Prime Minister Theresa May speaking at an event at The Space in Norwich while on the General Election campaign trail. PRESS ASSOCIATION Photo. Picture date: Wednesday June 7, 2017. See PA story ELECTION Main. Photo credit should read: Stefan Rousseau/PA Wire Image Credit: AP

General election campaigns don’t trundle along a set path. They never cruise punctually on to the forecourt of their predicted destination. Instead, they pitch and lurch, soar and sink. All that is a given. But the 2017 campaign is surely the most bewildering ever.

Nothing, at all, is following a script. Terrible events such as the London Bridge and Manchester Arena attacks bring campaigning to a halt. Pollsters are uncertain, anxiously explaining the variations in methodology that make their results so different from the others. YouGov has designed a model for predicting the state of the parties that shows the Tories losing 15 seats and their majority. A Survation poll has the Tories ahead by a single point. “Jez we can”, shout the posters. The shadow chancellor, John McDonnell, discusses the Labour manifesto with Whitehall officials. Not so fast — oh, not so fast.

Remember where you were on May 7, 2015? Remember how the final polls had the two main parties neck and neck, and then came that 10pm exit poll predicting Labour’s Scottish wipeout and putting the Tories just a few seats short of an overall majority? You can bet Ed Balls does, and Douglas Alexander — two people whose lives and careers were woven into the fabric of successive Labour administrations who lost their seats and their political futures that night. So do many of the 229 Labour MPs who held their seats, but whose dreams of returning to power after a single term of opposition were dashed. It’s the hope that’s so cruel.

The thing about a surge like the extraordinary wave that’s lifted Jeremy Corbyn off the floor is that merely being afloat can feel like flying. It’s not. The Labour leader has had a remarkable campaign. Even Tory voters think he’s done better than their team. He’s generated excitement and a sense of the possible that, if it lingers, could be transformative, at least for campaigning. It is always seductive to imagine that what is will somehow turn into what ought to be.

After all, the pollsters themselves acknowledge their uncertainty. The Guardian is not alone in treating its poll series from ICM (an 11-point Conservative lead in the penultimate survey) discreetly, rather than headlining the results week by week. Each pollster has a model of logic - as long as its basic assumptions are correct. Taken together they are a wild ride through thickets of differential turnouts and stratification analysis.

The US pollster Nate Silver has written a handy guide on his fivethirtyeight.com blog, giving seven different hypotheses for the polls’ inaccuracy in 2015 (and in 1992, the last time they predicted a Labour victory only for a Tory majority unfold) — ranging from lax turnout models to “herding pollsters”. It’s a timely reminder that polling is an art, not a science.

Here is a festival of opportunity for pattern spotters: local election vote share (in May, Tories 38 per cent, Labour 27 per cent) is a significant predictor of general election outcomes. Approval ratings for party leaders can be useful, except when they aren’t. Asking voters who they think will win, rather than how they will vote, is supposed to be highly accurate, but since it’s not often asked it’s hard to test.

All the same, in the campaign, the two parties have come closer together, in a way no one was expecting. The Tories are reported to be rattled. Yet consider this: The Tory vote share has declined, but no poll has put them below 40 per cent, nor yet put Labour above 40 per cent, although there is still time, just. Labour needs the 18-24-year-olds to come out and vote, but even if they do, there are many more over-65s. And the latest analysis from the National Centre for Social Research is sceptical about youth turnout and suggests that a significant group of former Labour voters now feel politically homeless.

And then there’s anecdote. Just because the wheels have fallen off the Tory campaign and Prime Minister Theresa May’s made-for-TV outings are a car crash doesn’t mean that things are a disaster away from the major cities. Instead, there is the kind of calm among party managers and candidates that is probably inspired by a certain confidence.

Watch where May’s battle bus takes her, ploughing down the motorways to Labour-held marginal after Labour-held marginal: To Bridgend and Stoke and Mansfield and Plymouth. Consider the spending that can’t be seen — the millions poured into daily polling aimed at teasing out possible Tory switchers. See the use of targeted attack ads. Check out the Ashcroft marginal-seat polls that still show a probable Tory majority of somewhere between 40 and 100 seats.

Like the campaign, the result will be a rollercoaster: Labour will do better in London than in 2015 (as may the Liberal Democrats). The Lib Dems will fail to make headway in their old heartlands in the south-west; and Nick Clegg, in Sheffield Hallam, is projected to lose to Labour (as it was in 2015). But the party may win St Albans from the Tories.

There is, however, no evidence that points to anything other than a Tory victory. On the polling average, they bob along seven points ahead. Labour Uncut has just done its second sobering analysis of canvass returns from Labour-held marginals, particularly in the north-east and north-west: There’s talk of a nuclear winter.

Sorry about the cold shower. Need something to keep your spirits up? Nate Silver advises that the most important thing to remember is this: Never trust conventional wisdom.

— Guardian News & Media Ltd

Anne Perkins has been a leader writer, lobby correspondent and feature writer for the Guardian since 1997.