Bryan Caplan  

I Changed My Mind About Betting

Am I an ideologue?... China is rapidly shifting to a...
The most profound way I changed my mind:

When I was a child, I thought betting was like playing with matches.  Only fools did either.  The habit stuck through my twenties until I met Robin Hanson

Hanging out with Robin didn't just make me lose my prejudice against betting.  He led me to my current view: intellectuals who refuse to bet on their beliefs are tricksters, frauds, phonies, eels, and blowhards. 

In slogan form: I used to think people who bet were fools.  Now I think people who don't bet are knaves.

COMMENTS (23 to date)
Thomas writes:

"Knaves" is rather harsh. Betting depends on two things. One is a person's confidence in his beliefs. The other is a person's risk aversion (or lack of it). I may be very confident in a belief (though not 100-percent certain of it), but also very risk-averse. As a result, if I'm less than 100-percent certain of a belief (e.g., the spelling of "antidisestablishmentarianism" or the chemical compound of which water normally consists) I won't bet even if the odds offered are extremely attractive. In the end, I will either win or lose. As a very risk-averse person, I'd rather not lose, even if I might win a lot because winning a lot isn't a sure thing.

In sum, you're assuming that everyone else should share your psychological traits. That strikes me as arrogant and even anti-libertarian.

Sam writes:


1) Being *very* risk averse is probably irrational. And probably those who cite risk aversion as an excuse for not betting are being inconsistent. Surely a lot of those people hold stocks in their 401(k) accounts...

2) Even at an extreme level of risk aversion, there is almost certainly *some* set of odds at which you're willing to bet, even on propositions on which you are less than 100% certain.

Combining 1 and 2, Bryan's diagnosis of "knaves" is actually pretty reasonable. A more logical explanation for why people refuse to bet is that they are actually less confident than they would like to profess. So in a sense, they're liars.

SMack writes:

Please take this in the nicest way, Bryan, for it comes from a fan.

Given the trivial sums you wager, self-congratulation seems well out of order. You risk a little bit of money on your opinions, over a long period of time.

What was your recent bet with Murphy? $100 on a five-year proposition? What's that as a percentage of your income over the same period?

So okay, this makes you better than people who use weasel words to avoid making testable predictions at all. And it makes you very slightly better than people who make such predictions but risk nothing. Yet it doesn't make you so much better that it yields a right to boast of some uncommon intellectual discipline.

Rather it seems to say something like: "I feel so strongly about this belief, I will risk one family dinner every half-decade! And let eternal shame fall on those who chatter from the cheap seats!"

Betting on this level seems less to do with futarchy, and more like the trifling of Mr. Randolph and Mr. Mortimer. Shall we say the usual amount?

blink writes:

Betting signals non-conformity given the current equilibrium. It is certainly unfair to call someone a trickster or blowhard simply because they prefer not to antagonize their associates.

Moreover, what is so special about betting for money? Whenever we make precise, testable claims, we put our reputation on the line, at least to a small degree. This is the margin we ought to improve: Clarify of predictions, perhaps even with confidence intervals.

James writes:


If you believe some claim to be true but your confidence in that claim is too low to overcome your risk aversion, that's fine. However, if your confidence is too low to overcome your risk aversion, then you also should not be confident enough to insist that other people take the risks that come with living as if they believe your claim were true.

Or from a more libertarian point of view, if you say your confidence in some claim is not sufficient to overcome yor risk aversion, but your confidence is sufficient to support a policy in which the state curtails people's freedom, you have a problem: You are willing to put other people's freedom at risk for reasons that you would consider insufficient grounds for putting a tiny fraction of your net worth at risk.

John S writes:

Given the trivial sums you wager, self-congratulation seems well out of order.

I'm with SMack. Do your total outstanding bets even come out to over $1,000? Compared to your salary, the amounts you wager are a pittance.

It's fine if you want to say "I think more academics should bet their beliefs," but to brag about your "betting" and call people names is laughable. They'd mock you mercilessly on for these micro-prop bets.

Greg G writes:

>---"Given the trivial sums you wager, self-congratulation seems well out of order."

That's a fair point but, in a different sense, this public self-congratulation is the very thing that makes these bets meaningful.

The fact that a public humiliation is likely to come with a losing bet is the real cost to a bad bet.

Usually, bad predictions are quickly forgotten. Public bets make sure they aren't. If no one is really hurt financially, that's OK with me.

John S writes:

The fact that a public humiliation is likely to come with a losing bet is the real cost to a bad bet.

Greg G, I respectfully disagree. Caplan shielded himself from humiliation with his own "Bettor's Oath," which states:

"When I win a bet, I will not shame my opponent, for a betting loser has far more honor than the mass of men who live by loose and idle talk."

I presume he expects the same from his opponents.

In terms of reputation risk, what Caplan is really doing is freerolling. If he wins, he basks in his predictive genius. If he loses, he still implicitly claims credit for being brave enough to take the bet in the first place.

I don't buy the idea that the financial aspect doesn't matter (*IF* one insists on calling non-bettors "tricksters, frauds, phonies, eels, and blowhards"). I'm not saying he has to put his kids' college funds on the line, but how can one claim to be a brave gambler without actually putting something significant at risk?

The bets he's posted are entirely lacking in potential sting.

anomdebus writes:

I don't bet, but nor do I give advice. Can I say anything without being damned?

To backtrack a bit, I bet against being wrong or ignorant whenever I type. I feel bad when I err, so didn't I lose those bets? How can we compare if my losses means more to me than any of your monetary bets?

Greg G writes:

John S.

I don't really disagree with anything you wrote. You are only describing part of the story though.

I do think that the knowledge that some real cash money will change hands, and that bad predictions will be revisited, has some real positive value in suppressing B.S. predictions.

I am probably closer to you than Bryan about just how heroic this act of betting really is in the larger scheme of things.

Aaron writes:

I understand the complaints that Bryan isn't betting huge sums of money, but in my experience getting people to give me 10:1 odds for bet $10 on an issue just moments ago they were 100% certain about is like pulling teeth. Sure $100 is a small portion of Bryan's income, but getting people to put up *any* amount of money, even with favorable odds, can be hard. As other commenters have noted, I think many people use risk aversion allow them to pump up their certainty risk-free.

(That being said, I'm not sure how much I would bet on that proposition. I guess it would rest on how the outcome was measured).

All in all, maybe not betting doesn't make a person a "knave" or "eel", but it might make them a "fraud" and it certainly makes them a "blowhard"

I also favor the idea of betting where a bet can be adequately specified. But I question whether concrete observables necessary for betting can be found in a field of study which is still in early stages of development.

To explain, arguments have been made by Kuhn and Feyerabend against the method of falsification advanced by Popper. Concepts are necessarily fuzzy, poorly articulated, and perhaps not yet named, in the early development of a science. Only in later phases, during Kuhn's period of normal science, do practitioners develop a language specific enough to enable falsification. Betting, as I now understand it, would not be possible in early phases.

This relates to my own project. I am trying to call attention to the Resource Patterns Model of Life (RPM), a model with great powers of explanation but which elicits yawns (at best) from others. I wish I could formulate some prediction from RPM upon which I could bet. But I doubt such specifics are possible this early in RPM's life, and, my weakness, I do not seem gifted to formulate bet-able predictions.

I may be willing to bet with Bryan, that he can not formulate a bet-able and significant prediction from RPM. If I lost this bet I would win the bet-able prediction.

SMack writes:

@ Greg G

That's what I was getting at with my Randolph and Mortimer Duke comment. If the bet is purely reputational, then just put up a single dollar and let no one be confused. Indeed, just call it a "public prediction" and don't use the word "bet" at all.

Because calling it a bet and then wagering only $100 feels like an attempt to make it seem as though money is being put where mouths are, when in fact the money isn't really doing any work, and the mouthing is all that matters.

Bear in mind, I already conceded that Caplan's willingness to make clear statements like "inflation won't go above X in Y years" is better than the vast majority of pundits who only ever say slippery things like "If Bernanke can keep steady at the controls, and if China doesn't pull any surprises, inflation shouldn't go above X".

What I'm saying is this: the plainness and falsifiability of Caplan's statements puts him way ahead on public intellectual honesty. Why turn around and ruin that by using the word "bet" in such an argghh-I-guess-that's-technically-accurate-but-just-barely sort of way? More importantly, why BRAG about the discipline which risking money has brought to your ideas, when you're not actually risking an appreciable sum of money?

Greg G writes:


Yeah, we are pretty much on the same page. I loved your Randolph and Mortimer Duke reference. That one cuts both ways though. Their bet turned out to be a lot more consequential than they expected.

John S writes:

Actually, Caplan's bet amounts are even less impressive than I initially thought. Let's generously assume he has a modest (and hardly brag-worthy) total of $1,000 outstanding in a given year.

Unless he's a spectacularly poor predictor, the expected value of his portfolio will be a lot closer to zero than -1,000. So he'd have to bet in the thousands per year to stand a realistic chance of losing more than a few family dinners per year on net.

He's hedged psychologically due to his Bettor's Oath, and the monetary risk is *truly* nil. So what exactly is he risking that gives him the right to boast?

Mark Bahner writes:
As a very risk-averse person, I'd rather not lose, even if I might win a lot because winning a lot isn't a sure thing.

The amount of money means nothing, as far as I'm concerned. Stephen Hawking and colleagues have bet on many things, with typically very small amounts at stake Total Baseball, the Ultimate Encyclopedia of Baseball...or even $1. It's the principle of the thing.

P.S. I think Long Bets is a particularly useful concept, and I have three predictions on that website.

Thomas Sewell writes:

The point of the bet isn't the money at risk, it's the way it requires the participants to distill their opinion down into something concrete enough for both sides to to bet on.

The "both sides" feature is important (and a big improvement over a mere prediction) because it removes the possibility of a straw man or other misrepresentation of the opposing side, as well as creating a measurable indication of who was right which can't be disputed later.

25+ years later, we'd have a lot more people arguing Ehrlich was right, except it's a matter of recorded history now who won the bet.

Rainmount writes:

Put up too much money and they can claim risk aversion, put up too little and they can say the amount isn't meaningful. But of course they're free to make a counteroffer.

SMack writes:

To everyone saying "the amount doesn't matter", two quick things:

1) Of course it does. The size of a bet is generally a good measure of one's confidence in the underlying proposition. $100 doesn't reflect as much confidence as $100,000.

2) In this case it definitely matters, for Bryan himself opened the door. He didn't say Hanson had changed his mind about "staking his reputation on falsifiable public predictions". He said that Hanson changed his mind about "betting".

anomdebus writes:

I would like to point out to those trying to defend Caplan, that he does not call out non-betting blowhards, but all non-bettors.

John S writes:

SMack, I definitely agree that betting amounts matter. But the thing that irks me most in this case is the name-calling and snootiness.

I wouldn't have a problem with advocating betting modest amounts if it were done in a dignified way (I may be wrong, but I get the impression Hanson isn't as self-righteous about betting; Surowiecki and Scott Sumner are also classy advocates for prediction markets).

But if you're going to deride and boast, you darn well better have something to show for it. This post is like saying, "Men who shy away from physical violence are cowards. Who else is out there thumb-wrestling like me?"

Eric Hammer writes:

I think those arguing against Bryan are forgetting just how hard it is for him to find people to take him up on his bets. He has in the past proposed very clear bets (in the sense that results are easily determined after the fact) with good odds for those who believe he is mistaken, and still found very, very few people willing to take him up.

His point is that people claim certainty, or at least speak as though they had 90%+ certainty in their claims, but then won't take 10:1 odds in favor of their being right in a bet. That suggests that those folks are either way overstating their certainty or are hugely risk adverse.

Considering further that those people are making claims that hugely affect other people's lives (assuming he is discussing the results of government programs for instance), then yes, those who don't bet are knaves. The knave claims certainty, but then won't take a tiny risk to claim a relatively large gain based on their claim.

Caplan probably should have written that as "I used to think people who bet were fools; now I think intellectuals who don't bet when they can are knaves." It isn't all people he means, but those who claim great knowledge and certainty who refuse to put any skin in the game, even a tenner.

Adriana writes:

If the amount is so trivial, then why won't more people bet? The amount doesn't seem like a mark against Bryan, but rather against those who make bold claims and predictions without being willing to take a small risk.

I'd also point out that these bets aren't intended to be a money-making venture, so they need not and even should not be exorbitant. A large bet would, in my view, seemingly undermine the friendliness, spirit, and purpose behind the bet. Modest stakes are all that's required to drive the point home. And clearly modest stakes are enough to discourage a great number from participating.

Comments for this entry have been closed
Return to top