Damien

Also whilst it's true most implementations of a standard random number algorithms in programming aren't truly random they pretty much are for the purposes we're talking about here. For picking from datasets, i.e a population, pseudo-random is fine. It's not going to be biased to a pre-determined outcome whereas if you were doing a coin-toss betting game there would be an advantage to knowing what seeded the random number generator (and what the generator was).

But polling clearly works. YouGov called the last election as Tories: 43%. They got 44%. Labour: 33%. They got 34%.

It's not precise. There are limitations. But it's not bad and it's the best we have outside of elections. It's certainly better than our own biases, political commentators or anecdotal evidence from our social circles.

If Remainers listened to polling rather than Twitter the Brexit vote wouldn't be such a shock. If Labour had listened to the polls rather than assume everyone loves Corbyn they would have gotten rid of him before December 2019.

Failed election campaigns almost always a mantra into why the polls are wrong.