Getting the expectations game wrong
I went away at the weekend to celebrate my dad's 69th birthday (which puts him in the right age bracket to be a Labour by-election candidate nowadays) and therefore missed the opportunity to react to Labour hitting 40% in the opinion polls and Tory Ealing candidate Tony Lit turning out to be even less partisan than anyone realised.
It strikes me that the Tories have not learnt two important things about by-elections:
1) The degree of scrutiny your candidate will come under in a by-election means you need a ruthless panel process to weed out, for instance, folk who just attended the other party's fundraising dinner, had their photo taken with the other party's leader and made a large donation to do so. It doesn't mean that you headhunt people with local status without checking their background. Labour learnt all about the need for really hardline selection procedures in by-elections the hard way, after Deirdre Wood lost us the Greenwich by-election and Bob Gillespie lost us the Govan by-election. In by-elections a dull candidate who offers no hostages to fortune is preferable to a controversial one.
2) Managing expectations is critical. If the Tories had said they hoped to move from third to second in a safe Labour seat, they might be on track to match that expectation and get good coverage. But they've implied they could win and sent Cameron down to the seat 4 times so that success or failure is linked to him personally. That's created a situation where as long as Labour just holds on, it will be portrayed as a Tory failure to match expectations.
David Cameron has gambled the credibility of his leadership by associating himself so personally with this by-election.
If Lit loses on Thursday - which I think he will - the vultures on the David Davis wing of the Tories will be circling. And that could make an autumn general election very tempting - how about Thursday 27 September as polling day so that Labour benefits from the PM's speech to conference on Monday 24th, then scraps the rest of the event, and the Tory conference is after polling day? Just an idea.
UPDATE: someone has correctly commented that my idea above would fall foul of media coverage impartiality rules. So maybe another scenario is that GB announces the date of the election during conference week, with a four week campaign ending on 25 Oct?