, , , , , , , , ,

Have a look at this picture (it comes from Nate Silver website here):


This image was on his web site on the 8th of November, meaning that his forecast was wrong right up to the last minutes of the election. Why? I can only speculate, but I base this speculation on election results in Poland and in Hungary and on recent successes of all populist parties in Europe. My speculation is this: all polls check what people say, not what people feel or really think. We live in times where all issues where politics have failed us (income inequality, immigrants, refugees, unemployment, lack of future) have become playing field for all kinds of populists. They know how people feel and use it to gain power – but have no real solutions to those problems. Look at Brexit. It was easy to play on people’s feelings, but after it was done, masters behind that referendum disappeared, leaving a big mess behind (here too, all sorts of forecasts showed Britain staying in the EU). I am sorry to say to all Americans, Trump will not leave. You are stuck with him for at least 4 years – unless you will impeach him for something. Which is not that unlikely.

Similarly I guess all the recent elections were lost to populists, and there is plenty to come, next one being Germany. Those parties promise money, or building walls, or whatever – and sometimes really keep their promises by paying for them with taxpayers money – and even though your reason tells you that they are not the best solution, your heart wins, because you feel in your heart that this is what you would like to happen, even if it logically cannot as there are no easy solutions to all current problems. And you vote for them. Lying at the same time to all polls you came across, so you don’t look stupid.

My advice to Nat Silver and all other guys (including the one from Stanford who ate a bug on national TV) is to add “feeling factor” to their algorithms. The more all issues are hot in terms of beliefs and feelings, the bigger impact that factor should have over cold data based on interviews. Eventually you will be able to adjust over people’s lies.