What if Sam Wang is right and Silver is wrong
I'm not sure that's possible. It's impossible to conclude on the basis of the results which prediction was 'correct' - theoretically, if Trump wins we can't say that Wang was wrong. In such a case, he's just assigned a very low probability that happened to eventuate. If you roll a die and say "it probably won't be a 6", you're not wrong if it ends up being a 6.
So if Clinton wins, although I would expect Silver to cop flack for having her at lower odds, it won't be possible to say that he was wrong. It might be possible to say the polls upon which he was basing his model were generally wrong if Trump wins or if Clinton wins by a significant margin, but he's just assigning probabilities to the chance that the polls are incorrect.
The problem for him at the moment is likely to be that he's going to have to come out with an article which 'calls' every state. At the moment, Nevada, Florida & North Carolina all show Trump marginally ahead, and Silver will have to say "my model shows Trump as more likely to win". Any slight deviation in the polls and he'll have blown three states. No-one will care that the three states going to Clinton doesn't actually make Silver's prediction 'wrong'.
On the other hand, I do actually think we can say that Wang's prediction is wrong, regardless of the result. It does not strike me as epistemologically plausible to assign from the current polls anything higher than, say, an 80% probability. If I randomly say that there's a 100% probability that Trump will win, and he subsequently does, I'm not some wise sage - I'm wrong regardless of the result, because that probability is simply not possible with the current information we have.