Mashable, in this instance, is just rushed and plainly false blog-spam. According to them, FiveThirtyEight "accounted for 20% of all visits to the New York Times website," probably implying page views. The original article says 20% of visitors to the New York Times website also visited the FiveThirtyEight blog.
Man he's hated by a lot of the media. Whole diatribes are being written about how it's impossible to boil an election down to numbers on a spreadsheet. It's crazy that these are the same people that ask us to respect their opinions. Not even a basic understanding of mathematical thinking.
By my scoring, it's something like Silver 1, pundits -75.
If they had just kept quiet, it would have been easy to treat them as providing a different take. But their eagerness to beat on Silver was overreach. I can only hope that some of them will have to find honest work after this.
As a counterpoint, there is of lot of knee jerk rejection of any criticism of Nate Silver. A lot of people protray anything that questions Nate Silver's work as rejection of math.
Well, if you just reject his conclusions without refuting any of his methods, you haven't really made any argument against him. Those rejections are pretty much all rejections of math in favor of wishful thinking. Mere intuition can never be an effective argument against quantitative analysis.
FiveThirtyEight is actually leased by The New York Times.
Nate, brand, and all can leave the New York Times in the middle of 2013 once his contract is up.
Depending on the election results, I wouldn't be surprised to see him ask for a whole lot more money, jump to a new highest bidder, or go independent again.
He's actually said that he doesn't intend to be doing this again in four years time, but that FiveThirtyEight (the algorithm/brand) may continue without him.
Judging from what he wrote about poker in his book, doesn't seem likely - talked about how much the game has changed and how he couldn't win with his old strategies/style any more.
From the book I get the feeling he'd like to use his forecasting skills in something that is a bit more demanding, he chalks up baseball and political forecasts as falling in the 80/20 pareto rule.
Tangential: If you believe in Nate Silver, Intrade is underpricing the Barack Obama contract by 20%+.
The contract that pays out $10 if Obama wins is currently trading at $6.70 per contract, implying a 67% chance Obama wins vs Nate Silver's 90%+.
Unfortunately, Intrade and American gambling regulations don't play nice so they don't accept US transactions. They also won't write you a margin account so you actually need to have funds in their system to execute a trade today.
Hard to say which party is inaccurately estimating the outcome, but given the regulatory hurdles around Intrade and the seemingly low volumes, I wouldn't be surprised if they're off.
SkyBet and Ladbrokers for example both have payouts closer to 538's forecast.
What is peculiar is that Intrade is the outlier. In a liquid market this shouldn't happen because it allows for risk-free payouts. So something seems a bit odd about Intrade.
It's possible the campaigns have caught on to Intrade's impact on the media and will now step in when prices fall to much. </conspiracy>
It isn't entirely liquid. It's a pain to get money into Intrade (esp. for an American).
IEM (http://iemweb.biz.uiowa.edu/quotes/Pres12_quotes.html) and Intrade were up until yesterday off by 10%. Eventually things moved closer, but regulatory hurdles in actually getting money in (+ limits on funding) really hurt arbitrage opportunities.
It's really not difficult at all to get money into Intrade as an American. I had wires show up day-of if I did it early enough or next-day in every other case.
Also of note is that other betting markets were consistently higher than Intrade was for all of the last 6 months basically- there are a lot of theories as to why this may be, but it's regularly been 5%+ lower on Obama.
I am still sad that fivethirtyeight got swallowed by NYT. fivethirtyeight has a much better reputation in my mind, I would have loved to see it grow into its own powerhouse.
Maybe they would have done better on their own, but I was much happier to see them get swallowed by the NYT rather than a site like the Huffington Post, where the association would lose them much more goodwill.
The things that annoy me with the NYT are the monthly article limit that I have to workaround and how painfully slow 538 is on my iPhone.
Yeah, it's such a terrible limitation too when all you need to do is remove the variable, and it goes back to normal. I wonder how many customers that inconvenience actually gets for the NYT?
Regardless, they've landed more than 500k digital subscribers in the past 18 months. I'm sure there are a number of things at play here (bundling home delivery and digital, including app subs in these numbers, etc), but that's still pretty impressive.
And now we know why: Nate Silver correctly predicted every state in the presidential election (Electoral College). So much for the pundits who criticized him, Rasmussen's party ID results that supposedly showed Republicans with a commanding lead in Republican party identification, and "UnSkewed" polls.
The accuracy of Nate and other quantitative modelers' forecasts depends on the quality and accuracy of the polls they use. Drew Linzer of votamatic.org has written some really good pieces the past few days examining poll accuracy and the potential for "systemic polling failure":
Nate's model accounts for presumed inaccuracy in the polls but it is partly reliant in those polls' historical leanings. I think it would be more correct to say that his model depends on the quantity and consistency of polls.
I don't really buy all this. 538 is a great resource and Nate Silver is a smart guy doing interesting things with data, but millions and millions of people were going to hit the NYT for election stuff regardless and 538 is the site's big election brand right now. It's great for him and the times, but not really remarkable.
Second (as is clear from source article but, once more): via https://twitter.com/JanetCoats/status/265925450696642561
@fivethirtyeight did not account for 20% of NYT traffic; 20% of visitors to NYT looked at 538. Big difference, my math-challenged friends.