RevStu has suggested today that if the Labour party splits in two following its leadership election, it could make it easier for the Tories to be defeated at the next general election, with a new "Real Labour" party able to motivate the "missing millions" of left-wing people to turn out to vote for the first time in decades, and a centrist "Middle Labour" able to take on the Tories in the marginals. I have to say I disagree. Firstly, it's open to question whether it's even possible to win by increasing turnout - there's always the danger that your success in motivating your own lapsed supporters will be noticed by the normally apathetic on the other side of the fence, who will be sufficiently scared to turn out to vote themselves. That appears to be what happened in the independence referendum.
But even if, for the sake of argument, it proves to be possible to increase turnout on the left but not on the right, a split in Labour is still not a viable strategy for defeating the Tories, because the current electoral system punishes split oppositions. 1983 is the classic example - contrary to popular belief, Mrs Thatcher's share of the vote actually fell from 43.9% to 42.4%, but her parliamentary majority rocketed from a relatively modest 44 to an overwhelming 144. That happened simply because the left-wing vote, which had been relatively united behind Labour in 1979, ended up more or less evenly divided between Labour and the SDP-Liberal Alliance.
RevStu has cited Douglas Carswell's re-election in May to support his belief that it will prove possible in practice for incumbent MPs from both of the post-split Labour parties to hold their seats. But that ignores the fact that for every Carswell, there is likely to be ten Mark Recklesses. The vast majority of incumbent SDP MPs lost their seats in 1983, including the hugely popular Shirley Williams. That such a thing could happen even when the SDP and their allies were almost level-pegging with Labour in the popular vote shows how utterly unforgiving the electoral system can be. Carswell is a very special case, and indeed Clacton is a very unusual constituency.
The only theoretical hope would be some kind of electoral pact between the two parties, but that's almost certainly a non-runner. In 1981, there were hopes that the social democrats who had decided to stick with Labour might at least be able to stay on good terms with their ex-colleagues in the SDP, but the reality is that within months they could have cheerfully strangled each other. That's the nature of a split - and indeed the same thing happened to the SDP itself a few years later, when one faction merged with the Liberals, and another tried to plough on by itself. It's true that the Liberal Democrats eventually agreed not to put up candidates against Rosie Barnes and John Cartwright at the 1992 election, but that was practically an act of charity towards individuals who by then had seen their party cease to exist.
In any case, I think a split is only likely if Corbyn wins. If he loses, some of Labour's new members and registered supporters will be bitterly disappointed and may drift away from the party, but the hard-core of Socialist Campaign Group MPs will stick it out in their traditional futile manner. When fellow travellers of Corbyn have ended up in other parties, it's mostly because they were expelled. (Think Galloway.) As Diane Abbott has observed a number of times, it's the right that tends to walk out on Labour, not the left.
If a Corbyn victory leads to the creation of a "new SDP", I would imagine that the people behind it will be betting everything on their party being perceived as "the real opposition", and the rump Corbyn-led Labour swiftly withering away into irrelevance, much like Murphy-led Scottish Labour in May. I very much doubt that would happen, but it's the only way the electoral system can be beaten.