We’ve just had our pre-election meeting at The Spectator, and agreed the usual drill for the big night. Election day itself is dead: we relax and steel ourselves for the evening. There'll be the normal 8.30pm curry as we wait for the exit poll and we'll lay on some wine (and desk space) for contributors who'll be near our Westminster office. Katy Balls will stay up late – that’s how she likes it – and I’ll try to grab some sleep early and come into the office for 2.30am. Katy, James Forsyth and Tom Goodenough will do the night shift; Will Heaven, John O’Neill and I will do the early morning shift. We’ll record a podcast after the 10pm exit poll, which we probably won't need to revise because they're usually right. Our plan is based around the consensus result: a comfortable (c70-seat) win for Theresa May.
And if, heaven forbid, there's a hung parliament or Jeremy Corbyn triumphs then we have a contingency plan. Katy’s night shift becomes a night-and-day shift, same with James. Someone will wake me up at when the exit poll is released and I’ll come straight in. The chances of us needing to deploy this scenario? Tiny. About 10pc.
This is our well-established drill. It never works. Every time, I get the midnight phone call, saying: some weird results are coming through and it actually looks like Cameron will get a majority/Brexit will edge it/Trump will get in. Then it's time to tear up the old plan and deploy the emergency one. The plan that we had thought there was only a 10pc chance of actually needing has, each time, become the plan that we ended up deploying. Given how often the unexpected has happened, I find it hard to understand why everyone is so confident now that Theresa May is going to walk this.
Let's just remind ourselves of the US election scenario. Never had an election been more extensively polled; never had there been so much political data collated and dissected by so many experts. This is the Nate Silver era; we've done a Newton on public opinion, we have the technology and algorithms to predict the future to a high degree of accuracy. The kit is so damn good we can go better than polls and give you a percentage chance: Hillary had an 80pc chance of winning. (Experts at Princeton University put it at 99pc.) Here's how the New York Times predictor looked as the evening went on...
Giving Trump a 20pc chance was more than the FT/Populus ‘predictor’ did. In the the 2015 general election, it gave Cameron a 0.2pc chance of victory...
And then the bookmakers' odds over Brexit...
The pollsters got blamed for missing Trump but their data showed that Trump had a real chance; it was the experts who dismissed this scenario out of hand. Just as Corbyn is dismissed out of hand now.The polls put Brexit ahead more often than Remain, and were always close, but not a single commentator predicted Brexit. Usually, the great 'surprise' is not the fault of data - it's the fault of those who extrapolate data, and use their past experience as a guide to the future (aka 'gut feeling'). And this time, yet again, commentators' opinion about the 2017 general election has a solidity that is simply not reflected in the polls that we have seen.
Wise is he who knows he knows not, goes the old Arab proverb, and the same is true for pundits. On the evening of the 2015 general election James Forsyth, Andrew Neil and I were at a dinner for the Addison Club, a Spectator offshoot where a group of people from politics, business and the arts get together. We had a guest speaker, a pollster who revealed his firm's final analysis: it was a monster poll, five times the normal sample size and (ergo) an error margin of 1pc. It was impossible, he concluded, for Cameron to win.
Andrew was then asked what he thought would happen. ‘How would I know?’ he replied. ‘I'm not Mystic Meg.’ He was teased for this: what a cop-out answer! Surely with all his experience, he has a feel for what would happen? Andrew replied his experience gave him a feel for when anything could happen – and this was one of those times.
Am I saying I think Corbyn will win? Nope. But I've overseen enough 'emergency scenario' election night plans in the last few years to make me doubt if we journalists really can predict the future. My guide is not YouGov but Karl Popper, who held up to ridicule the "historicism," or the notion beloved of political scientists that history runs on tram lines.
And if the last three years have taught us anything, they’ve taught us this: nothing, in politics, is so crazy that it couldn’t actually happen. Let’s just hope that a Corbyn premiership isn’t one of them.