The government has been urged to reconsider plans that would see the 2020 general election being held on the same day as the London elections.
The assembly has said that it needs to have time to procure the suppliers if it is to e-count votes – Photo credit: PA
The London Assembly’s election review panel has said that this could create a “perfect storm” of election complications for those in the capital city, with severe implications for the technology used to count the votes.
Speaking to PublicTechnology, Duminda Baddevithana, a programme manager within the Greater London Authority’s technology group, said that his team needed “the clearest of steers” from central government on the issue.
“We haven’t formulated a firm outline plan for the [technology for the] 2020 elections, mainly because of the impending perfect storm of having the general and London elections at the same time,” he said.
Running the London elections – which involve three ballot papers with three different voting systems – requires a lot of planning and coordination, with the electronic system used to count and process the votes needing repeated testing beforehand.
This e-count may prove too difficult to organise if it has to happen at the same time as the general election, Baddevithana said. “We have to ask: would it be better all-round if – if the London vote is on the same day – we did a manual count for the London elections, rather than an e-count?”
However, this is a time-consuming process and a better option would be for the London elections to be moved to a different day. But if that is the case, the elections team would have to begin procurement for the contracts to run the e-count soon.
“We need to get into procurement for an e-counting supplier by March 2017 at the latest, because it takes a year to go through the process,” Baddevithana said. “After negotiation, that would take us to mid to late 2018, and then we will have just under two years for the testing and planning, and time will disappear just like that.”
There is even greater pressure on the GLA to get the testing right next time, after a bug in the database used in the e-counting process delayed the results for this year’s mayoral elections.
The mayoral election counting process involves ballot papers being machine read and put into a database before the outcome is calculated based on first and second preference.
However, “small discrepancies” between the reported votes and the number of ballot papers were noticed during the count – a problem caused by a system overload meaning the code used to identify the first and second preferences didn’t return them in the right order. (Baddevithana stressed that the results, however, were correct and were also verified after the count.)
A review of the issue showed that the problem was unlikely to have been noticed without testing the system with the same number of ballot papers as received in the general election, which was 8 million in 2016. Because testing involves creating a set of fake ballot papers with a known result, this would have cost a lot of time and money.
But Baddevithana noted that the second part of the system, which calculates first and second preference, doesn’t necessarily require the ballot papers – for instance, if a database could be created instead this would cut costs and make testing at full volume possible.
He said that, based on the experience of this year’s vote, he would like to see the planning and testing process start earlier.
“The period of testing we had was pretty robust, but I would want us to be testing earlier and testing more frequently,” Baddevithana said. “That’s why it’s so important government decides about the timeline now.”