One thing about 2016 is it wasn't just that our model gave Trump a better chance than other models/conventional wisdom, but also that we really tried to lean into that with our headlines, on social media, etc. We were quite adamant about pointing out that Trump had a decent shot.

We actually worried toward the end that we were overdoing it to some degree—not with the model itself, but that we tended to focus a lot more on Trump 30% chance than Clinton's 70%. And I think we probably would have gotten some crap for it if Clinton had won.

It's weird to live through an experience—that we were sort of waving our arms around and saying "he's got a shot!" when very few people were doing that—and have a lot of people remember it exactly the opposite way. Especially in the internet era when you can look all of this up.