We'd have to see granular exit poll data to reach that conclusion. Based on aggregate vote counts, seems like the 2008 huge wave of volume is recurring.
538 certainly isn't perfect but they just model the inputs (polls) that they are given, along with their own algos. So I wouldn't say they're in on it. Earlier when ABC's gang interviewed Nate, it looked to me like he wasn't comfortable with the state of the election (i.e. either on a personal level, or because his probabilities haven't matched the results so far).
Hmm, there's a massive shift in NYT's needle for GA right now with 80% of precincts reporting. I don't know if that's a glitch. There was a ton of vote left to count around Atlanta metro but when I spot-checked it, I didn't think it was enough to close the gap.
Modeling key battleground states.
www.nytimes.com