Report On The Voting System For London Elections
Report on the voting system for London Elections (further to a report presented by the Open Rights Group)
The Open Rights Group recently presented a report into the electronic counting process used by London Elects for the London Mayoral and Assembly Election held on 1st May 2008. We concur with the findings of that report and make our own observations.
The system in place for the London Elections
The Count for the London Elections was conducted electronically. This had a number of implications:
The ballot papers had to be readable. This lead to a decision to have three separate ballot papers which was a sensible. One for the Mayor with first and second preference votes, one for the London wide Assembly top up list and one for the London Assembly constituency seat.
The ballot papers were each A4 sized but of different colours.
The ballot papers were designed to be posted unfolded into a specially designed, plastic self assembly ballot box.
At the close of polls the ballot boxes were to be sealed and delivered to a central location in each Borough (usually the council offices). From there the ballot boxes were loaded onto a lorry, usually in a cage and delivered to one of three counting centres.
The lorries were unloaded at the counting centres (Olympia, Excel and Alexandra Palace) and the unopened boxes were placed in the centre of constituency compounds within each counting centre.
When the count started, the seals on the boxes were opened and the contents of each ballot box was scanned. The first part of the scanning process checked that the number of ballot papers within the box matched the number of votes cast as certified by the Presiding Officer at the polling station concerned. There was a degree of tolerance to allow for voters walking out of the polling station with a ballot paper in their pocket. Observers were allowed to see screens showing this process taking place.
Once the total contents of each ballot box had been verified, the actual votes marked on the three different ballot papers were read by the scanners and applied to each candidate or Party.
If the voters intention was unclear, perhaps they voted for too many candidates in error (or deliberately), or none, or there was a mark outside the regulation box, then the ballot went to first level adjudication. The unclear ballot appeared on a terminal screen which observers could see. Two staff would look to see if the voter’s actual intention was obvious and if so correct the system to allow the vote. If it was unclear or if the voter had not expressed a clear preference (perhaps deliberately) then the ballot was forwarded to second level adjudication. There were one or two second level adjudication screens per constituency, one of which was a large screen. The Constituency Returning Officer would make decisions on the votes which had not been accepted at first level adjudication. Votes could only be rejected as spoilt at second level adjudication. Again observers could view this process.
Plasma screens were sited in each counting centre showing cumulative totals for the candidates or parties for each of the three contests (for Mayor, London wide top up and Constituency member) per constituency. The totals did not show relative percentages or aggregated total votes but merely displayed the information in terms of bar graphs.
The information as to the cumulative and final totals per constituency were electronically passed to City Hall. The constituency member result was calculated and announced at each relevant counting centre. The London wide top up and Mayoral results were declared one at a time as they came in late on the evening at City Hall.
The individual ward results breakdown was released some weeks later.
The traditional process for conducting a count
The traditional manual method for conducting a count allowed for scrutiny at every stage of the process to ensure transparency and confidence in the process, so that interested parties could satisfy themselves that there could be no question of cheating, or that any errors could be identified and corrected.
The differences between the traditional practice and the method employed by London Elects can be summarised as follows. (In both cases a representative of the candidate known as a Polling Agent is allowed to be present within a Polling Station to observe the proceedings, and they may affix their own seals on the ballot boxes).
The traditional method of polling was that a ballot paper (typically A5 sized) would be folded and placed in a narrow slot into a metal ballot box.
At the close of polls the ballot boxes are taken to a hall, often at the main council offices, for the count which would take place immediately.
At the count the ballot boxes are opened in turn. The candidate’s Counting Agents should be shown that the seals are unbroken first. The ballots are laid on a table, in front of the Counting Agents, unfolded and counted to ascertain the turn out, and make sure that this agrees with the figures compiled by the Presiding Officer of each polling station.
If there is a discrepancy the candidate’s agents are informed and may comment and object.
The ballots are then apportioned between the various candidates, and put in bundles of equal number of votes. The bundles are stacked and then counted.
The bundles of votes should be in clear view of the Counting Agents. It is possible to request a flick through of bundles to ensure that none have been mixed up.
The result is then declared. If it is close it is possible to ask for a recount, either of the bundles (in case a bundle for one party had been put in the stack of bundles for another party), by flicking through all bundles, or by a complete re-count.
It should be possible for any candidate to effectively know the result prior to it being declared by keeping accurate tallies of votes that are seen when the ballots are being unfolded, and by checking the number of bundles, by ensuring that the same number of votes are included in each parties bundles and by ensuring that the total number of votes for each candidate agrees with the turn out figure.
The decline of the traditional method
The ‘correct’ procedures are not observed in an increasing number of authorities and lax practices have become commonplace.
Increasingly flimsy plastic or even cardboard ballot boxes are being used.
It has become the practice in some authorities to delay the start of the count until next morning. The ballot boxes are stored at an unidentified site in a manner where they are out of sight and beyond the scrutiny of any candidate’s observers
Some authorities are known to put the bundles of counted votes on tables out of sight of the Counting Agents so that it cannot be verified that the stacks for each candidate are not getting mixed up – deliberately or by accident.
Some Returning Officers regard it as an impertinence for an Agent to request a flick through.
However in general, there are systems in place with manual counts to ensure that the process is carried out in a legitimate manner. It is not always easy to do this as the count takes place at the end of a long tiring day. But potentially a candidate may arrange for the process to be observed if it is carried out in the correct traditional manner, and everyone candidates, parties and the electorate may have confidence in the final result.
Postal votes
Traditionally only a small number of people were allowed to have a postal (or proxy) vote. The reasons for being able to have a postal vote were quite restrictive. As a result postal votes were opened on the morning or afternoon of polling day, at a set time usually in the council offices and candidates were allowed to send a representative to view the process. The opened votes would then be placed in a sealed ballot box and counted later at the proper count.
As there are now many more postal votes cast, they are usually opened several days before polling day, often over several different days. This makes it very difficult for candidates to arrange for the process to be observed. The opened ballots are invariably left lying around in boxes in an unsecured manner until polling day. It is certainly not possible for the candidates to verify that the postal votes are not being tampered with.
There is a further problem in that in certain communities large numbers of postal votes are applied for and but are collected by people with influence who then vote on behalf of the actual voter. There is also evidence that vulnerable people (e.g. the elderly in sheltered accommodation) are also preyed upon for their votes.
As postal votes constitute a large proportion of the total vote cast, the almost compete lack of any effective scrutiny over the process is a major cause for concern. However there is no real difference between the potential for abuse with respect to normal elections compared to the London Mayoral and Assembly Election, so further consideration of this issue is outside the scope of this report.
Personation
Personation could be when someone votes illegally on behalf of someone else. Perhaps they know the person is dead or has moved, or perhaps they know the person will not otherwise vote. In some instances cases have been known where someone has voted early by pretending to be an elector and the real elector may attempt to vote later in the day. In such instances the second person is theoretically allowed to vote with a special ballot that would be taken into account if the result is close.
The other type of personation is when bogus names are entered on the electoral roll. This is an offence but is rarely if ever investigated. Often bogus voters apply for postal votes.
Both of these types of malpractice could take place under either system and are outside the scope of this report.
Places where the electronic process did not allow scrutiny
The ballot boxes were flimsy and the seals were not secure. By the time the ballot boxes were opened at the counting centres many of the seals on the flaps through which the ballots had been posted were broken. This is a cause for concern.
The ballot boxes were beyond the scrutiny of candidates from the close of polls (10pm) until the counting centres opened next morning (at about 8am). Given that many of the seals were broken this is a major cause for concern.
It was not possible to accurately ensure that the number of ballot papers in each ballot box tallied with the number entered by the Presiding Officer of the polling station concerned. A report by the Open Rights Group suggests that large scale discrepancies involving up to 41,000 votes were glossed over. Given that the ballot boxes were kept out of site for a number of hours and many seals were broken, this is a very major cause for concern.
It was not possible to know how many votes for each party were coming through the scanning machines. Only error votes were showing at the first level adjudication level. It was impossible to tell how many votes were accumulating for each Party or candidate. The only point of reference was the inaccurate and very imprecise plasma screens that showed rough bar graphs.
It is apparent that the scanning machines were set with too low a threshold in detecting marks. Numerous ‘good’ ballots appeared at the first level adjudication level, because tiny specs of dust of very small specs of ink were in a box. By implication blank ballots (where the voter did not put a mark against any candidate either deliberately or in error) would in many instances have been counted as a good vote for one of the candidates as a spec of dust or ink would inadvertently be in one of the boxes. It was admitted by staff from London Elects that this would be the case (report by Open Rights Group).
At first or second level adjudication it was difficult, if not impossible, for an adjudication decision to be re-visited if the one the officer involved had passed onto a subsequent ballot for adjudication.
It was not possible to refer to the actual paper ballot in cases of dispute. The Open Rights Group reports than not a single actual ballot paper was examined to visually determine or confirm a voters intention.
The choice of how to re-apply disputed votes rested with the staff. As the threshold for detecting errors on the papers was set too low a very large proportion of votes went through to first level adjudication, which meant that it was difficult if not impossible for Counting Agents to observe the process thoroughly.
The cumulative totals were passed to City Hall and the result was announced there. Until the announcement was made it was completely impossible to know what it might be. There was ample opportunity to change the result prior to the announcement. A simple program could have been run over the result (ward by ward if necessary) deducting for example 10% from the vote for one party and equally apportioning it to the other Parties.
The company (Indra) that supplied the computer software had direct access live to the systems that ran the electronic counting process. The more people who have access to the data the less secure it is.
The whole electronic count process took from approximately 8 am to 12 midnight to complete. The counts for a General Election in London (on a much higher turnout) and London local elections (with complicated multi-voting) take on average a third of that time. Clearly electronic voting is highly inefficient.
Summary
There has been a steady decline in the level of scrutiny available to candidates and parties to ensure that our elections are conducted in an above board manner. Any electronic counting method effectively removes any prospect of scrutiny and replaces this with a system based on blind trust. On one level this trust is placed in the ability of the systems to accurately reflect the number of actual votes cast. Evidence shows that electronic systems cannot be trusted to accurately count votes.
On another level this trust is placed in the integrity of employees and officers to carry out their duties in a neutral manner and not interfere with the process. As officers will have political views it is not appropriate or fair (either to the officers or to the candidates or parties) to place a large burden of trust upon them. The whole process must be transparent and open to scrutiny. Electronic counting is not open to scrutiny and is a dangerously flawed system.
