Saturday, October 29, 2016

I have been reading Nate Silver in the run up to two presidential elections now. I think I can read him decently. He is as precise as he can be in his word choice, it seems to me, which, given his cautious word choice, sounds a little oxymoronic. Lord Tweedsmuir wrote of Henry Wilson: "He had a gift, too, of making a situation more clear than God intended it to be." Silver has no such "gift", lol. He will be as precise, which almost always means as vague, as God or his forecasting model intends the election situation to be. There is far less with Silver than with The New York Times, but still some, reading between the lines that is necessary. I think I can read between his lines decently, too.

Although Silver is admirably direct in expressing his concerns about the polls', and therefore his model's ability to forecast this election outcome, I detect more concern still. Take these excerpts from his article today. It is the only thing of Silver's that I have had to read 2-3 times.

I’ve heard from people who wonder whether Friday’s news – that FBI director James Comey was investigating additional emails that may be pertinent to Hillary Clinton’s private email server — might have come too late in the campaign to be reflected in the polls, and therefore in our forecast, before Election Day. While the situation isn’t ideal, there’s probably just enough time left to measure the initial impact. 
...we should have a pretty good read on the initial reaction to the news by the middle of next week.

But while I’m not that worried about the model having enough time to account for the reaction to the FBI news, I am worried about whether it will capture the reaction to the reaction as the story continues to develop.

The only word in those brief passages that has Silver's original emphasis is the underlined "just." The rest are mine.

"Probably," more than 50 %, not certainly, "just enough time TO MEASURE THE INITIAL IMPACT," NOT THE "REACTION TO THE REACTION." He is "worried" that the polls, ergo his model, will not get "the reaction to the reaction" and that therefore his forecast will be WRONG!

For the record, Nate Silver's 538 has, at this writing, HRC +5.5 on Trump. I have more faith in his weighted aggregation of the polls than in any other. There will be more writing in the next ten days. Will the writing on the eleventh validate Silver's final forecast? He is "worried" about that. I am worried that the writing on the eleventh day announces President-elect Donald J. Trump whatever Silver's final forecast is.