There's something fundamental I don't understand about election odds predictions like @FiveThirtyEight... 1/N

Are the odds based on uncertainty about conditions right now, due to polling error and lack of uncertainty? 2/N

Or are they also supposed to cover the probability of future events changing the outcome of the election?

The first seems like a data+math problem. The second like a risk management+sociology problem. 4/N

But when I see sites talk about tuning/correcting the model, I don't see those discussed separately. @NateSilver538 and @Chris_arnade help?