"On my great steed Sabermetric, I venture forth to distant shores."
July 22, 2013 12:06 PM Subscribe
Nate Silver will move FiveThirtyEight to ESPN when his contract with the New York Times expires in late August. Silver's new site will look to Bill Simmons' Grantland as a model for existing under ESPN's umbrella. His new move could be "genius," with a role at ABCNews and a larger audience, but did the New York Times know what it had in Silver? ESPN press release & Nate Silver 2.0 quote
Obviously, the first question to my mind upon hearing this was "Will 538 have full-content RSS feeds again?".
posted by These Premises Are Alarmed at 12:10 PM on July 22, 2013 [12 favorites]
posted by These Premises Are Alarmed at 12:10 PM on July 22, 2013 [12 favorites]
Sad as I will be to see him gone from the NYT, he makes more sense at ESPN. There are sports statistics stories every day of the year; the politics stuff really only captures people's imaginations every 2 to 4 years.
posted by yoink at 12:15 PM on July 22, 2013 [2 favorites]
posted by yoink at 12:15 PM on July 22, 2013 [2 favorites]
I hope the Times gets neuroscientist Sam Wang of the Princeton Election Consortium to replace Nate Silver. Wang didn't get the visibility Silver did during the last election even though his model seemed as good or better than Silver's in many respects.
Plus having a hard-ass, multiclassing statistician/neuroscientist in a pretty visible spotlight might help rehabilitate the public image of neuroscience in the wake of the voodoo correlations and the brain activity in dead salmon nonsense.
posted by logicpunk at 12:17 PM on July 22, 2013 [17 favorites]
Plus having a hard-ass, multiclassing statistician/neuroscientist in a pretty visible spotlight might help rehabilitate the public image of neuroscience in the wake of the voodoo correlations and the brain activity in dead salmon nonsense.
posted by logicpunk at 12:17 PM on July 22, 2013 [17 favorites]
Grantland is wonderful in that they are allowed _mostly_ free speech to say what they want. Simmons is probably privately wrist-slapped every time he makes oblique references to how much ESPN has turned into a tabloid with dumbasses like Stephen A Smith and Skip Bayliss allowed to open their mouths on TV ad nauseum. I do hope they give Silver a long leash, and don't force him to write a disproportionate amount of columns about Tiger Woods and Tim Tebow.
It will also be interesting to see what happens next election cycle. ESPN is owned by ABC/Disney, and damned if they're gonna let some geek use math to show something against their corporate interests. That corporation is far afield from NYT in that respect, to be sure.
posted by mcstayinskool at 12:19 PM on July 22, 2013 [2 favorites]
It will also be interesting to see what happens next election cycle. ESPN is owned by ABC/Disney, and damned if they're gonna let some geek use math to show something against their corporate interests. That corporation is far afield from NYT in that respect, to be sure.
posted by mcstayinskool at 12:19 PM on July 22, 2013 [2 favorites]
His entire probability-based way of looking at politics ran against the kind of political journalism that The Times specializes in: polling, the horse race, campaign coverage, analysis based on campaign-trail observation, and opinion writing, or “punditry,” as he put it, famously describing it as “fundamentally useless.”
God I hope this means he'll still be covering politics in 2014.
A number of traditional and well-respected Times journalists disliked his work. The first time I wrote about him I suggested that print readers should have the same access to his writing that online readers were getting. I was surprised to quickly hear by e-mail from three high-profile Times political journalists, criticizing him and his work. They were also tough on me for seeming to endorse what he wrote, since I was suggesting that it get more visibility.
God. The emphasis here though should be political journalists.
Many others, of course, in the Times newsroom did appreciate his work and the innovation (not to mention the traffic) that he brought, and liked his humility.
This speaks well of the rest of the Times.
The Times tried very hard to give him a lot of editorial help and a great platform. It bent over backward to do so, and this, too, disturbed some staff members. It was about to devote a significant number of staff positions to beefing up his presence into its own mini-department.
As does this. I hope they continue with this; you don't have to have Nate Silver on hand to use his techniques. I hope this signals the Times turning away from those political journalists mentioned above. Silver was the best thing about 2012, other than Romney being sent packing.
posted by JHarris at 12:20 PM on July 22, 2013 [5 favorites]
God I hope this means he'll still be covering politics in 2014.
A number of traditional and well-respected Times journalists disliked his work. The first time I wrote about him I suggested that print readers should have the same access to his writing that online readers were getting. I was surprised to quickly hear by e-mail from three high-profile Times political journalists, criticizing him and his work. They were also tough on me for seeming to endorse what he wrote, since I was suggesting that it get more visibility.
God. The emphasis here though should be political journalists.
Many others, of course, in the Times newsroom did appreciate his work and the innovation (not to mention the traffic) that he brought, and liked his humility.
This speaks well of the rest of the Times.
The Times tried very hard to give him a lot of editorial help and a great platform. It bent over backward to do so, and this, too, disturbed some staff members. It was about to devote a significant number of staff positions to beefing up his presence into its own mini-department.
As does this. I hope they continue with this; you don't have to have Nate Silver on hand to use his techniques. I hope this signals the Times turning away from those political journalists mentioned above. Silver was the best thing about 2012, other than Romney being sent packing.
posted by JHarris at 12:20 PM on July 22, 2013 [5 favorites]
From the NYT Public Editor's article:
I don’t think Nate Silver ever really fit into the Times culture and I think he was aware of that. He was, in a word, disruptive. Much like the Brad Pitt character in the movie “Moneyball” disrupted the old model of how to scout baseball players, Nate disrupted the traditional model of how to cover politics.
Disruptive, eh? Well we can't have any disruption of the NYT's political coverage now, can we?
posted by chavenet at 12:20 PM on July 22, 2013 [7 favorites]
I don’t think Nate Silver ever really fit into the Times culture and I think he was aware of that. He was, in a word, disruptive. Much like the Brad Pitt character in the movie “Moneyball” disrupted the old model of how to scout baseball players, Nate disrupted the traditional model of how to cover politics.
Disruptive, eh? Well we can't have any disruption of the NYT's political coverage now, can we?
posted by chavenet at 12:20 PM on July 22, 2013 [7 favorites]
Plus having a hard-ass, multiclassing statistician/neuroscientist
A game my friends sometimes play is putting things in real life into D&D (3E, specifically) terms. When we started playing Call of Cthulhu, one of them started doing the same thing with its rules.
posted by JHarris at 12:21 PM on July 22, 2013 [1 favorite]
A game my friends sometimes play is putting things in real life into D&D (3E, specifically) terms. When we started playing Call of Cthulhu, one of them started doing the same thing with its rules.
posted by JHarris at 12:21 PM on July 22, 2013 [1 favorite]
God I hope this means he'll still be covering politics in 2014.
If they're looking at Grantland as a model, it should all work out OK. Grantland covers a lot of non-sports stuff and does quite well, so if Silver's little area is half-focused on analyzing sports statistics and half on politics, it wouldn't break the model.
posted by LionIndex at 12:25 PM on July 22, 2013
If they're looking at Grantland as a model, it should all work out OK. Grantland covers a lot of non-sports stuff and does quite well, so if Silver's little area is half-focused on analyzing sports statistics and half on politics, it wouldn't break the model.
posted by LionIndex at 12:25 PM on July 22, 2013
Simmons is probably privately wrist-slapped every time he makes oblique references to how much ESPN has turned into a tabloid with dumbasses like Stephen A Smith and Skip Bayliss allowed to open their mouths on TV ad nauseum.
I watched some of the coverage of the NBA draft just so I could watch Simmons react when other people said things.
posted by srboisvert at 12:27 PM on July 22, 2013 [1 favorite]
I watched some of the coverage of the NBA draft just so I could watch Simmons react when other people said things.
posted by srboisvert at 12:27 PM on July 22, 2013 [1 favorite]
Disruptive, eh? Well we can't have any disruption of the NYT's political coverage now, can we?
She's using the term as praise (and is clearly thinking of it in the framework of the economic theory of "disruptive innovation" or "disruptive technology"). She also points out that the NYT management tried really hard to keep him on board. Whatever the takeaway from this story should be, "The NYT can't stand innovation" would be a really bizarre one.
posted by yoink at 12:28 PM on July 22, 2013 [4 favorites]
She's using the term as praise (and is clearly thinking of it in the framework of the economic theory of "disruptive innovation" or "disruptive technology"). She also points out that the NYT management tried really hard to keep him on board. Whatever the takeaway from this story should be, "The NYT can't stand innovation" would be a really bizarre one.
posted by yoink at 12:28 PM on July 22, 2013 [4 favorites]
Politico's breakdown of the negotiations between Silver and the Times, and Silver and ESPN/ABC.
posted by dry white toast at 12:30 PM on July 22, 2013 [1 favorite]
posted by dry white toast at 12:30 PM on July 22, 2013 [1 favorite]
Talking Points Memo about that Polito piece: Why'd Nate Silver Leave the Times.
Political blog story about a political blog story about a political blogger. Metafilter indeed.
posted by grajohnt at 12:41 PM on July 22, 2013 [2 favorites]
Political blog story about a political blog story about a political blogger. Metafilter indeed.
posted by grajohnt at 12:41 PM on July 22, 2013 [2 favorites]
Political blog story about a political blog story about a political blogger. Metafilter indeed.
"What role has social media played in this political story?"
posted by Celsius1414 at 12:45 PM on July 22, 2013 [3 favorites]
"What role has social media played in this political story?"
posted by Celsius1414 at 12:45 PM on July 22, 2013 [3 favorites]
But if he starts with the Grantland-esque ubiquitous footnotes, I will lose my shit.
What's wrong with them? I find the footnotes in the margins to be better solution than to have them at the end of a piece.
posted by AceRock at 12:50 PM on July 22, 2013
What's wrong with them? I find the footnotes in the margins to be better solution than to have them at the end of a piece.
posted by AceRock at 12:50 PM on July 22, 2013
NYTimes editor says Nate Silver didn’t fit in. Dowd, Brooks & Friedman just couldn't respect a guy who got stuff right.
posted by brundlefly at 12:55 PM on July 22, 2013 [23 favorites]
posted by brundlefly at 12:55 PM on July 22, 2013 [23 favorites]
Another +1 for the sidebar footnotes. I don't want them at the end and I certainly don't want paragraphs within parenthetical comments.
This morning Bill Barnwell had an NFL article in which Bill Simmons wrote snarky sidebar comments alongside. It was fun, and if you don't like to deviate from Barnwell's analysis-heavy descriptions, well then don't read the footnotes.
posted by mcstayinskool at 12:55 PM on July 22, 2013 [1 favorite]
This morning Bill Barnwell had an NFL article in which Bill Simmons wrote snarky sidebar comments alongside. It was fun, and if you don't like to deviate from Barnwell's analysis-heavy descriptions, well then don't read the footnotes.
posted by mcstayinskool at 12:55 PM on July 22, 2013 [1 favorite]
Brooks had an NYT article before the election talking about how bad statistical analysis is compared to good old-fashioned punditry. The headline should have read "I am Incredibly Threatened by Nate Silver"
posted by mcstayinskool at 12:58 PM on July 22, 2013 [20 favorites]
posted by mcstayinskool at 12:58 PM on July 22, 2013 [20 favorites]
grajohnt: "Talking Points Memo about that Polito piece: Why'd Nate Silver Leave the Times.
Political blog story about a political blog story about a political blogger. Metafilter indeed."
Indeed. Though to be honest in that case I was really happy to read what Politico said without, you know, actually having to click over to Politico.
posted by MCMikeNamara at 1:13 PM on July 22, 2013 [2 favorites]
Political blog story about a political blog story about a political blogger. Metafilter indeed."
Indeed. Though to be honest in that case I was really happy to read what Politico said without, you know, actually having to click over to Politico.
posted by MCMikeNamara at 1:13 PM on July 22, 2013 [2 favorites]
His entire probability-based way of looking at politics ran against the kind of political journalism that The Times specializes in: polling, the horse race, campaign coverage, analysis based on campaign-trail observation, and opinion writing, or “punditry,” as he put it, famously describing it as “fundamentally useless.”
it's odd to make a dichotomy between what Silver was doing and the "horse race." You can't have a horse race without *odds*, Silver was just the Times house bookie with everyone else doing color commentary. There's plenty of bullshit to spew when people are placing bets. What Silver was doing is the epitome of "horse race" coverage of politics i.e. no politics at all, just who's winning or losing the race.
posted by ennui.bz at 1:15 PM on July 22, 2013 [9 favorites]
it's odd to make a dichotomy between what Silver was doing and the "horse race." You can't have a horse race without *odds*, Silver was just the Times house bookie with everyone else doing color commentary. There's plenty of bullshit to spew when people are placing bets. What Silver was doing is the epitome of "horse race" coverage of politics i.e. no politics at all, just who's winning or losing the race.
posted by ennui.bz at 1:15 PM on July 22, 2013 [9 favorites]
I was really happy to read what Politico said without, you know, actually having to click over to Politico
Amen to that. The increasingly aggressive advertising on TPM has really started to drive me away from it lately, but I'd still rather read there than give Politico a single click.
Silver continually sticking it to Politico is probably my favorite thing he's ever done.
posted by Kosh at 1:17 PM on July 22, 2013 [7 favorites]
Amen to that. The increasingly aggressive advertising on TPM has really started to drive me away from it lately, but I'd still rather read there than give Politico a single click.
Silver continually sticking it to Politico is probably my favorite thing he's ever done.
posted by Kosh at 1:17 PM on July 22, 2013 [7 favorites]
also
In a better world, the Times would realize that there is good long-term money to be made in in having experts with informed opinions analyse the news (note: ezra klein is not an expert, he just plays one on the internet.) Instead it seems like they want to be a "media" company with personalities drawing traffic. You can see that a personality driven market is always going to be dominated by those with the deepest pockets, in this case, Disney.
posted by ennui.bz at 1:25 PM on July 22, 2013 [19 favorites]
did the New York Times know what it had in Silver?What Silver showed is that there is real value in news analysis. The Times wanted to treat Silver like a personality, a star, but his success was mainly about getting the analysis right when no one else was even bothering to do it. What Silver was doing was not "genius" work but statistical analysis that any expert could do. He's replaceable if you understand what was valuable about him. The Ivy educated courtiers at the Times have such a narrow "palace gossip" view of the world, that they don't understand that Silver is entirely replaceable.
In a better world, the Times would realize that there is good long-term money to be made in in having experts with informed opinions analyse the news (note: ezra klein is not an expert, he just plays one on the internet.) Instead it seems like they want to be a "media" company with personalities drawing traffic. You can see that a personality driven market is always going to be dominated by those with the deepest pockets, in this case, Disney.
posted by ennui.bz at 1:25 PM on July 22, 2013 [19 favorites]
What Silver was doing is the epitome of "horse race" coverage of politics i.e. no politics at all, just who's winning or losing the race.
You haven't actually read his stuff much, have you?
posted by yoink at 1:31 PM on July 22, 2013 [1 favorite]
You haven't actually read his stuff much, have you?
posted by yoink at 1:31 PM on July 22, 2013 [1 favorite]
What Silver was doing is the epitome of "horse race" coverage of politics i.e. no politics at all, just who's winning or losing the race.
True but what he did was deflate all the air out of anyone else's horse race pontificating. Once he's done his analysis there isn't much room for the traditional hunch based commenting. And it's not like anyone wants to talk about actual policies or platforms.
posted by octothorpe at 1:32 PM on July 22, 2013
True but what he did was deflate all the air out of anyone else's horse race pontificating. Once he's done his analysis there isn't much room for the traditional hunch based commenting. And it's not like anyone wants to talk about actual policies or platforms.
posted by octothorpe at 1:32 PM on July 22, 2013
I do hope they give Silver a long leash, and don't force him to write a disproportionate amount of columns about Tiger Woods and Tim Tebow.
I find it hard to believe ESPN's ability to "force" Silver to write about anything survived past the first round of negotiations.
posted by mediareport at 1:39 PM on July 22, 2013
I find it hard to believe ESPN's ability to "force" Silver to write about anything survived past the first round of negotiations.
posted by mediareport at 1:39 PM on July 22, 2013
Silver went to ESPN for the same reason Willie Sutton preferred to rob banks. Because "that's where the money is" (til 2016, anyway).
This is about the enormous volume and rapid growth of sports betting.
posted by jamjam at 1:44 PM on July 22, 2013
This is about the enormous volume and rapid growth of sports betting.
posted by jamjam at 1:44 PM on July 22, 2013
My response to hearing that Nate Silver was moving was something about one less reason to subscribe to the Times.
posted by immlass at 2:00 PM on July 22, 2013
posted by immlass at 2:00 PM on July 22, 2013
Brooks had an NYT article before the election talking about how bad statistical analysis is compared to good old-fashioned punditry. The headline should have read "I am Incredibly Threatened by Nate Silver"
My favorite piece of that ilk was Jonah Goldberg's column "Nate Silver's Numbers Racket," which was published election day and didn't reach his own publication until after the results were in. This is a column that half-heartedly asks who's to say whose statistical model is better, follows that up with "the soul . . . is not so easily number-crunched," and finishes with a nice bit of false equivalence wherein everyone is guilty of scientism. It's such a perfectly anemic bit of waffling, as if there's just no way to compare two differing claims.
I bet we'll keep seeing a lot of that as Nate Silver gets back into sports, except with "grit" or "the X factor" in place of "the soul."
posted by knuckle tattoos at 2:08 PM on July 22, 2013 [3 favorites]
My favorite piece of that ilk was Jonah Goldberg's column "Nate Silver's Numbers Racket," which was published election day and didn't reach his own publication until after the results were in. This is a column that half-heartedly asks who's to say whose statistical model is better, follows that up with "the soul . . . is not so easily number-crunched," and finishes with a nice bit of false equivalence wherein everyone is guilty of scientism. It's such a perfectly anemic bit of waffling, as if there's just no way to compare two differing claims.
I bet we'll keep seeing a lot of that as Nate Silver gets back into sports, except with "grit" or "the X factor" in place of "the soul."
posted by knuckle tattoos at 2:08 PM on July 22, 2013 [3 favorites]
What Silver was doing is the epitome of "horse race" coverage of politics i.e. no politics at all, just who's winning or losing the race.
If you're going to do it, you might as well strip all the bullshit out and do it right. A lot of supposed horse-race coverage is really just concern trolling. (Obama is losing, he needs to shift to the right/kowtow to my pet interest group!)
posted by empath at 2:12 PM on July 22, 2013 [2 favorites]
If you're going to do it, you might as well strip all the bullshit out and do it right. A lot of supposed horse-race coverage is really just concern trolling. (Obama is losing, he needs to shift to the right/kowtow to my pet interest group!)
posted by empath at 2:12 PM on July 22, 2013 [2 favorites]
I hope the Times gets neuroscientist Sam Wang of the Princeton Election Consortium to replace Nate Silver. Wang didn't get the visibility Silver did during the last election even though his model seemed as good or better than Silver's in many respects.
I'm not sure that would help much.
Silver's gifts aren't so directly in his modeling -- in the end, his model pretty much boils down to "People are going to vote for the candidate they told you they'd vote for," though the way he rates and aggregates polls while including some other information is pretty clever. But as modeling, it's not earth-shatteringly clever; I expect most ABDs working in behavior could whip up something similar, if not quite as complex.*
What makes Silver so special are his gifts at communication. He's able to describe the parts of the model that need describing in any particular post clearly and simply, and able to deal with sometimes subtle methodological issues in a clear and frank matter that doesn't drown the reader in detail but also doesn't patronize. I'm not sure I've seen anyone able to write with his combination of numeracy and clarity. Which makes him very hard to replace.
*I expect that the next phase of this will be combining his simulation technique with elements of multilevel regression and poststratification, especially for infrequently-polled states.
posted by ROU_Xenophobe at 2:12 PM on July 22, 2013 [16 favorites]
I'm not sure that would help much.
Silver's gifts aren't so directly in his modeling -- in the end, his model pretty much boils down to "People are going to vote for the candidate they told you they'd vote for," though the way he rates and aggregates polls while including some other information is pretty clever. But as modeling, it's not earth-shatteringly clever; I expect most ABDs working in behavior could whip up something similar, if not quite as complex.*
What makes Silver so special are his gifts at communication. He's able to describe the parts of the model that need describing in any particular post clearly and simply, and able to deal with sometimes subtle methodological issues in a clear and frank matter that doesn't drown the reader in detail but also doesn't patronize. I'm not sure I've seen anyone able to write with his combination of numeracy and clarity. Which makes him very hard to replace.
*I expect that the next phase of this will be combining his simulation technique with elements of multilevel regression and poststratification, especially for infrequently-polled states.
posted by ROU_Xenophobe at 2:12 PM on July 22, 2013 [16 favorites]
The Onion: Nate Silver Warns Against Overestimating His Value To ESPN
posted by wikipedia brown boy detective at 2:29 PM on July 22, 2013 [3 favorites]
posted by wikipedia brown boy detective at 2:29 PM on July 22, 2013 [3 favorites]
If you're going to do it, you might as well strip all the bullshit out and do it right.
That would make gambling boring. There's more tolife betting on horses than winning and losing and losing and losing.
True but what he did was deflate all the air out of anyone else's horse race pontificating. Once he's done his analysis there isn't much room for the traditional hunch based commenting. And it's not like anyone wants to talk about actual policies or platforms.
It's not like Silver was doing string theory. The real meta-question to ask is: why weren't even the polling companies doing basic statistical analysis? It's almost as if the predictive value of their polling was less important than other considerations.
US presidential elections are only vaguely democratic. I mean, Stalin was elected General Secretary: how much more democratic is it when you have two candidates instead of one? A horse track (like any sport) has general seating and VIP seating where the owners and their friends sit. Brooks, Friedman, et al are about communicating to the general seating about what the owners think the horse race is about. It's clear that they don't actually take the horse race very seriously, at least compared to Silver, but without the race they wouldn't have an audience. Silver is coming in and saying, hey, don't pay attention to those guys, lets get all sabermetric about the game. Which is to say that maybe he is misunderstanding what the game is actually about.
posted by ennui.bz at 2:39 PM on July 22, 2013
That would make gambling boring. There's more to
True but what he did was deflate all the air out of anyone else's horse race pontificating. Once he's done his analysis there isn't much room for the traditional hunch based commenting. And it's not like anyone wants to talk about actual policies or platforms.
It's not like Silver was doing string theory. The real meta-question to ask is: why weren't even the polling companies doing basic statistical analysis? It's almost as if the predictive value of their polling was less important than other considerations.
US presidential elections are only vaguely democratic. I mean, Stalin was elected General Secretary: how much more democratic is it when you have two candidates instead of one? A horse track (like any sport) has general seating and VIP seating where the owners and their friends sit. Brooks, Friedman, et al are about communicating to the general seating about what the owners think the horse race is about. It's clear that they don't actually take the horse race very seriously, at least compared to Silver, but without the race they wouldn't have an audience. Silver is coming in and saying, hey, don't pay attention to those guys, lets get all sabermetric about the game. Which is to say that maybe he is misunderstanding what the game is actually about.
posted by ennui.bz at 2:39 PM on July 22, 2013
I hope the Times gets neuroscientist Sam Wang of the Princeton Election Consortium to replace Nate Silver. Wang didn't get the visibility Silver did during the last election even though his model seemed as good or better than Silver's in many respects.
Sadly, it seems more like the lesson the Times learned was that Nate Silver is a Magical Genius Whose Powers We Can't Comprehend and I get the sense the feeling among the influential senior people there was that they could maybe put up with his style of coverage for his Magical Genius (and massive numbers of hits from people who wouldn't get near their website otherwise) but it doesn't seem super likely they'd deal with someone trying a similar approach who didn't come with a large pre-existing audience.
Grantland is wonderful in that they are allowed _mostly_ free speech to say what they want. Simmons is probably privately wrist-slapped every time he makes oblique references to how much ESPN has turned into a tabloid with dumbasses like Stephen A Smith and Skip Bayliss allowed to open their mouths on TV ad nauseum. I do hope they give Silver a long leash, and don't force him to write a disproportionate amount of columns about Tiger Woods and Tim Tebow.
Simmons has gotten into a regular cycle of saying something mean about ESPN every couple of months and having to shut down his Twitter for a few days and repent or whatever. I'd be embarrassed to be publicly affiliated with ESPN, too, but as long as they keep backing dump trucks full of money to his house and let him shoot the shit with Jalen Rose on national TV, he's probably sticking around. I can see Nate falling into a similar place, where he's trying to keep as much distance as possible from ESPN proper while still taking their money and using their resources, access, and platform.
posted by Copronymus at 2:45 PM on July 22, 2013 [1 favorite]
Sadly, it seems more like the lesson the Times learned was that Nate Silver is a Magical Genius Whose Powers We Can't Comprehend and I get the sense the feeling among the influential senior people there was that they could maybe put up with his style of coverage for his Magical Genius (and massive numbers of hits from people who wouldn't get near their website otherwise) but it doesn't seem super likely they'd deal with someone trying a similar approach who didn't come with a large pre-existing audience.
Grantland is wonderful in that they are allowed _mostly_ free speech to say what they want. Simmons is probably privately wrist-slapped every time he makes oblique references to how much ESPN has turned into a tabloid with dumbasses like Stephen A Smith and Skip Bayliss allowed to open their mouths on TV ad nauseum. I do hope they give Silver a long leash, and don't force him to write a disproportionate amount of columns about Tiger Woods and Tim Tebow.
Simmons has gotten into a regular cycle of saying something mean about ESPN every couple of months and having to shut down his Twitter for a few days and repent or whatever. I'd be embarrassed to be publicly affiliated with ESPN, too, but as long as they keep backing dump trucks full of money to his house and let him shoot the shit with Jalen Rose on national TV, he's probably sticking around. I can see Nate falling into a similar place, where he's trying to keep as much distance as possible from ESPN proper while still taking their money and using their resources, access, and platform.
posted by Copronymus at 2:45 PM on July 22, 2013 [1 favorite]
There is no doubt that the NY Times will miss Nate. During the last two months run-up to the general election his pages had to have been among the most heavily visited each day. A good number of people only went to his pages to see the latest projections.
posted by Rashomon at 2:57 PM on July 22, 2013 [1 favorite]
posted by Rashomon at 2:57 PM on July 22, 2013 [1 favorite]
The real meta-question to ask is: why weren't even the polling companies doing basic statistical analysis?
Polling firms do what they're paid to do. Most of the big firm have PhDs on staff who can do whatever kind of statistical analysis is needed (and anyway, all the raw data is there in the crosstabs), but those that are hired by newspapers/media outlets are typically only paid for the results.
Partisan firms hired by campaigns often do more analysis, or they work with other firms hired by the campaign to do the analysis, if there's the money for it.
posted by lunasol at 3:12 PM on July 22, 2013
Polling firms do what they're paid to do. Most of the big firm have PhDs on staff who can do whatever kind of statistical analysis is needed (and anyway, all the raw data is there in the crosstabs), but those that are hired by newspapers/media outlets are typically only paid for the results.
Partisan firms hired by campaigns often do more analysis, or they work with other firms hired by the campaign to do the analysis, if there's the money for it.
posted by lunasol at 3:12 PM on July 22, 2013
I predict Mr. Silver will find a welcome home for public predictions at ESPN, but he'll have another, less public forecast available at, let us say, Star's End.
posted by infinitewindow at 3:30 PM on July 22, 2013 [6 favorites]
posted by infinitewindow at 3:30 PM on July 22, 2013 [6 favorites]
What Silver was doing is the epitome of "horse race" coverage of politics i.e. no politics at all, just who's winning or losing the race.
Maybe so, but unlike most of the Times op-ed hacks, at least Silver was able to cover who was winning the actual election, as opposed to who was winning the pundit narrative of that day's news cycle.
posted by strangely stunted trees at 4:30 PM on July 22, 2013 [5 favorites]
Maybe so, but unlike most of the Times op-ed hacks, at least Silver was able to cover who was winning the actual election, as opposed to who was winning the pundit narrative of that day's news cycle.
posted by strangely stunted trees at 4:30 PM on July 22, 2013 [5 favorites]
How do I put this without sounding like a curmudgeon or a devil's advocate?
Well, here goes: Silver was and is massively overrated. He was a fine statistician and a good enough writer. And by hiring him, the Times seriously improved its political coverage; all the newspapers of record should have a polling statistician as good as him on hand to nail the 'horse race' coverage, as they call it in the biz. But here is the thing: there are plenty of polling statisticians around who ARE just as good as Silver: he was overrated because everyone pretended he was uniquely accurate or predictive when he was far from unique.
His vaunted model was not markedly better than the one Pollster (Simon Jackman) used. Nor did it beat the Princeton Election Consortium (Sam Wang) in 2012. In 2008/10/12, Silver was good but so were others; I think 'Nate Silver' just became a media shortcut for 'polling statistician'
The Times can easily replace him by hiring another good statistician, and pairing them with a serviceable writer: all they have to do is come up with a catchy brand name for the page. Five Thirty Eight always seemed pretty clunky to me anyway.
By the way I say this as someone who checked fivethirtyeight.com as compulsively as anyone else in 2012 (and 2008)!
posted by A Fine Mess at 4:51 PM on July 22, 2013 [1 favorite]
Well, here goes: Silver was and is massively overrated. He was a fine statistician and a good enough writer. And by hiring him, the Times seriously improved its political coverage; all the newspapers of record should have a polling statistician as good as him on hand to nail the 'horse race' coverage, as they call it in the biz. But here is the thing: there are plenty of polling statisticians around who ARE just as good as Silver: he was overrated because everyone pretended he was uniquely accurate or predictive when he was far from unique.
His vaunted model was not markedly better than the one Pollster (Simon Jackman) used. Nor did it beat the Princeton Election Consortium (Sam Wang) in 2012. In 2008/10/12, Silver was good but so were others; I think 'Nate Silver' just became a media shortcut for 'polling statistician'
The Times can easily replace him by hiring another good statistician, and pairing them with a serviceable writer: all they have to do is come up with a catchy brand name for the page. Five Thirty Eight always seemed pretty clunky to me anyway.
By the way I say this as someone who checked fivethirtyeight.com as compulsively as anyone else in 2012 (and 2008)!
posted by A Fine Mess at 4:51 PM on July 22, 2013 [1 favorite]
I think if you're in any business where you're transparent enough to enable copycats, it's smart to get out while you're on top, and he pretty much aced that 2012 election. Anything less than perfection in the coming years would be seen as a failure. But it's nice that it sounds like he's still going to do some coverage during election years.
That said, sports is a far more lucrative and emerging market for someone like him, and there are a million different directions he could go. Obviously advanced statistics are becoming more and more well-known to less-knowledgeable fans, and I think his ability to explain why a certain advanced stat is worth more than another is valuable to a network like ESPN. On top of that, there's even more money to be made by using that knowledge to influence the sports betting world.
Personally, I wish he were more of a hoops fan than a baseball fan - basketball is so fluid and dynamic that advanced analysis is very difficult to quantify for the common fan. I actually attempted to buy access to Synergy Sports (which does advanced basketball video analysis for NBA teams), but they do not sell to those who are not affiliated with a team. Baseball is much easier to quantify because of the frequent stops in action and limited "scenarios," so it's easier to get an apples-to-apples comparison for most situations.
posted by antonymous at 5:19 PM on July 22, 2013
That said, sports is a far more lucrative and emerging market for someone like him, and there are a million different directions he could go. Obviously advanced statistics are becoming more and more well-known to less-knowledgeable fans, and I think his ability to explain why a certain advanced stat is worth more than another is valuable to a network like ESPN. On top of that, there's even more money to be made by using that knowledge to influence the sports betting world.
Personally, I wish he were more of a hoops fan than a baseball fan - basketball is so fluid and dynamic that advanced analysis is very difficult to quantify for the common fan. I actually attempted to buy access to Synergy Sports (which does advanced basketball video analysis for NBA teams), but they do not sell to those who are not affiliated with a team. Baseball is much easier to quantify because of the frequent stops in action and limited "scenarios," so it's easier to get an apples-to-apples comparison for most situations.
posted by antonymous at 5:19 PM on July 22, 2013
I read the NYTimes Public Editor's blog post. That blog post reads like something one would write on Facebook at 3am after heavy drinking, perhaps during a nasty breakup.
posted by humanfont at 5:28 PM on July 22, 2013 [1 favorite]
posted by humanfont at 5:28 PM on July 22, 2013 [1 favorite]
I don't think Nate Silver will find a comfortable future at ESPN unless he can somehow parley his statistical work into becoming a knowledgeable sports commentator. ESPN's big edge online is not so much television but fantasy sports where he may just as well find a home as some kind of back office statistical manager.
posted by parmanparman at 5:41 PM on July 22, 2013
posted by parmanparman at 5:41 PM on July 22, 2013
Some strange comments here.
Nate Silver basically averages polls, tossing in some overly complicated twists that make his analysis a bit less of a commodity. His work hardly paints polling operations in a bad light, he depends on them for his projections.
"Horse race" coverage is not a synonym for bullshit, although most horse race coverage is bullshit. Most of Silver's political commentary is horse race coverage.
Silver's background is in sports commentary, I'm pretty sure he'll find a way to churn out interesting ESPN content.
posted by leopard at 5:48 PM on July 22, 2013
Nate Silver basically averages polls, tossing in some overly complicated twists that make his analysis a bit less of a commodity. His work hardly paints polling operations in a bad light, he depends on them for his projections.
"Horse race" coverage is not a synonym for bullshit, although most horse race coverage is bullshit. Most of Silver's political commentary is horse race coverage.
Silver's background is in sports commentary, I'm pretty sure he'll find a way to churn out interesting ESPN content.
posted by leopard at 5:48 PM on July 22, 2013
Frank Conniff's tweet got it exactly right. With a few exceptions on the editorial page, Charles Blow mostly, the NYT is about one particular Nobel Prize in Economics winner away from being purely irrelevant.
Also, A Fine Mess: Silver was and is massively overrated...there are plenty of polling statisticians around who ARE just as good as Silver: he was overrated because everyone pretended he was uniquely accurate or predictive when he was far from unique.
I don't think that there's much doubt that Silver could be replaced, the question is whether or not he will be. Silver's "massive overrating" is a result of his having built his personal brand up, a non-trivial task for something as dry and (for the average American) mysterious and largely incomprehensible as statistics, to the point that he could fend off the likes of that "unskewed polls" lout, who used the highly-scientific method of casting aspersions on Silver's masculinity, made a shamefaced walkback the day after the election, and has now walked back his walkback and is working the voter fraud scam because he finally figured out where the wingnut money is. Anyone who has to follow Silver's act will be just another random numbers cruncher by comparison.
Also, too, humanfont: I read the NYTimes Public Editor's blog post. That blog post reads like something one would write on Facebook at 3am after heavy drinking, perhaps during a nasty breakup.
Oh, don't I wish. It reads like someone who makes a very good living at being able to say "shitstorm" without, you know, actually saying it.
posted by Halloween Jack at 6:28 PM on July 22, 2013 [2 favorites]
Also, A Fine Mess: Silver was and is massively overrated...there are plenty of polling statisticians around who ARE just as good as Silver: he was overrated because everyone pretended he was uniquely accurate or predictive when he was far from unique.
I don't think that there's much doubt that Silver could be replaced, the question is whether or not he will be. Silver's "massive overrating" is a result of his having built his personal brand up, a non-trivial task for something as dry and (for the average American) mysterious and largely incomprehensible as statistics, to the point that he could fend off the likes of that "unskewed polls" lout, who used the highly-scientific method of casting aspersions on Silver's masculinity, made a shamefaced walkback the day after the election, and has now walked back his walkback and is working the voter fraud scam because he finally figured out where the wingnut money is. Anyone who has to follow Silver's act will be just another random numbers cruncher by comparison.
Also, too, humanfont: I read the NYTimes Public Editor's blog post. That blog post reads like something one would write on Facebook at 3am after heavy drinking, perhaps during a nasty breakup.
Oh, don't I wish. It reads like someone who makes a very good living at being able to say "shitstorm" without, you know, actually saying it.
posted by Halloween Jack at 6:28 PM on July 22, 2013 [2 favorites]
Things In Politico That Make Me Want To Guzzle Antifreeze, Part The Infinity
posted by homunculus at 7:09 PM on July 22, 2013
posted by homunculus at 7:09 PM on July 22, 2013
He was a fine statistician and a good enough writer.
Ain't that enough? The combination of the two seems to be in short supply. He never said he was doing anything magical, and he explained his methodology often in his column and his book. The whole point of his book is that he's not a genius, and that anybody can do what he does, but for some reason, they don't.
posted by empath at 7:12 PM on July 22, 2013 [3 favorites]
Ain't that enough? The combination of the two seems to be in short supply. He never said he was doing anything magical, and he explained his methodology often in his column and his book. The whole point of his book is that he's not a genius, and that anybody can do what he does, but for some reason, they don't.
posted by empath at 7:12 PM on July 22, 2013 [3 favorites]
He never said he was doing anything magical, and he explained his methodology often in his column and his book. The whole point of his book is that he's not a genius, and that anybody can do what he does, but for some reason, they don't.
This this this. How soon everyone forgets at the time that the media was trying to build up Romney vs. Obama into this big thing. Then comes Nate Silver saying it's Obama 90%... no, 93%... no, 95%. Even though when Silver says 90% he means it, that is, Romney still has a 10% chance, he was the only prominent voice who was saying anything like what he was, and he was right. As a result, Nate Silver has a big career now, and Karl Rove is on the outs.
And the lesson the NYT takes away is they need more personal brands, not to hire people who know numbers. SIGH.
posted by JHarris at 7:42 PM on July 22, 2013 [2 favorites]
This this this. How soon everyone forgets at the time that the media was trying to build up Romney vs. Obama into this big thing. Then comes Nate Silver saying it's Obama 90%... no, 93%... no, 95%. Even though when Silver says 90% he means it, that is, Romney still has a 10% chance, he was the only prominent voice who was saying anything like what he was, and he was right. As a result, Nate Silver has a big career now, and Karl Rove is on the outs.
And the lesson the NYT takes away is they need more personal brands, not to hire people who know numbers. SIGH.
posted by JHarris at 7:42 PM on July 22, 2013 [2 favorites]
What makes Silver so special are his gifts at communication.
For the record, Sam Wang is really good at clear communication as well. Check his series of posts on gerrymandering from last winter, e.g. He'd be great at the NYT.
posted by mediareport at 8:03 PM on July 22, 2013 [5 favorites]
For the record, Sam Wang is really good at clear communication as well. Check his series of posts on gerrymandering from last winter, e.g. He'd be great at the NYT.
posted by mediareport at 8:03 PM on July 22, 2013 [5 favorites]
It's understandable why some people will be concerned at the idea of a statistician being involved with an event that revolves around a series of 'revelations' of secret information. Mathematical analysis of anything traditionally qualitative can cause a great deal of distress.
"You just let the machines get on with the adding up," warned Majikthise, "and we'll take care of the eternal verities, thank you very much. You want to check your legal position you do mate. Under law the Quest for Ultimate Truth is quite clearly the inalienable prerogative of your working thinkers. Any bloody machine goes and actually finds it and we're straight out of a job aren't we? I mean what's the use of our sitting up half the night arguing that there may or may not be a God if this machine only goes and gives us his bleeding phone number the next morning?"
"That's right!" shouted Vroomfondel, "we demand rigidly defined areas of doubt and uncertainty!"
posted by The Zeroth Law at 12:32 AM on July 23, 2013 [2 favorites]
"You just let the machines get on with the adding up," warned Majikthise, "and we'll take care of the eternal verities, thank you very much. You want to check your legal position you do mate. Under law the Quest for Ultimate Truth is quite clearly the inalienable prerogative of your working thinkers. Any bloody machine goes and actually finds it and we're straight out of a job aren't we? I mean what's the use of our sitting up half the night arguing that there may or may not be a God if this machine only goes and gives us his bleeding phone number the next morning?"
"That's right!" shouted Vroomfondel, "we demand rigidly defined areas of doubt and uncertainty!"
posted by The Zeroth Law at 12:32 AM on July 23, 2013 [2 favorites]
The NYT wants to be more like Politico and less data and fact driven. Got it.
posted by professor plum with a rope at 1:31 AM on July 23, 2013
posted by professor plum with a rope at 1:31 AM on July 23, 2013
professor plum with a rope: "The NYT wants to be more like Politico and less data and fact driven. Got it."
I'll rephrase this. "Elite" journalists, ie the status jockeyers who fought tooth and nail through newsroom politics to get hired by the WaPo, or NYT and manage to get themselves to work on national politics don't have skills beyond the "horse-race," "analysis", or whatever else they pat themselves on the back for doing. That's because their primary skill is people politics.
This is fine in and of itself, cut-throat organizations and bureaucracies refine these skills in people. But remember, these are the people who are hired to understand and relate national policy and they have ZERO interest in doing this job. They give two shits about the costs of dysfunction, revolving door payouts in DC, and how the 3 years of crying about debt was the largest concern trolling since the morality of Clinton's dick, because as a group of fucking functional sociopaths, understanding and relating what's actually important is not and has never been their job. They're very jobs, insider-status, and sinecures were all secured by being very good at essentially reading the pettiest of politics, and that's what they want to write about.
So of course Silver is a fucking threat to their way of life.
Fuck.
posted by stratastar at 1:53 AM on July 23, 2013 [2 favorites]
I'll rephrase this. "Elite" journalists, ie the status jockeyers who fought tooth and nail through newsroom politics to get hired by the WaPo, or NYT and manage to get themselves to work on national politics don't have skills beyond the "horse-race," "analysis", or whatever else they pat themselves on the back for doing. That's because their primary skill is people politics.
This is fine in and of itself, cut-throat organizations and bureaucracies refine these skills in people. But remember, these are the people who are hired to understand and relate national policy and they have ZERO interest in doing this job. They give two shits about the costs of dysfunction, revolving door payouts in DC, and how the 3 years of crying about debt was the largest concern trolling since the morality of Clinton's dick, because as a group of fucking functional sociopaths, understanding and relating what's actually important is not and has never been their job. They're very jobs, insider-status, and sinecures were all secured by being very good at essentially reading the pettiest of politics, and that's what they want to write about.
So of course Silver is a fucking threat to their way of life.
Fuck.
posted by stratastar at 1:53 AM on July 23, 2013 [2 favorites]
They're very jobs, insider-status, and sinecures were all secured by being very good at essentially reading the pettiest of politics, and that's what they want to write about.
They're not even paid to inform readers. They are paid to express a point of view that the newspaper's advertisers and owners want propagated to the public.
That's the main reason Nate Silver didn't 'fit in'. He actually tries to accurately predict the race and inform people about the news. All of the other political reporters only use the 'horse race' as a hat-rack to hang whatever agenda they're paid to push that week. The whole business of political journalism, at least in the mass media, is corrupt from top-to-bottom.
posted by empath at 2:14 AM on July 23, 2013
They're not even paid to inform readers. They are paid to express a point of view that the newspaper's advertisers and owners want propagated to the public.
That's the main reason Nate Silver didn't 'fit in'. He actually tries to accurately predict the race and inform people about the news. All of the other political reporters only use the 'horse race' as a hat-rack to hang whatever agenda they're paid to push that week. The whole business of political journalism, at least in the mass media, is corrupt from top-to-bottom.
posted by empath at 2:14 AM on July 23, 2013
Color me a Nate Silver non-fan. Which isn't to say I don't respect him and what he does -- he's as good a poll analyst as ever wrote for the Times, as far as I can tell. That's not to be taken lightly.
But as with what A Fine Mess wrote, and leopard as well -- when he reduces the whole thing to some numbers, it deflates a lot of the point of the thing. That's very comfortable when the result is what you want, but speaking as someone in Wisconsin, the opposite makes it all very uncomfortable.
Put it this way, as an aside. We had a guy who tried to be Nate Silver for the recall elections. He was very comfortable talking numbers, coming up with exit poll screens, and would tell you six ways to Sunday that Walker was sure to lose the recall. But he didn't; he won solidly (and with a huge number of votes from union households, no less). To look at that result and just be blandly told the Nate Silver inevitability pitch is pretty debilitating to any sort of viable political discourse (of course, the result by itself was mostly to blame, but I'm getting at the message being touted by Nate here). Silver's predictions of victory for Obama were widely shared as a means of sort of emotional support, especially in that brief first-debate surge period Romney had, but I completely understand what more traditional pundits are talking about when they say it takes the air out of everything. "Oh, he's gonna win. Well, what the hell are we doing here? Anybody for a brew?"
And despite the I-emphasize-deserved criticism of the pundit class above, I think that's something that makes our democracy poorer.
I'll bring up another point. Silver looked at some other prediction models as a sort of filler piece here and there (because he'd taken all the air out of the balloon -- what is there to discuss?), and one of them was Allan Lichtman's Keys to the Presidency model. [previously on mefi] Oh, he savaged it, slingshot it out past Saturn. He said the model was too perfect, so it couldn't possibly be useful, as if Lichtman had been tuning it and tuning it right up to the day of Nate's analysis.
But he didn't seem to understand how it was built (literally with the aid of a statistical modeler of volcanic eruptions), nor that it was essentially unchanged in twenty years -- twenty years in which it has been right every time.
The one exception being 2000, when it correctly predicted that Gore would win the popular vote, to which it was tuned by choice versus the electoral wins of the 19th century that went against the grain. So it was right in that sense.
Now, I'm not going to tell you that Lichtman's Keys model is perfect. It doesn't tell you the percentage of the popular or electoral vote. It doesn't break down easily for states. It involves a lot of judgement in its application which Lichtman uncomfortably waves off as the realm of a professional historian. It's also hard to reconcile the model with the significant changes in the American political scene -- thouh by the same token it tries to use a larger data set (more elections) than most models.
But I'll tell you this -- it's one of the few models that accounts for much more than the economy, and one of the few that, how can I put this, engages with politics in a significant way. I think there's value in that, and it's specifically why I am less enamored with Silver's approach. I have been an advocate and defender of the Keys system in the past (I even ran a small forum back on Dave Winer's platform for a while), but I really don't think this is just about Lichtman -- I think that Silver's criticism of it revealed some of the severe limitations of his approach.
Especially underlined here in Wisconsin, where we have governance, but virtually no politics anymore. The Republicans are just doing stuff, they're not selling it to voters, asking for opinions even of Republican rank-and-file, they're just passing legislation from the darkest corners of the Capitol -- which for Wisconsin is so very weird. I think, now, that this is what a country without politics looks like. A bit like Warsaw Pact nations, or China, or other nations gripped by a dictatorship. Politics is important. Politics is, to coin a phrase, made of people. And that's what I worry that Silver's grinding it down to the numbers can do.
If we know the outcome, and we have 90% confidence in that prediction, what indeed are we to talk about?
posted by dhartung at 2:55 AM on July 23, 2013
But as with what A Fine Mess wrote, and leopard as well -- when he reduces the whole thing to some numbers, it deflates a lot of the point of the thing. That's very comfortable when the result is what you want, but speaking as someone in Wisconsin, the opposite makes it all very uncomfortable.
Put it this way, as an aside. We had a guy who tried to be Nate Silver for the recall elections. He was very comfortable talking numbers, coming up with exit poll screens, and would tell you six ways to Sunday that Walker was sure to lose the recall. But he didn't; he won solidly (and with a huge number of votes from union households, no less). To look at that result and just be blandly told the Nate Silver inevitability pitch is pretty debilitating to any sort of viable political discourse (of course, the result by itself was mostly to blame, but I'm getting at the message being touted by Nate here). Silver's predictions of victory for Obama were widely shared as a means of sort of emotional support, especially in that brief first-debate surge period Romney had, but I completely understand what more traditional pundits are talking about when they say it takes the air out of everything. "Oh, he's gonna win. Well, what the hell are we doing here? Anybody for a brew?"
And despite the I-emphasize-deserved criticism of the pundit class above, I think that's something that makes our democracy poorer.
I'll bring up another point. Silver looked at some other prediction models as a sort of filler piece here and there (because he'd taken all the air out of the balloon -- what is there to discuss?), and one of them was Allan Lichtman's Keys to the Presidency model. [previously on mefi] Oh, he savaged it, slingshot it out past Saturn. He said the model was too perfect, so it couldn't possibly be useful, as if Lichtman had been tuning it and tuning it right up to the day of Nate's analysis.
But he didn't seem to understand how it was built (literally with the aid of a statistical modeler of volcanic eruptions), nor that it was essentially unchanged in twenty years -- twenty years in which it has been right every time.
The one exception being 2000, when it correctly predicted that Gore would win the popular vote, to which it was tuned by choice versus the electoral wins of the 19th century that went against the grain. So it was right in that sense.
Now, I'm not going to tell you that Lichtman's Keys model is perfect. It doesn't tell you the percentage of the popular or electoral vote. It doesn't break down easily for states. It involves a lot of judgement in its application which Lichtman uncomfortably waves off as the realm of a professional historian. It's also hard to reconcile the model with the significant changes in the American political scene -- thouh by the same token it tries to use a larger data set (more elections) than most models.
But I'll tell you this -- it's one of the few models that accounts for much more than the economy, and one of the few that, how can I put this, engages with politics in a significant way. I think there's value in that, and it's specifically why I am less enamored with Silver's approach. I have been an advocate and defender of the Keys system in the past (I even ran a small forum back on Dave Winer's platform for a while), but I really don't think this is just about Lichtman -- I think that Silver's criticism of it revealed some of the severe limitations of his approach.
Especially underlined here in Wisconsin, where we have governance, but virtually no politics anymore. The Republicans are just doing stuff, they're not selling it to voters, asking for opinions even of Republican rank-and-file, they're just passing legislation from the darkest corners of the Capitol -- which for Wisconsin is so very weird. I think, now, that this is what a country without politics looks like. A bit like Warsaw Pact nations, or China, or other nations gripped by a dictatorship. Politics is important. Politics is, to coin a phrase, made of people. And that's what I worry that Silver's grinding it down to the numbers can do.
If we know the outcome, and we have 90% confidence in that prediction, what indeed are we to talk about?
posted by dhartung at 2:55 AM on July 23, 2013
If we know the outcome, and we have 90% confidence in that prediction, what indeed are we to talk about?
How to convince people to change their minds? He's not predicting the future, he's merely extrapolating from current trends. The course of the election changed multiple times because of news stories (Romney's 51% disaster, for one) during the 2012 race, and Nate's projections changed accordingly.
All he can tell you is that if something doesn't change, then a certain result is likely to happen. Your job, as a politician, or as someone engaged in politics is to change things. His projections don't prevent anyone from doing that. They may even help you do that, so you don't blindly walk off a cliff you don't see coming.
posted by empath at 3:22 AM on July 23, 2013 [3 favorites]
How to convince people to change their minds? He's not predicting the future, he's merely extrapolating from current trends. The course of the election changed multiple times because of news stories (Romney's 51% disaster, for one) during the 2012 race, and Nate's projections changed accordingly.
All he can tell you is that if something doesn't change, then a certain result is likely to happen. Your job, as a politician, or as someone engaged in politics is to change things. His projections don't prevent anyone from doing that. They may even help you do that, so you don't blindly walk off a cliff you don't see coming.
posted by empath at 3:22 AM on July 23, 2013 [3 favorites]
It's interesting how many commenters here are bashing the NYT. Silver presumably went to ESPN because they offered him more money. Silver built up 538 into a popular site all on his own, but the Times gave him a prominent, very respectable platform. Has anyone here ever received a better job opportunity from a new employer? I'm not seeing any evidence that Silver was driven away from the Times.
dhartung, Silver is able to predict the outcomes of things like "who will win the Presidential election in a few months" because elections are decided by vote counts and because there is an army of pollsters who simply ask people who they are going to vote for and because most people have fairly stable voting preferences. Politics is not just winning elections however. Can Silver predict things like "what will happen to the economy in the next 2 years," "should Obama focus on health care reform or economic stimulus," "how will intervening in Syria affect the American political landscape"? He can offer up some relevant data but he hardly has a crystal ball. And that stuff is the heart of actual politics.
posted by leopard at 6:48 AM on July 23, 2013
dhartung, Silver is able to predict the outcomes of things like "who will win the Presidential election in a few months" because elections are decided by vote counts and because there is an army of pollsters who simply ask people who they are going to vote for and because most people have fairly stable voting preferences. Politics is not just winning elections however. Can Silver predict things like "what will happen to the economy in the next 2 years," "should Obama focus on health care reform or economic stimulus," "how will intervening in Syria affect the American political landscape"? He can offer up some relevant data but he hardly has a crystal ball. And that stuff is the heart of actual politics.
posted by leopard at 6:48 AM on July 23, 2013
Wang didn't get the visibility Silver did during the last election even though his model seemed as good or better than Silver's in many respects.
Agreed, but in what respects is the Princeton model better and how are you defining that? I do know that Wang predicted the Presidential vote results in 49 states, that Silver predicted the Presidential vote results in 50 states in 2012 and that the opposite was true in 2008. It seems to me that they have different systems, each with strengths & weaknesses, that are about equal in terms of accuracy.
posted by snottydick at 7:21 AM on July 23, 2013
Agreed, but in what respects is the Princeton model better and how are you defining that? I do know that Wang predicted the Presidential vote results in 49 states, that Silver predicted the Presidential vote results in 50 states in 2012 and that the opposite was true in 2008. It seems to me that they have different systems, each with strengths & weaknesses, that are about equal in terms of accuracy.
posted by snottydick at 7:21 AM on July 23, 2013
Here is an analysis comparing various 2012 election predictions. Wang was actually slightly more accurate than Silver in forecasting the Presidential race.
posted by leopard at 7:54 AM on July 23, 2013 [1 favorite]
posted by leopard at 7:54 AM on July 23, 2013 [1 favorite]
Wang's model seems better mostly in terms of parsimony - not as many assumptions go into it. Wang's model is mostly about the polls that are used, whereas Silver includes a number of other factors that may/may not be useful. For instance, in Silver's model, he includes an incumbency factor (incumbent candidates are more likely to be elected than challengers). I believe Wang's argument against some of these assumptions is that they are unnecessary since they are already reflected by the polling data itself.
So, yeah, the simpler model that performs as well as a more complex model is usually preferable.
posted by logicpunk at 8:00 AM on July 23, 2013 [2 favorites]
So, yeah, the simpler model that performs as well as a more complex model is usually preferable.
posted by logicpunk at 8:00 AM on July 23, 2013 [2 favorites]
If we know the outcome, and we have 90% confidence in that prediction, what indeed are we to talk about?
Silver never claimed that he knew the outcome. He was pretty clear about the fact that if the model says somebody has a 70 percent chance of winning, he or she is not guaranteed to win, but should win 7 out of 10 times. And it's odd for you to say that there was nothing to talk about when Silver wrote two or three blog posts a day about what the model's predictions said about what each candidate was or wasn't doing, and how that resonated with the public, and how that might change in the future.
Politics is, to coin a phrase, made of people.
If Silver's model is successful at predicting the outcomes of elections and it is based mostly on economic conditions, shouldn't that tell you that politics is not actually 'made of people'? Shouldn't that tell politicians that if they want to win elections, they have to focus on creating favorable economic conditions instead of pandering to the electorate? Isn't that what we want them to do?
posted by ultraviolet catastrophe at 8:15 AM on July 23, 2013 [3 favorites]
Silver never claimed that he knew the outcome. He was pretty clear about the fact that if the model says somebody has a 70 percent chance of winning, he or she is not guaranteed to win, but should win 7 out of 10 times. And it's odd for you to say that there was nothing to talk about when Silver wrote two or three blog posts a day about what the model's predictions said about what each candidate was or wasn't doing, and how that resonated with the public, and how that might change in the future.
Politics is, to coin a phrase, made of people.
If Silver's model is successful at predicting the outcomes of elections and it is based mostly on economic conditions, shouldn't that tell you that politics is not actually 'made of people'? Shouldn't that tell politicians that if they want to win elections, they have to focus on creating favorable economic conditions instead of pandering to the electorate? Isn't that what we want them to do?
posted by ultraviolet catastrophe at 8:15 AM on July 23, 2013 [3 favorites]
That's very comfortable when the result is what you want, but speaking as someone in Wisconsin, the opposite makes it all very uncomfortable.
Arguing for the virtues of false hope seems like an odd position to me. If the polling is telling you that you aren't going to win then you should listen to the polling and start thinking about what your next steps are going to be after the election so that the next time out you're the one feeling "comfortable" as the election approaches.
Some of the criticisms of Silver in this thread are truly bizarre. I get the feeling that most of you only looked at this table of predictions during the lead up to the election and didn't actually read the columns. It's simply not true, at all, to suggest that Silver is simply a "horse race" commenter who does nothing but talk about who is/isn't going to win. He writes a lot of very engaging and thoughtful political analysis about why they are going to win. He goes back over the history of opinion polling on issues and talks about what approaches have and haven't worked to change people's attitudes on important issues, he talks about why a candidate is leading in a given race in the light of economic, social, demographic and other data. He does all these things extremely well, rooting his commentary, always, in actual, observable and testable data.
The notion that this inculcates some kind of defeatist "welp, the numbers can't be beat!" attitude is just bizarrely ass-backwards. It's precisely Nate Silver's evidence-based, analytic approach to politics that, for example, Obama's data-driven election team deployed to win two historic campaigns.
Oh, and dhartung, this:
posted by yoink at 8:28 AM on July 23, 2013 [2 favorites]
Arguing for the virtues of false hope seems like an odd position to me. If the polling is telling you that you aren't going to win then you should listen to the polling and start thinking about what your next steps are going to be after the election so that the next time out you're the one feeling "comfortable" as the election approaches.
Some of the criticisms of Silver in this thread are truly bizarre. I get the feeling that most of you only looked at this table of predictions during the lead up to the election and didn't actually read the columns. It's simply not true, at all, to suggest that Silver is simply a "horse race" commenter who does nothing but talk about who is/isn't going to win. He writes a lot of very engaging and thoughtful political analysis about why they are going to win. He goes back over the history of opinion polling on issues and talks about what approaches have and haven't worked to change people's attitudes on important issues, he talks about why a candidate is leading in a given race in the light of economic, social, demographic and other data. He does all these things extremely well, rooting his commentary, always, in actual, observable and testable data.
The notion that this inculcates some kind of defeatist "welp, the numbers can't be beat!" attitude is just bizarrely ass-backwards. It's precisely Nate Silver's evidence-based, analytic approach to politics that, for example, Obama's data-driven election team deployed to win two historic campaigns.
Oh, and dhartung, this:
He said the model was too perfect, so it couldn't possibly be useful, as if Lichtman had been tuning it and tuning it right up to the day of Nate's analysis.is a spectacularly misleading summary of Silver's incisive and absolutely accurate analysis of the problems with Lichtman's "keys to the White House" model.
posted by yoink at 8:28 AM on July 23, 2013 [2 favorites]
Shouldn't that tell politicians that if they want to win elections, they have to focus on creating favorable economic conditions instead of pandering to the electorate? Isn't that what we want them to do?
The question then becomes, "does the political class have the power to create favorable economic conditions?" Certainly few or no members have that power if they try to act as individuals, which already creates plenty of problems.
posted by kewb at 8:41 AM on July 23, 2013
The question then becomes, "does the political class have the power to create favorable economic conditions?" Certainly few or no members have that power if they try to act as individuals, which already creates plenty of problems.
posted by kewb at 8:41 AM on July 23, 2013
yoink, I maintain that he has fundamentally misunderstood how the model was devised. He claims that Lichtman has chosen the ones that he, Lichtman, can manipulate to make the model work, but the initial stab at the system looked at dozens (50, 100, I don't recall) of potentially usable variables and, with the help of the volcano prediction model devised by his colleague, they applied that to those potential keys to find out which ones gave the most useful results. It was the exact opposite of selectivity; it was fitting the model in the way one tests hypotheses against reality. (" It’s less that he has discovered the right set of keys than that he’s a locksmith and can keep minting new keys until he happens to open all 38 doors.") I don't know how Silver missed the origins of the Keys system, but by not addressing it he seems to be ignorant of this, and thus his analysis of the model is suspect.
I mean, I'm not sure how you would develop a model if it wasn't by finding out which variables test better against the historical record than others. If there's a way, tell me.
I'm not quibbling with Silver's model, to be sure. It's useful when you have polling data. But you don't have polling data most of the time, so sitting on your hands and waiting for it is the only response with his model.
Ultimately, I think that Lichtman and Silver have somewhat different objectives with their models, and I'm more in tune with those suggested by the Keys than those where you have to wait around until the pollsters kick in. Again, I wish that Silver had addressed this question of goals, rather than simply evaluating the Keys as if they were a polling model or a competitor to his polling model.
posted by dhartung at 6:24 PM on July 23, 2013
I mean, I'm not sure how you would develop a model if it wasn't by finding out which variables test better against the historical record than others. If there's a way, tell me.
I'm not quibbling with Silver's model, to be sure. It's useful when you have polling data. But you don't have polling data most of the time, so sitting on your hands and waiting for it is the only response with his model.
Ultimately, I think that Lichtman and Silver have somewhat different objectives with their models, and I'm more in tune with those suggested by the Keys than those where you have to wait around until the pollsters kick in. Again, I wish that Silver had addressed this question of goals, rather than simply evaluating the Keys as if they were a polling model or a competitor to his polling model.
posted by dhartung at 6:24 PM on July 23, 2013
dhartung, what you describe here --
the initial stab at the system looked at dozens (50, 100, I don't recall) of potentially usable variables and, with the help of the volcano prediction model devised by his colleague, they applied that to those potential keys to find out which ones gave the most useful results
is exactly what the kind of selection Silver describes would look like -- taking your known outcomes and throwing a huge pile of variables at them, then taking whatever happens to be most highly correlated with your known outcomes. It's some variant of stepwise regression, and stepwise regression is shameful and evil.
I don't know why you keep mentioning the volcano prediction model as if it's a good thing. Unless there's some strong reason to expect that presidential election outcomes are like volcanic eruptions -- which is unlikely as elections are the aggregations of millions of human decisions and volcanic eruptions aren't -- it would actually be a bad idea to use a model developed to predict volcanic eruptions.
I mean, I'm not sure how you would develop a model if it wasn't by finding out which variables test better against the historical record than others. If there's a way, tell me.
There really is a simple better way to do this: have a theory. Have a theory about what causes presidential election outcomes. Then look at the variables that your theory says should be correlated with election outcomes and see how well you do. Then, if you want to be really good, think about what correlations about *other things* your theory says should exist, and look for them too. Maybe your theory also predicts that presidential approval is correlated with crop yields. Is it?
Statistical tools are really good at some things, like testing the implications of theories. They're extraordinarily bad at the process you described of trying to learn which factors cause presidential election outcomes. This is because in almost any social science setting, you're going to be throwing whatever causal variables actually exist at the outcomes... along with a bunch of other variables that are other consequences of those causal factors and a size extra humongous number of variables that happen to be correlated with the outcomes by dumb luck.
It's also the case that many processes social scientists examine are much noisier than the ones physical scientists usually deal with. In a social science setting, it's unlikely that the set of variables that most successfully predict a known set of outcomes are the actual causal forces. Instead, the best set of predictors are likely to be a random set of variables that happen to correctly predict the random noise in the sample really well. Again, you really need to have a theory here.
posted by ROU_Xenophobe at 8:50 PM on July 23, 2013 [1 favorite]
the initial stab at the system looked at dozens (50, 100, I don't recall) of potentially usable variables and, with the help of the volcano prediction model devised by his colleague, they applied that to those potential keys to find out which ones gave the most useful results
is exactly what the kind of selection Silver describes would look like -- taking your known outcomes and throwing a huge pile of variables at them, then taking whatever happens to be most highly correlated with your known outcomes. It's some variant of stepwise regression, and stepwise regression is shameful and evil.
I don't know why you keep mentioning the volcano prediction model as if it's a good thing. Unless there's some strong reason to expect that presidential election outcomes are like volcanic eruptions -- which is unlikely as elections are the aggregations of millions of human decisions and volcanic eruptions aren't -- it would actually be a bad idea to use a model developed to predict volcanic eruptions.
I mean, I'm not sure how you would develop a model if it wasn't by finding out which variables test better against the historical record than others. If there's a way, tell me.
There really is a simple better way to do this: have a theory. Have a theory about what causes presidential election outcomes. Then look at the variables that your theory says should be correlated with election outcomes and see how well you do. Then, if you want to be really good, think about what correlations about *other things* your theory says should exist, and look for them too. Maybe your theory also predicts that presidential approval is correlated with crop yields. Is it?
Statistical tools are really good at some things, like testing the implications of theories. They're extraordinarily bad at the process you described of trying to learn which factors cause presidential election outcomes. This is because in almost any social science setting, you're going to be throwing whatever causal variables actually exist at the outcomes... along with a bunch of other variables that are other consequences of those causal factors and a size extra humongous number of variables that happen to be correlated with the outcomes by dumb luck.
It's also the case that many processes social scientists examine are much noisier than the ones physical scientists usually deal with. In a social science setting, it's unlikely that the set of variables that most successfully predict a known set of outcomes are the actual causal forces. Instead, the best set of predictors are likely to be a random set of variables that happen to correctly predict the random noise in the sample really well. Again, you really need to have a theory here.
posted by ROU_Xenophobe at 8:50 PM on July 23, 2013 [1 favorite]
While Nate Silver made his name in statistical punditry, I've always thought (that's from previous election cycles, in 2004 and 2008) Sam Wang did it better; there's a quiet elegance in seeing a range of options at "landfall", as Sam Wang does it, rather than a statistical probability of something happening.
HOWEVER, here's the thing: unlike other quantitative analysts, such as The VoteMaster (Tanenbaum) and Sam Wang, Silver does awesome qualitative analysis as well. _That_ is the real reason why the Friedman's or Brooks' of the world should feel scared; even if you separate quants and hacks as two different species with different skills (which you shouldn't, but heck), Nate Silver kicks some solid ass in both respects.
posted by the cydonian at 9:00 PM on July 23, 2013 [2 favorites]
HOWEVER, here's the thing: unlike other quantitative analysts, such as The VoteMaster (Tanenbaum) and Sam Wang, Silver does awesome qualitative analysis as well. _That_ is the real reason why the Friedman's or Brooks' of the world should feel scared; even if you separate quants and hacks as two different species with different skills (which you shouldn't, but heck), Nate Silver kicks some solid ass in both respects.
posted by the cydonian at 9:00 PM on July 23, 2013 [2 favorites]
« Older Hello extremely! I hope you like to *play*. | Ghost Beach Is Closed Due To Jellyfish Newer »
This thread has been archived and is closed to new comments
posted by These Premises Are Alarmed at 12:09 PM on July 22, 2013