Nate Silver’s website

Someone who wishes to remain anonymous writes:

I believe you are aware that Nate Silver spoke at last year’s JSM and that he began a publication under ESPN (http://fivethirtyeight.com/). Do you have any opinions on the publication? Maybe some you wish to share with the public. I was hoping to hear your opinions about 538 in particular and data-driven reporting in general.

My reply: I have no real organized thoughts on this yet. I suspect that Nate was getting overwhelmed by the implicit demands on him, in which he was expected to write about and perform original and accurate analysis on a range of topics as they arose, also he was being put into a no-win situation regarding future election analyses. He predicted 50 states out of 50 in the 2012 election. What comes next, predicting 52 states? So I think that for his own sanity he had to get out of this trap. Setting up a site where he is the editor rather than the (nearly) sole author: that seems like a logical step. I hope it goes well.

19 thoughts on “Nate Silver’s website

  1. Most people do not know that for the 2012 election Nate Silver was wrong on two of the U.S. Senate races: Montana and North Dakota where he predicted the Republicans would win when in fact the Democrats came out on top. Most likely the reason for his wrong predictions was the lack of polling in these sparsely populated states.
    With regard to the presidential race, it is not clear what Silver predicted for Florida. He forecast 332 electoral votes for Obama and 203 [misprint, should be 206] for Romney which turns out to be the exact result. However, on Silver’s blog itself, Obama got only 313 electoral votes and Romney got 225. Perhaps because Florida was so very close in the polls, he merely split Florida’s 29 electoral votes.

  2. One of the early articles on his blog was all about “decoding” health care news by assigning a number to your “gut feeling” about the claim in question, then to the study, then multiplying for an absurd result. (it was by Jeff Leek but I’m guessing it was Silver’s idea). http://fivethirtyeight.com/features/a-formula-for-decoding-health-news/ I find the science articles bland, quite superficial, and boring–very disappointing, I expected more. I don’t see how it can possibly compete with other science news sources on-line. Perhaps it should stick to sports and politics. I have a post on draft on 538 that I might post some day.

    • Actually, if you had looked closely enough you’d see:
      “Jeff Leek is associate professor of biostatistics and oncology at Johns Hopkins Bloomberg School of Public Health and co-director of the Johns Hopkins Specialization in Data Science. He writes for the blog Simply Statistics.”

      So I doubt it was “Silver’s idea”. I thought the point of that article was to put some rough numbers to one’s prior beliefs about headlines and how the evidence presented could be used to update those beliefs. What’s wrong with that? Labeling a back-of-the-napkin calculation “absurd” is a bit extreme.

      538 has material contributed by academics aside from that by the full-time writers and editors. The content varies (and sometimes the quality), but so do readers’ interests. Their sports analyses are excellent and overall the site fills an interesting niche. The real fun starts when readers from ESPN start trolling the comments because the statistics don’t match their personal perceptions. But 538’s material certainly adds weight to pub debates.

      • Dan: I know who he is, and that’s why I said what I did. Read the article and see what you think–it was severely criticized by numerous others. I’ve no reason to doubt that 538’s material certainly adds weight to pub debates”. I actually pondered for awhile what i thought was really missing from the attempt at thoughtful data-driven journalism, or whatever they’re calling it these days. I’m sure he wants something deeper. I came up with a short paper that delineates specific ways to make it work. I never shared them, but will at some point.

        • Mayo, the website told me it had 0 comments until I clicked the expander. Sure enough there were some gems. I see now that Jeff’s explanation of the Bayes factor was a bit confusing and incorrect, and as one commenter pointed out, this can result from the intentional dumbing down of something that is relatively complicated.

        • It was in other forums that I heard alarmed gasps. Then there was the episode on the article on climate change costs, and the author having to leave 538: not sufficiently politically correct.

    • guessing it was Silver’s idea? That’s a bit presumptuous Jeff doesn’t need anyone to spoonfeed him article ideas on statistics. Put your biases aside please.

  3. I also think he made a good move. FiveThirtyEight isn’t dead though. He has a staff now, which in my mind relieves the pressure cooker. Being sole author, analyst, and app developer for the site could not be less of a 24 hr detail within the timelines he needed to work.

    Still, social/political analysis aside, he no doubt has a love for studying sports, and I’m sure given his past life making his living in the poker room, doesn’t shy away from a parlor game now and then.

    I can relate here as my own first interest in probability & stats came through playing way too much backgammon. I began that sojourn by studying day/night how to deal with positional decisions statistically from folks like Kit Woolsey and Paul Magriel. To this day the game is still quite a sickness with me, whether playing a speedy game with a few Turkish friends, a tourney, or worse, the internet :). I can see why he would ‘return migrate’ to sports/gaming stats.

    ESPN won, too, having picked up a great secret weapon for bookies everywhere :)

  4. The best material so far has been about sports because that’s where he comes from and perhaps is most fluent in the positives and negatives of each statistical decision and approach. I see in those articles more discussion of what this or that means or how the stat has limited meaning. Some of the material seems put out to be a take on an issue using a degree of analysis. I don’t think it pretends to be more.

  5. I think the statistics aspect of the site are so so. But they find really nice variables to tell a story. And the visuals are compelling, even if they are not as fancy and interactive as those you might see, say in the NYTimes. Yes, the best stuff is on sport. I don’t work in sports, and I work with fancier statistical tools, but some posts in fivethirtyeight have inspired me to find simple ways to tell a visual story with well chosen variables.

Leave a Reply to Steve Sailer Cancel reply

Your email address will not be published. Required fields are marked *