Realistic scope for e-voting implementations, given the time available. Employ e-voting mechanisms that widen rather than restrict the public scrutiny of the election process. Fostering of partnerships between political parties, electoral administrators, media organisations and the electorate themselves in order to distribute information about the pilots. Widening the scope of the Electoral Commission to begin gathering and distributing information on the practical experience of election administrators during pilots as well as evaluation of the e-voting technologies tested during pilots. Crucially, we would argue for a central database of cases of electoral fraud. Such a database would be indispensable if quality evaluations of different voting mechanisms (and their exposure or resistance to fraud) are to be made. We feel that current practice relies too much upon anecdotal allegations rather than sound evidence or convictions. Such a database may well demonstrate the widespread fraud that occurs in the presence of remote voting. Alternatively, the database would indicate that fraud is prevalent at extremely lowlevels regardless of voting mechanism employed. However, without such information, it is impossible to speculate and much more difficult to recommend any voting mechanism for conducting elections 6
References [1] Representation of the People Act, 2000 Ch. 2. [2] European Parliamentary And Local Elections (Pilots) Bill[HC](2002-03). [3] House of Commons Library Research Paper 03/76: The European Parliamentary and Local Elections (Pilots) Bill. [4] Roy Williams. Electoral Pilots at the C.E.P. and Local Elections - Consultation Paper. http://www.odpm.gov.uk/stellent/groups/odpm\_localgov/documents/page/ odpm_lo%cgov_024004.hcsp. [5] Help America Vote Act, 2002 (P.L. 107-252). [6] HC, Vol 411, cols 510 603, October 21, 2003. [7] Rebecca Mercuri. Electronic Vote Tablation, Checks and Balances. PhD thesis, University of Pennsylvania, October 2000. [8] Peter Jackson, Colin Rosenstiel, and Seamus O Connell. Ballot secrecy. Electoral Reform Society, 1997. [9] Sam Younger(Chairman). Modernising elections: A strategic evaluation of the 2002 electoral pilot schemes. The Electoral Commission, July 2002. [10] Sam Younger (Chairman). The shape of elections to come: A strategic evaluation of the 2003 electoral pilot schemes. The Electoral Commission, July 2003. [11] Stephen Coleman, Nicola Hall, Chris Quigley, Peter Kellner, Jason Buckley, Kate Mc- Carthy, Mark Pack, Beccy Earnshaw, and Greg Sedberry. 2001: Cyber Space Odyssey: Spinning on the Web: e-campaigning and Beyond. http://www.btinterface.co.uk/ reports/report\_pdfs/cyberodyssey.pdf. 7