If you need a break from UK election politics, here's a cool metric I came up with to measure how settled or unsettled the polls are in the primary.
I call it the "diffusion index" (better name coming later). Here's how it works. Take all candidates' vote shares, square them, sum the squares, then subtract the result from 1. An index of 1 means the race is extremely unsettled whereas 0 means one candidate has 100%.
As you might gather, this year's D race is relatively unsettled, so has high diffusion. Here it stands in comparison to other races at the same point in time (57 days before Iowa). It's most similar to the '08, '12 and '16 GOP races, and the '88 and '04 Democratic ones. 
Higher diffusion tends to predict higher polling volatility. That is to say, when the race is unsettled, you'll typically get bigger and more unpredictable polling swings. So this is somewhat good news for candidates who might not like what the polls say lately.
