Graeme Archer: You Can't Always Get What You Want

By Graeme Archer

Look, I know, do you think I want to be writing about AV again, anymore than you want to be reading about it again? Sorry about this, and I did have a half-completed article about Tower Hamlets, but I can’t stop obsessing about the referendum. While I don’t think either Yes or No campaign has covered itself in glory (ego-speak for “I would have done things differently”, the rallying cry of bloggers everywhere), there are some myths being pumped out by the Yes campaign which should be addressed in a public forum. (It is politeness, rather than honesty, which leads me to label them as “myths”).

First Myth The Conservatives use AV to elect their leader, or: It’s the same voting system that The X Factor uses. This claim is made over, and over, again, and it is nonsense. The mechanism of the Tory leadership (s)election is so bizarre I don’t think it should be used as a model for anything (other than, of course, for the election of Tory leader, in which realm it is unsurpassable; this rapid revision of my opinion has nothing to do with the fact that Candidate List re-assessment is underway), but the X-factor comparison is both false, yet actually quite instructive.

What happens in X Factor is a sequence of exhaustive run-offs. Every week you (oh alright then, we) phone in and vote for our favourite act. The singer with the smallest number of votes is expelled (I’m ignoring the judges’ intervention for simplicity). All of us then get to change our mind about who to vote for next. (Readers of a certain age will have an image of the late, great Mike Reid in their head, shouting “Runaround NOW”.) This is a crucial difference with AV, which requires every voter to list preferences regardless of the actual outcome at the end of each round.

In X Factor terms, it would be like being locked into voting for Olly all the way through, even when it becomes apparent in later stages that Olly’s doing fine, but probably can’t win, while Stacey, whom you've grown to love, is at risk from a surge in support for Jedward. (I may be mixing up the years here, after a while they all blur into one. Sorry Olly). (Actually the system which most resembles X Factor/exhaustive run-offs is … a sequence of FPTP elections, held over time. It’s how seats change hands under FPTP, absent changes in background levels of national party support).

AV supporters dismiss this as somehow unimportant. But it is not unimportant, it is a devastating flaw in AV’s claim to be a proper representation of the conditional preferences of voters. Forget The X Factor (if only). Suppose five candidates are standing in, ooh I don’t know, one of the Brighton seats. At the end of round 1, what if the top three are close together, including my favourite, who’s come top, but the second placed is my least favourite, while I could live with the 3^rd, while we can assume that the 4^th and 5^th placed, who got my 2^nd and 3^rd preferences, are not going to make it. Why can’t I move my 2^nd preference to the currently 3^rd placed candidate? (Yes I know this is complicated, but bear with, bear with). (If you think I’m being arcane, go back to look at the votes David Davis received at the end of the first and second rounds of the Tory leadership election, and ask yourself if it could have occurred under AV: it couldn’t). After all, the second preferences of the voters for the least popular candidate are about to be redistributed by the AV algorithm. Why not mine? This is the …

Second Myth Everyone only gets one vote in AV. What is meant is that in the final round, each voter is represented once. But (see First Myth), voters who gave their lead preferences to unpopular candidates do have their lesser preferences redistributed – that is, they have more than one vote. If I gave my first preference to the candidate who came out on top in Round 1, I am not allowed to redistribute any of my lesser preferences, in order to either protect my top choice or switch horses to protect a candidate I mind less than another one who’s likely to pick up momentum in Round 2. If you vote for a candidate who proves to be unpopular, you get to vote more than once. If you vote for the most popular candidate, you don’t. That’s a fact. AV supporters should stop insisting on the contrary, as though the mechanism by which the final outcome is declared is shrouded in secrecy and unavailable for inspection.

Third Myth AV leads to MPs with the support of more than 50% of their electorate. (So, by the way, do safe seats – which I thought under the esoteric theology of AV was A Bad Thing? Only teasing). Of course it’s not true, or rather, it’s not necessarily true. I was asked by a few people after the last article to ‘prove’ this, as though it was some claim that relied on the higher mathematics. Of course it’s just a trivial outcome of the fact that each voter need not list n preferences for n candidates. If enough list only one or two, the victor may well have less than 50% support (Proof: An election where every voter listed only one preference is permissible under AV and is identical to FPTP. QED). Actually let’s go back to the tease about safe seats. Safe seats are those where an MP really does have more than 50% support of his or her electorate – of their first preferences. AV can only ever, in general, get to the 50% figure by giving first-preference weights to the second-, third-, fourth- etc preferences of voters for unpopular candidates. (This is what I loathe most about AV but I’ve given up trying to convince anyone else to share my skepticism about calling a system “fair” where the last preferences of voters who support unpopular candidates are given an equal weight to other voters’ first choices).

Just a last word on safe seats. The lament is that supporters of the non-dominant party are effectively disenfranchised, as their votes “don’t count”. Of course they do. You just need to organise, to work harder, to bring about the change you want, all the while humming that Stones song inside your head. When I was born, in North Ayrshire & Bute, our MP was Sir Fitzroy Maclean. I started as a very young Tory – there is a photo somewhere of the great man at my christening. That very safe Tory seat is now an SNP/Labour marginal. Now, I’m getting on in years, but I’m not that old. From safe Tory seat where the votes of socialists “didn’t count”, it went to safeish Labour seat where the votes of Nationalists “didn’t count”, to a safe leftwing seat where the votes of Tories “don’t count”: in (ahem) forty years (alright, forty-one). I’ve heard something similar said by Tory friends about our experience in East London; that there is apparently a sea of second and third preferences out there in Hackney, which AV will let us tap into, if only we can change the vote counting algorithm. I doubt it’s true, and even if it were, I’d still prefer to fight to be the most popular candidate in an election, and not just the guy who hopes to hoover up the second and third choices of people who’d really rather have someone else. The rotten-ness of some of our inner London boroughs is a function of the socialism which runs them, not of how we count the votes.

Look. I doubt this will change a single person’s mind. Either you are in thrall to one of the myths, and you’ll vote yes (it is a property of a myth, isn’t it, that repeated assertion of it lends it a potency regardless of its foundation in fact), or you’re, like me, a fan of having the most popular candidate elected by one person one vote, and you’ll vote no. But between now and May 5^th, for the sake of this statistician’s sanity, if nothing else, could we focus on discussion of the attributes of the AV algorithm? And give all the made-up blarney about which unpopular national figure is more or less likely to benefit from AV a rest? I could have produced another page about the negligible impact of AV on minor parties’ electoral results: take that as a warning! Now go and enjoy the sunshine. And remember to vote No to AV.