Pro-Brexit demonstrators protest outside the Supreme Court in London. File/Reuters

John Rentoul, The Independent

There is nothing to stop Boris Johnson slicing through Labour Leave seats like a hot knife through butter — according to an opinion poll in Workington, that is. It found that if they were voting “tomorrow”, the voters of the Cumbrian constituency would replace their Labour MP with a Conservative one, on a swing of 10 per cent.

It is true that constituency polls have a poor record of predicting results, and that the election campaign has hardly started yet. I remember voters on the doorstep early on in the last election saying how much they admired Theresa May.

Yet Survation’s result is supported by the only attempt to do seat-by-seat modelling so far. The great polling sensation of the last election was YouGov’s MRP model, which predicted a hung parliament when all the conventional pollsters (bar Survation) suggested a comfortable majority for the Tories. The problem with MRP modelling (it stands for “multilevel regression with poststratification”) is that it is 25 times as expensive as normal polling; requiring a sample of about 50,000 people instead of the usual 2,000. So while we won’t see many MRP polls in this election, another has already been produced, by the anti-Brexit tactical voting campaign, Get Voting. Using a sample collected in September and October, this poll, too, suggests the Tories are ahead in Workington, if only by a few points.

The significance of the Workington polling is that it suggests Johnson could succeed where May failed. “I am interested in these Labour seats that Johnson has to win if he is to get his majority,” one veteran of the last campaign told me. “The list will be similar to the one Team May drew up in 2017, where they made zero progress.”

The conventional wisdom is that Johnson is a southern Tory toff who will never break through in such places — but this may be based on southern Remainer assumptions about what a terrible person the prime minister is.

I certainly assumed Johnson had lost the cross-party appeal that made him a two-time mayoral winner in a Labour city, and it is true that young Remainers have gone right off him since the referendum. But this polling suggests that he still appeals across party lines among the Leave half of the population.

Even more than in 2017, this election will be fought between the Tory Leavers and Labour Remainers. This time the Tories are led by a committed Leaver, who has what he calls an “oven-ready” Brexit deal, while Labour is offering a referendum, which is the only realistic way to stop Brexit.

The question for this campaign is which side can add most and subtract least from its core Brexit proposition. Johnson is running a more “Labour” campaign than May did last time, and he has to keep Nigel Farage at bay. Corbyn is running the same campaign as last time, with even more spending pledges, and he has to fend off Jo Swinson. So far, Johnson is more successful.

The ban on fracking shows the prime minister is not going to be trapped, as May was with hunting, into defending unpopular “legacy” policies. His repetition of spending promises on the NHS, schools and police makes him competitive with Labour on that party’s favoured ground, and polls do indeed suggest Johnson is level with Corbyn on the question of which leader would be “best” on the NHS. Admittedly, May, too, was more trusted on the NHS than Corbyn at the start of the last election campaign — but I suspect Johnson is less likely to lose ground on this issue. It is a small thing, but someone who lives in the Middle East this week asked one of my colleagues why, in TV coverage of the UK, the prime minister is always in hospitals.

Relying on Johnson to make as bad a mistake as the dementia tax or to campaign as terribly as May — “a wooden spoon in a trouser suit”, as Matt Chorley of The Times called her — is not an adequate campaign strategy for Corbyn.

The polling in Workington suggests that Johnson can make the big inroads in Labour heartlands where May tried and failed.