Did you glance at his charts? He's saying a "rigorous updater" would have estimated Trump's chances at almost nothing all year long, until dramatically changing the estimate just before the conclusion. In contrast, 538 estimated Trump's chances as much higher all year.
He has a point that the chances of winning shouldn't shift so rapidly, but a rapidly moving gauge is a decent indicator of uncertainty.
Nassim Taleb has been calling BS on FiveThirtyEight's methods and models for months: https://twitter.com/nntaleb/status/762032883414556674
He's now gone as far as calling FiveThirtyEight "clueless" and posted further comparisons of his model with theirs: https://dl.dropboxusercontent.com/u/50282823/binary%20foreca...