This is to put the "538 was wrong" thing into perspective. The 538 "polls only" election eve page is at
https://projects.fivethirtyeight.com/20 ... -forecast/.
First of all, 45 of the 50 States went the way they were shaded based on the point estimates. As I've said before, by convention, what you are looking for when making a call in statistics is 95% confidence. There was not 95% confidence in ANY of the 5 States that didn't go the way they were shaded.
Secondly, there is the overall "percent chance" projection. That's a confidence level. It's 71.4% for Clinton winning. That is not even CLOSE to what would be required for making a decision in statistics. A drug company that went to FDA with data saying their drug would have an effect with 71.4% confidence would be laughed out of the office. No way.
The idea that 538 "got it wrong" or made a "call" that Hillary would win based on that page is just ridiculous. And saying 538 should be discredited because Trump won the electoral college is saying that college football oddsmakers should be discredited because Clemson beat Alabama in the following 2017 College football championship game.
Saying that there is 71.4% confidence that something will happen then having it not happen is no big deal. One instance of it not happening doesn't discredit anything at all.