Data and Democracy

Out for the count : the £9 million white elephant in London’s next election

Electronic counting in London – the subject of criticism from the Electoral Commission and Open Rights Group for many years – is now spiralling in cost. The cost of the election in 2020, for which the contract was tendered in 2018, will be over twice that of the last tendering process in 2010. Worse still, the contractual arrangements are highly obscure and problems with transparency are yet to be addressed.

At the end of July,  the Greater London Authority (GLA) Oversight Committee met to review the activities of the Greater London Returning Officer (GLRO), in the 2020 GLA elections. Election results for the London Mayoral and London Assembly elections have always been counted by machines, which is known as ‘e-counting’. This round, the GLRO and politicians are keen to avoid the problems they had with software and machines in 2016, which led to long delays in announcing the results.

E-counting has been criticised in successive reports by the Electoral Commission for being untransparent, as independent observers are unable to effectively monitor the count inaccurate as clearly marked ballots are frequently counted as unmarked; and expensive as this service cost £4.1 million for the 2012 elections.

In particular the Electoral Commission advised that the GLRO carried out a ‘cost benefit analysis’ of e-counting, which should include a cost benefit analysis of manual counting. The GLRO has ignored this advice, twice.

In 2018, the tendering process for the e-counting contract for 2020 elections began. The contract was eventually split between two companies: CGI, the lead contractor, and Smartmatic the sub contractor. The cost of the contract between the last tendering process in 2010 and the most recent of 2018 has more than doubled – from £4.1 million to more than £8.9 million.

The increase in cost partly reflects the fact that the tendering process was uncompetitive – the market for e-counting providers is small, and they tend to operate in consortiums. In this case a very small number of expressions of interest were given, and the GLRO was itself very worried that they might not find a company able to meet the bid criteria, which included new measures to avoid the problems they experienced in 2016.

As part of the approval process for commencing e-counting procurement the GLRO stated that:

“Should no tender prove satisfactory either quality-wise or price-wise, there will be no commitment to award a contract and consideration will be given to counting manually” .

Nevertheless, even though the cost has spiralled to almost £9 million, the GLRO have decided to press ahead.

We are concerned that the methodology employed to assess the various bids for the contract is not being made publicly available. Without this information, it is impossible to know why £9 million pounds of public money is being spent.

There are serious questions about the sub-contractor, Smartmatic. Smartmatic’s “election solutions” have been the subject of criticism in severalcountries where they have run electronic voting systems. Smartmatic says the criticisms are unjustified.

For example, there were serious technical issues with the machines used in the 2012 Belgium elections in Flanders, with second preference votes incorrectly allocated. Additionally, there is no definitive account of why the Committee on Foriegn Investment in the United States were investigating Smartmatic. For now, Assembly members and the GLRO can only read the reports from the countries concerned, or view the hearings.

Given that the GLRO reported before the meeting of the Oversight Committee that many of the functions of e-counting had been subcontracted out further. The GLRO have stated that Smartmatic and CGI are:

“Working with Rathmhor, who delivered user training for the programme in Scotland in 2017. Hamilton Rentals will supply hardware and audio-visual services at the count centres, using the latest generation of Fujitsu scanners. FDM will print the ballot papers.”

In this long and unwieldy chain of command, is hard to see what value Smartmatic brings, and where accountability lies if something goes wrong. Only full information about the contracts and subcontracts will make it clear what Smartmatic’s role is, and whether that is appropriate.

The GLRO previously asserted that CGI’s and Scotland’s performance in Scotland where e-counting at local elections is also the norm, warranted the awarding of the contact – that they are tried and tested. 

In its report to the GLA Oversight Committee the GLRO stated: “CGI and Smartmatic have previously delivered successful electronic vote counting systems for Scottish elections.” This was false.

The Scottish Government have confirmed to us that Smartmatic have never been awarded an e-counting contract for Scotland. The GLRO have subsequently corrected the record and explained to Assembly members that Smartmatic have not in fact had any involvement with the Scottish e-counting systems.

The other delivery partner, CGI, have also not been without controversy. The company went to court in Scotland after repeated repudiatory breach with a subcontractor on an IT contract for Scottish councils. Agilisys and CGI tussled over who was responsible for the contract not being properly delivered. The judge described CGI’s expert witness as “one sided”, “not balanced” and its submissions as “unconvincing”.

The GLRO were unable to answer many of the questions put to them by the GLA oversight committee. On costs, the GLRO, at one point attempted to attribute much of the £4.8 million increase in cost on ‘inflation’.

Eventually, they made it clear that much of the cost was due to new safeguards against error, alongside a further large proportion of the price hike being simply unexpected costs at the real market. This may simply mean the cost of having an uncompetitive bidding process where the two participants do not have to constrain their estimates as they cannot be replaced.

The GLRO did not explain why they have still not done a cost benefit analysis of the whole exercise, instead preferring  “stakeholder consultation” and “soft market testing”. Nor were the GLRO able to explain why they went ahead with such an expensive contract despite previously saying that they would not. Thankfully, the GLRO have now committed to cost benefit analysis of manual counting.

However, scrutiny has been avoided for too long. The GLRO has failed to provide answers. We look forward to gaining clarity on the facts in a future meeting with them. We believe that London Assembly members urgently need answers to the following questions:

  • Given the expense and the non competitive nature of the market for e-counting, why did the GLRO press ahead despite saying “Should no tender prove satisfactory either quality-wise or price-wise, there will be no commitment to award a contract and consideration will be given to counting manually”?

  • Will the GLRO release the details of the methodology used to award the contract and how individual companies scored?

  • Was Smartmatic’s controversial reputation and poor performance record taken into consideration in awarding this contract? What role are they actually performing?

Until Assembly members have answers, neither they nor the public should be confident that next year’s elections will be problem free, or value for money. The GLRO should have stuck to its promise, given the difficulties they have had getting competitive bids, and given “consideration to counting manually”. Voting systems need to be robust, trustworthy and inspire confidence. It is vital that this is done right.