Basically, First Past The Post is fine for binary decisions: e.g. Cake or Death? But if Chocolate is offered too, the 'We just want to live!' vote may be split, allowing Death to win. (Unlikely, of course, but who knows? There may be a lot of suicidal people living in your constituency.)
In the real world, FPTP was fine in the 1950s, when most voters were making a binary choice. In 1951, 97% of people voted either Conservative or Labour – so most MPs (94%, to be precise) were able to win a clear majority of support in their constituency. But in 2010, only 65% of people voted Conservative or Labour. Our votes are now spread across a wider range of parties – and the result is that two-thirds of MPs in the House of Commons today hold less than 50% of the total vote in their constituency. They've secured the largest minority, but do they have the support of the majority? We don't know, because FPTP doesn't check that.
At its worst, FPTP has allowed BNP candidates to be elected in some council seats because the majority non-BNP vote was split. Someone on Facebook posted this image of the results from Coalville ward in Leicestershire:
But it's not the only case of its kind. See also Mixenden in Calderdale, Mill Hill in Blackburn, or Fenside in Boston, Lincolnshire.
The extremist nature of the BNP means it isn't hard to imagine that the majority of the voters in these wards did not actually want a BNP councillor. But because the non-BNP-supporting majority spread their votes across a variety of other candidates, that's what they got.
The wider point about FPTP is that any candidate who wins on less than 50% of the vote may be just as loathed by the majority of their constituents as (we can guess that) these BNP councillors were. But FPTP doesn't check up on this by probing the views of the split majority. AV resolves this problem by eliminating trailing candidates one at a time, bringing split votes for similar candidates back together, and identifying a majority consensus. Put simply, it is a way of double-checking that the result which we would have had under FPTP really reflects what the majority want.
There's another issue here, too. One thing which the council seat examples above have in common is that each seat was won by a BNP candidate standing in that ward for the first time. (You can tell because the percentage swing to the BNP in each case is identical to their total share of the vote). This effectively means that the non-BNP-supporting voters in these wards were taken by surprise. They must have seen a BNP candidate on their ballot-papers, but they didn't know how many other voters were going to support them. If they had known that the BNP candidate stood a real chance of winning, a lot of them would probably have switched their support from the candidate they really supported to another candidate whom they thought would have a better chance of defeating the BNP. That is, they would have voted tactically.
UK voters have got so used to employing tactical voting to compensate for FPTP's flaws that we practically think of it as part of the system. (Page 9 of this paper shows that 23.6% of people vote tactically when they believe it is an appropriate strategy). But it's a very imperfect fix for the split votes problem. For it to work properly, all voters would have to have full information about the intentions of their fellow voters. The BNP candidates standing for the first time are a particularly extreme example of why this isn't always the case. But of course no voter ever has a really complete picture of how other people are going to vote when they cast their own ballot.
What's more, when people vote tactically, they are actually defining for themselves a hierarchy of at least three distinct preferences:
- The candidate they really want to win, but don't believe can do so.
- The candidate they don’t like as much as number 1, but whom they think has a real chance of winning.
- The candidate they can't stand and would take almost any other option rather than see them win.
In other words, a tactical vote under FPTP is essentially a second preference vote, based on the voter's (probably imperfect) understanding of other people's voting intentions. But AV can take the tactical voter's preferences into account formally and systematically, and in the light of a full and perfect knowledge of all other voters' similar hierarchies.
You can't decide to whether vote for the candidate you really support or the one you think can actually win? Maybe you don't even know how your favourite candidate is really going to perform, anyway. What if they could beat your most hated candidate – if only you dared to vote for them? And just how many people in your constituency are planning to vote for the Absolutely Loathsome Party, anyway?
None of this is a problem under AV. The AV system automatically takes account of your first preference for as long as that candidate remains a viable option. It then switches you to your second preference at exactly the point when other people's votes render your first preference non-viable. You don't have to second guess anybody - and even if a new candidate is standing, or an existing candidate performs surprisingly well or surprisingly badly, AV will still ensure that only a candidate who has demonstrated majority support can win.
So if you want a system which:
- Ensures that each constituency elects an MP whom the majority of the voters there really support.
- Prevents widely hated candidates from winning because the majority vote against them was split.
- Allows people to express their preferences honestly, without having to make (imperfect) guesses about how other people will vote.
- Deals with those preferences systematically and in the best interests of the individual voter, no matter what other voters do.
Click here if you would like view this entry in light text on a dark background.