Nate Silver will tax your crap!

By Razib Khan | November 5, 2012 1:22 am

I am currently reading Nate Silver’s The Signal and the Noise. The review will go live concurrently with Jim Manzi’s Uncontrolled, which I finished weeks ago. The two works are qualitatively different, but fundamentally they’re both concerned with epistemology. I do have to admit that halfway through  The Signal and the Noise I long for Manzi’s density and economy of prose. As someone on the margins of the LessWrong community I’m already familiar with many of the arguments that Silver forwards, so perhaps this evaluation is not fair.

But this post isn’t about The Signal and the Noise. Rather, I want to state that I’ve put some bets down on the upcoming election. Silver attempted to do the same, and got some blowback from The New York Times public editor. I’m not surprised, the idea of betting on ideas strikes many as transgressive. But what people have to understand is that this isn’t gambling, it’s putting your money where you mouth is. As Alex Tabarrok has stated it is a “tax on bullshit.” I really hate bullshit. I hate it from other people, and I hate it from myself. The more bullshit there is in this world, the less clarity we have about the world around us. Bullshit is the clarity killer. It is the enemy of objectivity.

Over the past few weeks I’ve tried to get people like Dwight E Howell to bet me on their skepticism on the polls. In fact, I’ve come close to harassing them in the comments, chasing them down on Twitter, etc. And yet I’ve received precious few wagers. Why? I think most of them are bullshitters. They might not even know they’re bullshitters consciously, but deep down they don’t have confidence in their views, else they’d be lining up to take my money.

Hank Campbell though stepped up to the plate. If Nate Silver calls 48 or more out of 50 states correctly, I get $50. If not, then he gets $50. Daniel Gonzalez Buitrago also proposed a $40 wager, and I accepted. These bets aren’t just about making money. They’re about sharpening my own sense of what’s true. Am I willing to lose money? My confidence in these wagers is modest, at best. Rather, my point is that if I make enough wagers, and I am confident in enough of them, then I should come out on top. This is how I view my attempts to understand the world. I’m not aiming to understand one fact perfectly; I want to understand many small facts to some approximation. You win some, you lose some.

Finally, if you haven’t read Colby Cosh’s take down of Nate Silver, you should. I like Silver because he’s presents himself as a humble Bayesian, but with great influence comes great expectations. Ultimately what Silver does that’s great is that he popularizes a method of inference and forecasting. But the problem may be that people are more enamored of the result of his current forecast.

CATEGORIZED UNDER: Politics
MORE ABOUT: Politics
  • http://www.exoticjewishhistory.com Moshe Rudner

    And of course you owe me $30 if Romney wins while I owe you $10 if Obama does.

    I’m on my knees praying to any deity with local sway to inspire Obama toward something like a nude chicken dance at his next event. Any additional prayers you may know would be appreciated.

    http://www.exoticjewishhistory.com

  • idea2go

    I’m a fan of Nate Silver but how likely is it he will get 48 out of 50 right? Maybe even money?

    I bet Nate could run some simulations and answer that question!

  • soren

    “Silver attempted to do the same, and got some blowback from The New York Times public editor. ”

    No he didn’t. His mouth was spouting 3:1 odds and he wanted even.

  • David Kane

    Are you still accepting bets? I will bet you $50 at even odds that Nate’s current forecast for Obama’s electoral votes (307) is 10 votes too high. (Feel free to replace this number with his final forecast.) That is, you pay me $50 if Obama wins < 297. If not, I pay you $50. If you think that Nate's forecast is unbiased, then you should accept this bet. If not, then you must have some (implicit) reason for thinking that Nate's model is biased high for Obama.

    I am making this bet because Colby Cosh's take-down makes me doubt the quality of the 538 model and because my own posterior estimates, which I am happy to anchor on 538, have been updated by the views of knowledgeable non-quants like Michael Barone and Jay Cost.

  • April Brown

    I wish I could place a bet, but I really can’t figure out what’s up with the polls. I stare at the numbers and the methodologies and my brain sort of stalls.

  • Bill

    You guys are thinking incorrectly on the odds. Nate Silvers using his stats approach thinks Obama will win the election, and he shows his methods and calculated odds of that occurring. Obviously, he thinks Obama has a greater than 1:1 chance of winning. The pundits that think Romney will win obviously think that Romney has a greater than 1:1 chance of winning. If the bets are made, it should be for even money. If you are asking the bet to be made with 3:1 odds, then you are admitting that Nate Silver’s calculations are correct.

  • azoomer

    @soren

    that’s exactly the point of silver’s bet.
    joe said everybody who’s saying it’s not a toss up should shut up.

    toss up is 1:1.

    if joe really believed that it was a true toss up, he should go for that 1:1 bet.

  • toto

    t if I make enough wagers, and I am confident in enough of them, then I should come out on top. I want to understand many small facts to some approximation. You win some, you lose some.

    Yes, but shouldn’t you make single bets on many different things, rather than several bets on the same thing? Otherwise correlation is going to kill you like a small hedge fund in 2008.

  • simplicio

    @2 He wanted even odds from Joe Scarborough, because Joe said the race was a “toss-up”. If Nate offered the odds he actually thought were correct, the exercise would be pointless as he could only expect to break even. The entire point of the exercise was that he was confident in the odds he published, while Joe S. was full of it.

    I read some commentator (Matthew Yglesis, maybe) make a somewhat tongue-and-cheek proposal that the NYT’s should make all of their pundits put some of their salary in an account that would automatically be used to make bets on any prediction they might make.

  • https://plus.google.com/109962494182694679780/posts Razib Khan

    but shouldn’t you make single bets on many different things, rather than several bets on the same thing? Otherwise correlation is going to kill you like a small hedge fund in 2008.

    i’m hedged. so i did. (there are other bets i haven’t mentioned int his post :-)

    david, not confident enough that he’ll come within 10 electoral votes. i’m tempted though, cuz i’m almost there. but it’s too similar to hank’s vote.

    p.s. don’t post long bullshitty comments trying to interpret what i’m saying. not going to publish it.

  • RocketDoc

    Since you’ve decided not to post my comment, you can read this one and not post it too. I had legitimate questions, and your article is far from clear as to why you think Silver should be held to a different standard. But hey, it’s your column, and if you decide you were perfectly clear, and my attempt to interpret you was “bullshitty”, then more power to you.

  • https://plus.google.com/109962494182694679780/posts Razib Khan

    #11, you totally misread the intent of my comment. that means you didn’t read closely, or, i wasn’t clear, though long time readers will know i’m quite a fan of silver’s work. not interested in having long debates about positions which i don’t hold. and now hopefully new readers will be more gun-shy.

  • TIRM

    The problem with Silver’s analysis is that it is based off the premise that the underlying polling model favors Dems and he is simulating a bunch of results off the same model.

    If the turnout model (D/R split like 2008) then Obama wins with the odds that he is giving. If the split is not like 2008 then Silver will be wrong in many states and Romney will likely win with roughly the opposite odds.

  • https://plus.google.com/109962494182694679780/posts Razib Khan

    #13, how much money are you willing to bet on skepticism? i know your identity so we can make a bet right now! give me specific values so i can make money off your belief. times’ a wastin’

  • dave chamberlin

    I like anything that sharpens my mind and have found small wager betting on football to be a delightful waste of time and money. Suddenly I care if the Peoria Lugnuts can cover the 5 point spread against the Chattanooga Chickenchokers. I wager 10 or 20 bucks on a game knowing full well I’ll be losing in the long run, but each year I get a little better. What I am really doing is spending a dollar to make a football game that I would otherwise find meaningless to be enjoyable. I find gamblers on the whole to be amazingly delusional, they give the deeply religious and the seriously political a real run for their money. The religious believe in deep meaning concentrated, the political believe in simple answers to complex questions and gamblers believe in bad math. I believe in reading words of people that are smarter than me in their respective discipline and loving it till I die. Gambling as 99.9% of the people play it is a fools errand, but blame the player not the game. Nate Silver cut his statistical teeth handicapping baseball before he became a featured New York Times political analyst. Thanks for the plug of two good books, I’ll get around to reading them. Just E-mail me whatever address you want five books sent to if Romney wins, as per our earlier bet, and be patient, finding five books Razib hasn’t read but would enjoy reading will be a real challenge.

  • Chris_T_T

    This could well turn into a reverse 2000 – Obama wins the electoral while losing the popular (particularly if Sandy results in depressed voter turnout in the Northeast). It’s going to be an interesting night tomorrow.

    Looking forward to a break from political commercials for a year.

  • Brian Too

    This of course is one of the founding ideas of capitalism. You may wish many things in the absence of any skin in the game. However once your money is on the line, most people become far more, ah, practical? Pragmatic? Hard-nosed?

  • https://plus.google.com/109962494182694679780/posts Razib Khan
  • http://www.isteve.blogspot Steve Sailer

    azoomer says:

    “if joe really believed that it was a true toss up, he should go for that 1:1 bet.”

    Personally, I’m not that interested in testing the guy on TV’s expertise, I’m interested in testing Silver’s. I would have been more impressed if Silver had offered 2 to 1 odds as a midway point between his model’s 3 to 1 odds and pure guess. But, Silver wouldn’t do that.

  • pconroy

    @Razib,

    Don’t forget our $1 bet!

    I also have a “Romney to Win” bet with Sarah of LW, and there it $1 vs Cup of Coffee – so potentially very favorable odds to me!

  • Neuro-conservative

    Is it possible to make a bet without outing my pseud? I am willing to bet straight up even odds on R vs O (push if PV/EV split) for $50. If you win, I will post $50 to an online paypal or Amazon account or whatever. If I win, you donate the $50 to the ZOA and post a snapshot of the Thank You screen on your blog or twitter account.

    While I would like to admire Silver’s popularization of Bayesianism, I think his electoral model is pure GIGO sprinkled (liberally) with obscurantism and actually does a disservice to the cause of greater clarity and numeracy.

  • https://plus.google.com/109962494182694679780/posts Razib Khan

    I am willing to bet straight up even odds on R vs O (push if PV/EV split) for $50.

    can you elaborate? i don’t get what you’re saying here. and yes, i trust you to honor the bet.

  • Neuro-conservative

    I mean that I will pay if Obama wins both the electoral and popular vote, and you will pay if Romney wins both electoral and popular, but no one pays if one candidate wins the electoral college while the other wins the national popular vote.

  • https://plus.google.com/109962494182694679780/posts Razib Khan

    Agreed. Can you at least email me at contactgnxp-at-gmail-dot-com your real name so I can track you down if you whelch? also, i don’t get exactly what you mean in terms of how i pay you, so include those instructions too!

  • Anthony

    Razib, if it’s not too late, I’ll wager $20 on an Electoral Vote over/under.

    Nate Silver’s electoral vote probabilities show 5 outcomes with a probability of 4% or more, the lowest EV total at about 300 EV for Obama; the 5 total 60%. There are a lot of lower-probability events in the range O > 299, also. I think Silver hasn’t adjusted well enough for the party identification in the polls, and that the odds of Obama getting 300+ EV are less than 50%.

    So my wager is $20 that Obama will receive fewer than 299.5 EV at even odds. Payable via paypal, or other methods at your option. My paypal account is linked to my email address; you can find my real name via whois on my email address.

  • https://plus.google.com/109962494182694679780/posts Razib Khan

    #25, sorry. i am not as confident in silver’s interval as he is. so i won’t take that bet.

  • pconroy

    OMG – I just experienced VOTER SUPPRESSION

    I went to a school nearby to vote, but am registered elsewhere, and while waiting on the Information line heard an East Asian woman being turned away, then I was turned away as well – as neither of us were on the list. But I said that I wanted to make an Affidavit Vote, and was then reluctantly directed to a desk to do that!

    Note: The person on the Info Desk never mentioned this option to the East Asian woman or to me!

  • dave chamberlin

    Just went large on Obama at 4 to 1 at Bovada. Nate Silver has him at 9 to 1 as of now, it is now either a damn good bet or I am going to be very pissed at Nate Silver tommorrow.

  • Anthony

    That’s ok. I spent a while trying to figure out a way to structure a bet that would support my belief that the Ds are being oversampled in most polls (or at least those which Silver is using), but couldn’t come up with anything better than that which wouldn’t be too complex.

  • AG
  • Sandgroper

    Be very suspicious of any interpretations of those data. Look at the numbers for Hong Kong.

  • dave chamberlin

    Silver can make you silver.
    Nate is great. He projected the actual winner in 49 out of 49 states and a tie in Florida. Florida is the one state too close to call at the moment with Obama holding a tiny lead. What is going to happen before the next election is on line gambling will become a new revenue source for the US. I expect future oportunities when right before the election Nate Silver of 538 will again have insights that will allow wagers to be placed that are profitable over the long run. When that happens I will let my friends at GNXP know. My earlier contention that you can’t win in gambling still holds. If you do this get your money in (I will tell you where to go) and then get it all out.

  • http://www.i-finance.co.k Marcus Holland

    So what was the outcome of the wager? Sorry if I missed it from within your post.

  • Sandgroper
NEW ON DISCOVER
OPEN
CITIZEN SCIENCE
ADVERTISEMENT

Discover's Newsletter

Sign up to get the latest science news delivered weekly right to your inbox!

Gene Expression

This blog is about evolution, genetics, genomics and their interstices. Please beware that comments are aggressively moderated. Uncivil or churlish comments will likely get you banned immediately, so make any contribution count!

About Razib Khan

I have degrees in biology and biochemistry, a passion for genetics, history, and philosophy, and shrimp is my favorite food. In relation to nationality I'm a American Northwesterner, in politics I'm a reactionary, and as for religion I have none (I'm an atheist). If you want to know more, see the links at http://www.razib.com

ADVERTISEMENT

See More

ADVERTISEMENT

RSS Razib’s Pinboard

Edifying books

Collapse bottom bar
+

Login to your Account

X
E-mail address:
Password:
Remember me
Forgot your password?
No problem. Click here to have it e-mailed to you.

Not Registered Yet?

Register now for FREE. Registration only takes a few minutes to complete. Register now »