There are five days to go, but even the best coverage of the US presidential election cannot give us any sense of which way things will go. If you believe the polls, the race is a dead heat. If you believe the so-called prediction models, Donald Trump is slightly more likely to win than Kamala Harris.

I believe neither. I decided to treat polls as uninformative after the 2022 midterm elections, where many people whose judgment on US politics I trust more than mine took the polls to show a “red wave”. It didn’t happen, and I have seen no totally convincing explanation as to why that would make me trust US political polls again. (My own attempt to make sense of this concluded that not just abortion, but the economy counted in Democrats’ favour — on which more below.) The 2022 failure came on top of the poll misses in 2016 and 2020.

Not that I’m less of a poll junkie than the next journalist. Polls are captivating in the way that another hit of your favourite drug is, as my colleague Oliver Roeder suggests in his absolute must-read long read on polling in last weekend’s FT. And, of course, pollsters have been thinking hard about how they may get closer to the actual result this time. But none of this makes me think it’s wise to think polls impart more information beyond the simple fact that we don’t know.