I can understand models not figuring out the early vote issue yet. The punditry and writing ignoring it is far less forgivable. 538 was supposed to be more than just Nate's model, but it's really not outside of a handful of Enten articles.
I totally agree with this. The models can be forgiven for not incorporating the early votes. Silver's punditry in the face of the data (again!) is unforgivable.