At some point, a predictor must predict.
My friend Dave Winer suggests that the polling analysis website FiveThirtyEight.com is engaged in some link-spamming effort with an article carrying the headline, “Trump is just a normal polling error behind Clinton.”
I’ve been fascinated with the challenges Nate Silver & Co. have faced in trying to convince an audience to stay engaged in a process that has been like a year-long version of the two weeks leading up to the Super Bowl.
In addition to trying of make the presidential election a series of daily swings (who’s winning in Hillsborough County, Fla., today?), Silver and those sites that have tried to mimic his approach, all flunked the Republican Primary race–caution now permeates each article and podcast. Every few days, Silver runs a story on why anything that sounds definitive should be taken with a grain of salt.
Silver’s mea culpa after the Republican nominee set the stage for his summer and fall of providing endless butt-covering explanations of the many ways Trump could win, despite having only a 30% chance of doing so.
Silver’s most difficult challenge recently has been crafting new caveats.
But at some point, a predictor must predict.