It's time to ditch polling averages that don't also show uncertainty

The polls are tight in Georgia — and that's all we can really know

AFTER the election, pollsters and savvy (ie, empirical) political journalists spent a lot of time arguing that polls weren't actually THAT, wrong, but rather, our expectations for them have been set far too high. It seemed that election handicappers and forecasters might have even come around to the idea that this was partly our fault for not constantly emphasizing the inherent errors in polling and forecasting. Maybe the next time around, we could do so properly.

But now, with nearly two months of new election commentary available for analysis, the evidence suggests that that was too optimistic a view. Forecasters have apparently fallen into their same old habits of underplaying uncertainty in the polls.

If you navigate over to FiveThirtyEight's home page right now, you'll see what I mean. Their polling average for the upcoming US Senate runoffs in Georgia are shown in a sidebar on the right side of the page, consisting of lines for each candidate and predicted vote shares — to one d…