Hat Tip to Stephen Tall's blog for pointing to this cartoon and article which explains in a humorous way why rules based on precedent to predict the results of an election cannot be relied on.
The trouble is that in the run up to every election you can always find something which has never happened before which would have to happen for any given candidate or party to win.
Sometimes the argument by precedent is a fairly powerful one - because what would have to happen for a party to win is genuinely unlikely. An example would this article on "Labour Uncut" which suggests that Labour's poll lead is a lot less solid than those people who are betting on a Labour win at the next election think, points out that in recent years no opposition has gone on to form a government after the following general election without being at least six points ahead two years before the election. They also argue that to have a good chance of winning the opposition has to be twelve points clear. This is because there has nearly always been a swing back from the opposition to the incumbent in the run up to an election.
Sometimes the argument from precedent is obviously contrived and ludicrous (e.g. "nobody with two middle names can get elected"). But sometimes an argument which appears strong at first sight (e.g. nobody can get ten million votes and still lose) is a lot weaker than it appears.
(What happens if someone else gets twelve million votes?)
At the end of the day the voters can pick whoever they choose, and the fact that they don't always do what the pundits expect is a good thing.