When I quit investment banking in search of daylight in 2014 I thought my life was going to be little easier crunching numbers for political campaigns. It wasn’t to be. Over the last few years, I’ve worked on the Scottish independence referendum in 2014, the 2015 general election, the Scottish Holyrood election in 2016, the EU referendum and the 2017 snap election. What I’ve never been able to wrap my head around through all these campaigns is why we’ve seen so many political upsets. Just why has the political consensus been wrong so often these past five years? When I worked on the Remain campaign, the upending of the consensus – against my own expectations – was a painful experience. Working for Ruth Davidson and her team in Scotland, the surprise result felt exciting and heady. Some political commentators have blamed innumerate polling companies and politically deaf journalists for failing to spot these upsets. These attacks have always struck me as wide of the mark, mistaking causation with correlation. On a plane last week I was struck by a scene which got me thinking about these political upsets. Bear with me. There were two groups of returning stag parties, each of about nine or ten people. The first group – let’s call them group A – had group booked all their seats together in three rows of three. They were riotously talking to each other throughout the whole flight, mostly oblivious to other passengers. The other stag group – group B – found themselves, due to inferior planning, split up from each other. This group found it hard to recount details of their weekend as random strangers sat awkwardly between them. So this less-organised gang ended up variously reading and nodding off; some of the group engaged in conversation with the strangers sitting next to them.