fivethirtyeight.com is a website, perhaps categorized as a blog, that follows politics, with a focus on polls and the math and statistics of them. The name refers to the five hundred and thirty eight votes in the electoral college. While the math is non-partisan, the people who put together the site are admittedly in favor of Barack Obama. The level of bias towards him they have causes a good amount of commentary on the threads attached to the posts. Along with presidential elections, he also is following senate races, but not any other contests.
What is different about the site, and probably the reason for its success, is the focus on the analysis of data. This is especially important because major news sources often give maps of electoral votes that declare states "toss-up" without explaining why. There has been some suspicion that some news outfits are trying to make the race appear closer than it is so that they will continue to attract interest and ratings. But fivethirtyeight is beholden to no one. The site was founded by Nate Silver, who apparently was also a well-respected expert in the field of baseball statistics. He started the site anonymously, but eventually unmasked himself as an already famous (in the field) baseball guru. New York magazine described the surprise this would cause amongst baseball followers as "a high-fashion Website turned out to be Howard Cosell." What is more surprising is that Nate Silver, along with his collaborators Sean Quinn and Brett Marty, appear to be doing very well, reportedly becoming the number one site on google's blogger service, and a site referenced in the mainstream media.
Not that he has done anything specifically innovative...he just does what he is doing very well, and explains most of his methodology. He takes all polls that are not commissioned by a partisan group for its own uses, and then averages them together. Based on prior reliability of a pollster, he weighs their polls more or less highly. Old polls are also not discarded, they are simply weighed less and less. They are then combined with national polls and trendlines to give a trend prediction. The polls are then all ran 10,000 times, and each state is given an average win/loss percentage. The overall results are then displayed in a number of graphs and maps. Along with that, posts are made on the daily polls, as well as on other topics related to the election, some serious, and some being rather speculative or almost silly. The site is required reading for me in the morning, and many of its fans refresh constantly, looking for updates, especially on the daily poll summary.
Other than learning a fair amount about politics, I have learned a fair amount about math, and learned even more about other people's ignorance of math. This mostly comes from commenters on the threads (see below), but also from other people I have talked to, who will throw out a random poll they read about in a newspaper article several weeks ago as proof that "X candidate could never win Y state". I used to assume these people were being stubbornly partisan (which they were), but I also realize that they really don't know what things like "margin of error", "sample bias" and "outliers" are. It has made me realize how important understanding statistics is to critical thinking, something people can always use more of.
The site has two major problems. First, it was started as the site of a math geek pontificating, for his own amusement, upon electoral politics, not as an interactive blog. Thus, the reader response part is not well developed. There is no moderation, no nesting, and blogger only shows the first 200 posts (without pulling a trick). Since the site has become popular, there have been hundreds, and sometimes over a thousand, comments on each post. They are full of redundancy and trolling. Some of which is amusing, but most of which is annoying and stupid. It is too late in the game to fix this now. Which leads us to another problem: in two weeks, the site will have lost its raison d' etre, as the election is finally settled and we see how the model worked. At that point (or shortly after), what will become of the site? Will it become dormant for another four years? Will it turn into a normal political blog? Will Nate Silver move on to a high paying job in baseball or politics? What site will we obsessively hit "refresh" on until the next horse race? That is just one of the things we will know about in two short weeks.