Building predictive models for politics relies on understanding that polls are imperfect instruments and calculating the probability of outcomes relative to the consensus.

“So essentially the model is handicapping what is a chance that the polling is wrong by enough to cause a change in the result right.”

He argues that while Trump won, his victory does not validate his policies and that those who believe in democratic values should continue to fight for them.

“Am I allowed to say I was wrong to think that she would win but I'm still right to think she should have won and that her values were better than Trump's and her ideas are better than trumps and she'd be better for the world than Trump that the fact that he he won doesn't suddenly make him right no”