If there hadn't been a (not actually all that large!) polling error in 2016, you can imagine the conventional wisdom being *incredibly* certain right now that Trump was going to lose, even more so than four years ago.

Are people overcompensating in the other direction? Probably a bit. But here's the thing. Unless you're interested in gambling/investing or forecast calibration, I'd argue that say a 10 or 15% chance of Trump winning isn't that different from say a 35% chance in practical terms.

On the other hand, a 10 or 15% chance is *really* different than 0%, both mathematically (1 in 8 is a lot different than 1 in ∞) and practically. So if people's takeaway is that Trump still has a meaningful chance, that's the right takeaway.