Monthly Archives: June 2017

Out for the Count: What I learned At the Election Counts.

It’s now over a week since the count for the general election. I’d not been to an election count before; now I’ve been to two in five weeks. The first was the count for the City of Edinburgh council elections, the second the count for the general election, which was called part way through the council campaign, for the five Edinburgh constituencies. The two counts were in the same place, the halls of Meadowbank stadium. (The general election counts are commonly at Ingilston, but that was already booked for the garden show.) They had a lot in common, but were also very different.

The council elections in Scotland are run under the single transferable vote proportional system. City of Edinburgh council wards each have three or four councillors, and voters rank as many candidates as they want, from 1 up to the number of candidates. If your preferred first candidate is knocked out, your vote goes to your second preference, and so on. And if your first choice is elected, any “surplus votes” for that candidate are also transferred.

The maths in this can be quite daunting – for instance, surplus votes can consist of second or third preferences which need to be reallocated. Complicated. Fortunately it is done by computer. I have no idea how the programmes work, but there is an audit trail. You can even follow the votes flow from one candidate to another.

Here’s his the count worked. It was the day after the election – very civilised. They start with postal votes. Postal ballots from each ward, or perhaps voting area (a division of a ward – each voting area has a single polling place, where you cast your vote) are out on a table and unfolded. Representatives from political parties frantically try to record the votes as they are unfolded, because they can get useful data on how the poll went and where people voted. (As it happens, this was a complete waste of time, because the data shared by the council included all that. Whether it does every time I’m not sure – people were surprised at the level of detail released, down to voting area. Each ward has about 35 pages of data – you can download PDFs from the Edinburgh council elections, should you choose to do so; I presume other councils in Scotland are equally open with their results.)

The same is then done for all the ballots placed in person into a ballot box – when you actually go and vote in person. The party representatives again frenetically try to record the data. The ballot papers are counted, and the number of ballots reconciled – each polling place knows how many papers were given to voters and those left unused.

All the ballot papers are placed in a cardboard box and taken to a scanner. There are several scanners so that the ballots from a whole ward can be scanned more or less simultaneously. The ballots are scanned at great speed. An image of each ballot paper is kept, the number of images reconciled to the number of ballots.

Computers read each image, and record the votes, from, say, 1 to 8 (if there are eight candidates). Any images that can’t be read satisfactorily are brought up in a screen, in public, where the candidates, their agents, and the council staff agree on the voter’s intent, if they can.

There are quite a few spoilt papers. Most were because critters had not used numbers to rank candidates, but other marks, such as an X. If only one such mark was used, that vote counts – the voter’s intention was clear. But if they had marked an X beside two or more candidates, no intent could be inferred, and the ballot paper was spoilt. Similarly if one number was used more than once to rank candidates, the rest of the votes were spoilt: if a voter ranked candidates 1, 2, 3, 3, 4, 4, 4 votes 1 and 2 were counted, the rest discarded.

Sometimes voters spoilt their papers in more imaginative ways. There are some who drew penises in the paper; a single penis might be taken as intent (perhaps against the voter’s real intent), more than one would be spoilt. One voter drew frowning smileys in each box. (I did wonder if a single smiling face amongst them would have been acceptable.) My favourite was the voter who wrote beside each candidate their view of their attractiveness; I thought “pure belter” should have been taken as an indicator of preference, but this wasn’t the common view.

The computers did the hard mathematical work, council staff the hard physical work and managing the count, including the machines. The whole thing was very efficient, and open – it was a very well run process. I was impressed: it maintained an audit trail between a voter casting their vote on paper through digitisation to the final digital result. Some enterprising people have even taken the data and produced animations, so you can see how the votes were cast and allocated in any ward in Scotland: it is very impressive.

The general election count last week was also very well run. But completely analogue. I was surprised by the importance of rubber bands and PostItâ„¢ notes in the process, as will become apparent. I don’t doubt the result at all, but the reliance on paper when it is clear better, more rigorous methods (such as scanning the ballot papers) exist seems baffling and archaic. And since it is done immediately after the vote, starting at 10pm, everyone is exhausted, too.

Here’s how it went. Ballot papers (either postal or in person) are unfolded from a ballot box and put into bundles of fifty, secured by a rubber band and signed off on a PostIt by the person counting it. Someone else counts the bundle to check it really is a bundle of fifty rather than 48 or 52, and (if it is) countersigned on the PostIt. All the bundles are counted up, with the last balancing (less than fifty) so that the number of votes front each ballot box is known. This is then reconciled to the number of papers handed out and those returned, ensuring the integrity of the ballot and giving the turn out in each voting area.

Whilst the ballot papers are being unfolded, party representatives are once again busy recording what votes they can see. This time it’s really important – because there are no digital results by voting area: no results but the final one. No data analysis is provided, because everything is on paper. Entirely analogue.

Once all the votes are bundled and counted, the counters are given a hard earned break and a sandwich.

Then it’s back to the bundles. The bundles are sorted by candidate (in our representative democracy, you vote for a candidate not a party), with piles of votes for each candidate. The party representatives keep eagle-eyes on this bit of the process, making sure their candidate’s votes don’t get put on their rival’s pile. (They don’t care if a rival’s vote gets put on their pile, obviously.) Except that the party representatives have been up since 5am and it’s now 2am and most of us were hallucinating by this time anyhow, and just wanted it all to stop.

Spoilt ballot papers are weeded out at this point. On some it apparent what was intended – for instance, where a voter uses a number rather than a cross, or a cross was placed next to the candidate’s name rather than in the box. Some were blank; one had “wankers” scrawled through the candidates’ names. Others had “none of the above” added.

Each pile is then counted – into bundles of fifty, secured by a rubber band, and signed off on a PostItâ„¢. And then counted again to make sure, and countersigned. The numbers for each candidate in each ballot box are not released, though it would be possible to work it out if one wasn’t knackered.

All the bundles are then moved to a central table, and piled up beneath signs for each candidate. It was pretty easy to see how close the vote was. The biggest pile is the winner. These piles are counted to give the final result. This is why, when there’s a recount, the numbers of votes often change in steps of fifty. And why that change can be significant. I heard that in one English constituency, a recount had found an additional 1000 votes: presumably 20 bundles had been transferred to the wrong pile, and the error only discovered when a recount was requested. (This is why it is important that party representatives tally up votes as ballot papers are unfolded: whilst you can’t see every paper, you do see a sample, and added up across all the ballot boxes, each party had a pretty good idea of the result before it is announced.)

The thing is, this is all done by tired people. It is easy to imagine mistakes being made. There are many recounts, sometimes because mistakes are made (and corrected) sometimes because the result is so close. In Fife North East, a few miles north of the Edinburgh, there were several recounts: in the end the majority was just two votes; if three people has stayed in from the rain, the result might have been different. (I wonder if there are many people in Fife feeling guilty that they didn’t vote; or pleased that they did. They know their votes count.)

Being at two election counts so close together, using different systems, has strengthened my view of democracy: but it is has also reinforced my belief that our parliamentary democracy would be better served by a proportional system, like that used in Scottish local elections. Seeing those piles of votes, each ballot paper bearing a representation of a voters views – and most of them are ignored. The winner-takes-all result of first past the post, as well as analogue method of counting, feel like something from the nineteenth century (as indeed they are), rather than a twenty first century means of electing are representatives – in which most votes don’t count.