This article is by Julian Yon and originally appeared on his blog here.
“Ankh-Morpork had dallied with many forms of government and had ended up with that form of democracy known as One Man, One Vote. The Patrician was the Man, he had the Vote.” (Mort, Terry Pratchett)
I am again writing in response to a specific concern raised on Twitter. The context is the forthcoming referendum to decide whether the UK should adopt AV (the Alternative Vote system, also known as Instant Runoff Voting) for its general elections.
It is worth dispensing with a red herring before continuing. AV is not a way of achieving Proportional Representation, and it does not try to be. Therefore all arguments relating to PR are moot. There are no PR options in this referendum, and therefore a vote for or against AV is not a vote for or against PR. The only other option being offered is FPTP (First Past The Post), the non-proportional status quo.
I’d also like to dispense with the “AV is complicated” myth. Offer a typical 5 year old strawberry, chocolate and vanilla ice cream. Ask her which one she likes most. Then which she likes next best. And then which she likes least. Chances are she can answer you without much difficulty. My daughter (who is far from typical and has significant learning difficulties) can manage it. This is how you vote in an AV poll. You don’t need to know exactly how the votes are counted (not that it’s hard) in order to cast your vote.
AV is a system for minimising “wasted votes” while, in the context of a general election, maintaining the strong constituency link which has often been used as an argument for FPTP. It is used for electing an individual person to a single seat, and therefore has no bias against independents. The intention is to eliminate tactical voting and “safe seats”, by ensuring that every vote is counted and that a majority is required to win.
AV is not flawless. It is probably the smallest and simplest improvement that could be made to our existing system. Taken alone, it falls short of what many campaigners for electoral reform would have liked to have seen. However, it is an improvement. It is somewhat harsh to have to vote for a “lesser of two evils”, but this is what is on offer. Neither outcome guarantees further reform, and this may be the only chance we get for a generation. Because of this, I will try to keep my examples simple. It’s easy to get hung up on small details when the way we elect our MPs is only one of things wrong with our constitution.
So, what is a “wasted vote”? Put simply, it is a vote that is correctly cast but cannot make a difference to the outcome of the election. They happen under FPTP because it is designed for two party elections. In that simplest case the two systems are equivalent. Either A or B will get >50% of the vote, and win with an outright majority. Every person’s vote has an equal weight and a direct effect on the outcome. One man, one vote.
It starts to go wrong when we allow a third candidate to stand. As long as that candidate has some support, there is the possibility of a democratically incorrect outcome due to wasted votes. The probability of such an outcome depends on the circumstances of the election, but it can happen in both marginal and “safe” seats.
Let’s pretend there are three candidates, imaginatively named A, B and C. People go to the polls. The results come in. A has 41%, B has 31%, C has 28%. Under FPTP this is a win for A, but no candidate actually has a majority. There are many different political situations which can lead to this outcome, and not all of them give A a democratic mandate. AV provides a way to differentiate between these situations.
In order to highlight just how badly wrong FPTP can theoretically be, let’s say that B and C are two candidates with very similar manifestos. This is the notorious “split vote”. If B was not in the running, every one of B’s supporters would vote for C, giving a win with 59%, a clear majority. And vice versa. By the accident of two people standing on the same platform, the minority candidate wins the seat.
The important way that AV differs from FPTP is that it insists on a majority to declare a win. This is achieved by taking second (etc.) choice votes into account. But, crucially, each vote only counts once. Nobody gets multiple votes. Here’s how it works (but you don’t need to know this to vote):
It’s Friday afternoon, and the teacher has promised the class something special. But they must choose collectively which activity they want to do. The teacher puts signs up in each of the corners with a letter that represents each activity, and the children have to gather under the signs. For argument’s sake, these made-up children are not influenced by their peers’ choices.
They gather as follows: 11 under A, 10 under B, 6 under C and 4 under D. There is no majority, but there is a loser. The teacher removes the sign for D and tells those 4 children they’ll have to choose again. One heads to B, and the others to C.
We now have 11 for A, 11 for B and 9 for C; still no majority. The teacher removes the sign for C, which is the new loser. Five children move to B, and the rest to A. Now there are 16 votes for B, which is a majority, and the decision has been made.
Now of course, only in fairy tales would children not be influenced by their peers. But this is where AV is actually simpler than the above example. You’re not asked to choose again, you get to vote in private, and you cast your vote once. All you have to do is write your order of preference, and let the ballot counters do the rest. They will play out this algorithm, taking your choice into account. One person still has one vote, just like in the two party FPTP, but as minority options are eliminated that vote is transferred to your less preferred candidates, in the order you specify. The winner will have a majority, and therefore a mandate.
If you have any trouble understanding the mechanism, remember that you’re only casting the vote, not counting it. It’s just like choosing your favourite ice-creams. A five year old can do it.
 As a tie in a large-scale election is statistically extremely unlikely, I will assume it cannot happen.