Sunday, February 21, 2010 14 Comments

One-sided conversation with a headless professor

I don't actually want Professor Hanson's head, of course. I don't want it in a box; I don't want it on my socks. I don't want it stuffed, embalmed, cremated or even cryopreserved. Physically and biologically, it should remain screwed firmly to Professor Hanson's neck - and thence, indirectly, to his chair. A post to which it in fact brings much credit. As I told the Professor at our debate, he's as far ahead of his profession as his profession should be ahead of him.

But a man can only go so far. Metaphorically, only Professor Hanson can prune, pack and ship his own capital peduncle. He can do so by agreeing with me on these two statements:
A prediction market can be used as a decision market if (a) it is accurately integrating genuine distributed information, and (b) it is robust against any causal feedback from its potential decisions.

Assessing (a) and (b) is a nontrivial, generally non-computable task that demands good intuitive judgment, sometimes known as "wisdom."
Thus, a prediction market is not an effective decision market if it is unable to predict accurately due to Knightian uncertainty, asymmetric information, etc, etc. (The uncertainty in Feynman's "Emperor of China's nose" example is Knightian - a prediction market cannot predict the length of the Emperor of China's nose, if no one has ever seen the Emperor of China's nose. And no, Feynman was not making an argument about biometric distributions.)

For instance, a prediction market in terrorist actions should not trade, because it is subject to both Knightian uncertainty (no one really has a computationally accurate model of terrorism) and asymmetric information (except the terrorists, whose predictions of their own actions will always be the best.) Therefore, rational actors will not trade in a terrorism prediction market. It is thus a machine for transferring money from fools to terrorists - a sort of high-tech Islamic relief fund. Alas, it's really quite typical that the Pentagon would fund such a thing.

In addition, whatever the financial strength of the predictive forces in a market, these forces must be stronger than the financial strength of anti-predictive forces, which can profit from a decision error that balances the expected loss of their anti-predictive bets. The cost of manipulating a market is a function of the players in that market; it is not a matter of theory. It can be anywhere from zero to infinity. So can the anti-predictive profit, of course.

The good news is: if Professor Hanson will agree to these declarations, I think we can all agree that he's earned the position of Sith Lord and the honorary title of "Darth." When the DHL van shows up with the box, I'll send out that red-lightsaber kit. Some assembly required. The Professor can start putting it together as soon as his new, Sith head grows back. He doesn't even need to resign his professorship, although the tattoos may arouse some faculty attention.

The bad news is: he probably won't. At least, I've already offered him the chance to turn his path away from the lies of the Jedi Council and the grotesque, stuffed corpse of the dead Republic. And he has already refused. I enclose my side of the conversation, and replace the Professor's with codes - which he can disclose if he wants, and not if he doesn't. I've also elided paragraphs in which I disclose dark secrets of the Sith order.

Subject: head

Let me know when you get the thing off, and I'll send you that DHL label:
Professor Hanson:
Really? I'd be happy, of course, to comment in response if you have a reply.

You might focus on the Popperian question of what empirical evidence would discredit your design - if this empirical evidence does not, or if it is not evidence at all. Because frankly (unless you deny the allegations against Berlacher et al), if this doesn't, I can't imagine what would! Maybe you can fill that imagination in.

[SS #1]
[SS #2]
Professor Hanson:
[SS #3]

If the strategy does not succeed, why do they keep doing it? What is the cause of the "familiar fact pattern" which is "rampant?" I quoted from three separate areas of financial expertise: prosecutors, journalists, and academics. You can't possibly deny that PIPE shorting is rampant.

Therefore, if it is not as you posit successful, the onus is on you to explain this irrational behavior. (In fact, it would probably make a good paper.) I have a perfectly sensible explanation: because it works. It is not irrational, but rational.

And I really wonder what, say, Charles Darwin, would make of your persistent complaints about my word count. Dear Lord Jesus, do we need to think in sound bites just because it's the 21st century? Must we twitter, just because we can?
Professor Hanson:
Laying off, at least if it's this:

is totally different. It's hedging a bet already made. Hedging is not manipulation.

In a PIPE, the bet is not already made, because the stock issue is not yet priced. The goal of manipulation is to affect the price at which the deal closes.

I can answer your question, actually. In a PIPE short, the direction of manipulation is always downward. That is, the PIPE manipulator is always shorting, never pumping. That's why it's a PIPE short, not a PIPE pump (the other direction of manipulation being the pump-and-dump).

As you know, a predictable price error is a predictable profit. So why are there no wolves? I'm not sure. But I suspect that there is no predictable profit, because there is no reliable way for a wolf to see that (a) the PIPE is being shorted by the new investors, and (b) if so how much.

That is, if you assume the PIPE is shorted, the error is predictable. But since the activity is after all illegal, it is not known whether it is done - thus not predictable. Thus there is no predictable profit and various sensible assumptions are not violated.

Are you ready for that DHL label yet? I've already paid for the shipping. You'll just need to perform the procedure.
Me, again:
Also, another reason these markets are not predictable, despite the "rampant" unidirectional manipulation, is that, since it dilutes the company, PIPE shorting actually decreases the value of the shares it attacks. The false prophecy of the ganked decision market is also a self-fulfilling prophecy.

Thus, it is not sufficient to notice a pattern in which PIPE shares decline between the announcement and the closing, or even before the announcement. If you buy in against the shorts and your purchase is not large enough to overcome the manipulators, you will just lose money along with everyone else. You are now a dolphin, too.

Is it impossible, still, to profit from this pattern? I'm sure it isn't. I am not a trader, but I'm sure there is action in the area. There are no trivial strategies, however, or the manipulation would not work and hence would not exist.
Professor Hanson:
I can't imagine what you mean by "increases price accuracy." How would that be measured? What do you even mean by "increases price accuracy?"

Are you saying that PIPE shorting is a self-fulfilling prophecy? In that case, we agree! It is most certainly a self-fulfilling prophecy. My point is that futarchy will be manipulated by those who can profit from self-fulfilling prophecy. Eg, that in the case of PIPE shorting predicts that the stock will decline - and thus dilute it, decreasing its value.

In a self-fulfilling prophecy, there is a circularity: because the prophecy is made, the prophecy comes true, because the prophecy is made. For instance, in one of these PIPE scams, the profit is made because the stock goes down. But the same action that made the stock go down - putting on the shorts - is itself the bet.

My point is that futarchy will be manipulated to be non-predictive, even anti-predictive, in precisely this way. Government will make insane decisions for the sake of funneling profit to self-interested players. It does this already! But in your system, it will do it far more efficiently. Futarchy is basically automated graft.

Therefore, if I am right about what you mean by "ex ante the possibility of such situations increases price accuracy," it is by no means a defence. Quite the contrary - it is a confession. Susceptibility to self-fulfilling prophesies is very much a bug in a decision market. It is not a feature, and I will indeed require that head. (Blame it on too much Japanese history.)

Moreover, I am not at all convinced that this defence is even true. All self-fulfilling prophesies are profitable and (by my definition) false, but not all profitable false prophesies are self-fulfilling. Rather, the bet can take a loss, and the bettor can still win by some other mechanism.

There is an easy way for you to clear up my confusion. If my empirical data, such as it is, does not falsify your theory - what would?
Me again:
Any more thoughts? If not, I will post this exchange, or at least my side of it. I'll also include your side if you give me permission.

In my (basically industry-centric) view, accepting the limits of prediction markets makes an argument *for* prediction markets. Obscuring these limits is an argument against them. So, if you invent the suspension bridge, and then claim that the suspension bridge can be used to bridge the Atlantic, you are sabotaging your own invention.

A prediction market can be used as a decision market if (a) it is operating correctly and accurately integrating genuine public information, and (b) it is robust against any feedback process involving the decisions it makes. Assessing (a) and (b) is a nontrivial task that may even demand good intuitive judgment, or "wisdom."

If you can accept this result, I will harass you no more!
- and here the matter resteth.


Anonymous Anonymous said...


February 21, 2010 at 4:55 PM  
Anonymous Robin Hanson said...

I think you don't even understand the claim I'm making, so you don't understand what would or wouldn't disprove that claim. I don't see these blog posts as working to help you to understand - we need to have an actual conversation, phone, IM, in person, something. Without I don't have much hope for communication.

February 21, 2010 at 5:17 PM  
Anonymous Anonymous said...

OK, heres my mencius inspired poem i just wrote

i used to love my unicorn
he looked the part, a fine strong steed
undappled sides and mane fresh shorn
the very sight inspired greed

his flaw was hidden in the plain:
the twisted spear upon his head,
afixed above his fevered brain,
was over-apt to leave men dead


it was supposed to be about the democratic state, but now i think it might be more about mencius (except i still love him (no homo))

February 21, 2010 at 7:32 PM  
Anonymous josh said...

Is the claim anything more than that using decision markets to make decisions at the sovereign level would lead to good governance?

If not, it seems trivial to show that this is, in fact, retarded.

February 22, 2010 at 5:29 AM  
Anonymous Anonymous said...

The claim is that it would be better than the present US regime.

February 22, 2010 at 3:49 PM  
Anonymous Pals said...

"The claim is that it would be better than the present US regime."

Farting into a Ouija board would produce decisions far better than those of the present US regime.

I think it's worth our while to take a minute to admire the majesty of what Moldbug has done here. Robin Hanson is by far the most creative, original, honest and interesting academic in America today. His claim to fame is that he does not fear ruffling feathers and that he speaks his mind. The man's life's work is about defeating all forms of bias and thinking honestly.

And yet, when it comes to his own academic ideas, the guy becomes as retarded and tired a hack as Brad deLong, Sean Hannity or Nancy Pelosi.

Down with the University!

February 23, 2010 at 12:04 AM  
Anonymous Leonard said...

as retarded and tired a hack as Brad deLong, Sean Hannity or Nancy Pelosi.

Come now. There are degrees.

February 23, 2010 at 7:13 AM  
Anonymous Anonymous said...


Mencius, I personally think you can leave off now. While entertaining, these exchanges with Professor Hanson no longer seem to be accomplishing anything constructive. You have made your case quite explicitly. Either you are correct, and Professor Hanson is wrong, or as he indicates in his comment, you two are not even speaking the same language, in which case this exercise is pointless. And consider the Sapir-Whorf hypothesis: learning his language is not necessarily conducive to improved logical cerebation.

You should be wary of accepting his suggestion that you speak in person rather than through written exchange. There is no logical argument which can be made better in person or orally than it can be written. Professor Hanson, however, is (in part) in the business of selling his ideas to key decision-makers, and a good salesman knows that the best pitch is made in person. That said, beers with Robin Hanson would obviously be fantastic fun - just remember to keep your "whether" eye open (Whether he is charming and intelligent or not, remember to listen to your own counsel.)

-- Dirtyrottenvarmint

February 23, 2010 at 7:14 AM  
Blogger Aretae said...

Three responses.

1. My read of Mencius here, and over the related posts was the same as Robin's (as indicated by his comment).

@Pals -- magnificent opener. However, my experience suggests that defending (hell, even clearly explaining) your own independently developed ideas against all comers is several orders of magnitude harder than (a) attacking ANY set of ideas, or (b) defending old ideas that have been kicked about for a million man-hours. If this is true, your criticism is unfair. On the other hand, I figure the academy has 15 more years before it's as dead as newspapers.

@ANONYMOUS Sapir-Whorf. Good call on the stop-now. You also claim that "There is no logical argument which can be made better in person or orally". Perhaps that is narrowly true. However, it assumes 2 things...1. We're doing formal logical arguments, not understanding of other positions. 2. Logical argument is an effective way to do this kind of communication.
However, most communications theorists suggest that communication is a 2-way business with feedback essential. If there is actually a misunderstanding, then in-person interaction is a much better path to finding the problem in understanding.
Also, many people (myself included) are much more effective communicators talking than writing. This doesn't mean that there is no danger of persuasion...but that type of persuasion isn't Robin's specialty.

February 23, 2010 at 10:37 PM  
Anonymous Anonymous said...

I have to say I agree with the above two comments. That's enough on the futarchy & sparring with Robin Hanson, it's starting to get a little childish.

If there's one thing I wish you'd write about, it's concrete implementation plans for your own perfect government. How it would work, the separations and functions of departments/levels, the duties of the people and the mechanics of the stock ownership, etc. I'd be fascinated to read that, as it would provide a real foundation for thought and discussion, rather than just beating the dead horse of your dispute with the good professor.

February 24, 2010 at 12:03 AM  
Anonymous Anonymous said...

I have to say I agree with the above two comments. That's enough on the futarchy & sparring with Robin Hanson, it's starting to get a little childish.

If there's one thing I wish you'd write about, it's concrete implementation plans for your own perfect government.

I don't know about childish, but when I zzzzz when I check this blog and it is still on econ. Moar American Revolutionaries: Cesspool of Puritanical Blowhards. That's gold, Jerry. Gold!

Scott W.

February 25, 2010 at 7:11 AM  
Blogger TGGP said...

Knightian uncertainty was one of the major points in dispute in the Caplan-Boettke debate. People familiar with decision theory and Bayesian probability don't bother with the concept. That Feynman didn't consider biometric distributions is a mark against Feynman, not support for Knightian uncertainty. If we were really that uncertain we would have no idea if the nose was as big as Manhattan. In fact, our information is much better than maximum entropy.

"Asymmetric information" is not an argument for why a trade SHOULD not happen. It argues that it WILL not happen. And the adverse selection resulting from it does not apply to the markets it was crafted to describe.

"no one really has a computationally accurate model of terrorism"
Does anyone have a "computationally accurate model" for most of the things we have speculative trading in? The information to be aggregated is greater than zero, that's good enough.

"It is thus a machine for transferring money from fools to terrorists"
Terrorists already have ways of making money off attacks, but don't normally engage in them (perhaps because that would reveal information early and possibly trace back to them). But if terrorists actually did participate and provide accurate information that would be BENEFICIAL!

"Alas, it's really quite typical that the Pentagon would fund such a thing."
I recall you praising the wisdom of Congress for putting the kibosh on it. Of course in my view they were just grandstanding about how nobody should be gambling on death rather than arguing that it would be defective. And in the original proposal trading would have been restricted to employees of our information agencies, as you recommended for fire-the-CEO. I doubt Robin thought that restriction was a good idea, just that it was an easier sell to said agencies.

"The cost of manipulating a market is a function of the players in that market; it is not a matter of theory"
As Friedman told you, everyone who likes making money is a potential player in the market. The number of participants is not some stochastic variable, but INCREASES if there is more money to be made from anti-predictive "sheep".

"It can be anywhere from zero to infinity"
No, the maximum cost is determined by the available funds that can bet on the other side. So 50% of the world's wealth should be the ceiling.

In the debate Hanson said exactly what empirical evidence would be relevant: the performance of decision markets. If firing the CEO based on his market leads such companies going bankrupt, I'm sure Hanson would admit that falsifies his theory. On the other hand, Mencius has said NO EMPIRICAL EVIDENCE can ever convince him, the philosopher disdaining induction for deduction. Even if Uncle Sam creates a utopia through the use of futarchy, Mencius indicates that would not change his mind.

"My point is that futarchy will be manipulated by those who can profit from self-fulfilling prophecy"
Futarchy for a corporation would mean implementing decisions that increase stock price. If a "self-fulfilling prophecy" increases the stock price, that's good! For a nation, it would be GDP+. There are certainly many complaints you can make about GDP+, but they don't have anything to do with PIPE shorting.

"(by my definition) false"
How is it false?

February 25, 2010 at 3:04 PM  
Blogger TGGP said...

In the off-chance you're still reading this, why not give the ok to make public your responses?

There are a great many existing governments worse than the U.S, and countless more have existed in the past. Remember Callahan's law.

"And yet, when it comes to his own academic ideas, the guy becomes as retarded and tired a hack as Brad deLong, Sean Hannity or Nancy Pelosi."
Most people who read him don't diagnose him Rothbard's law. Perhaps he is just generally retarded and you give him too much credit.

Dirtyrottenvarmint (thanks for signing your name):
"And consider the Sapir-Whorf hypothesis"
Or don't, it is just a hypothesis, and according to Pinker in "The Stuff of Thought", quite wrong.

For those interested, Paul Hewitt's big critique of futarchy is here. Hard to boil down to essentials, but he gets credit for citing the historical accuracy of actual prediction markets. I recently had a long back-and-forth with Paul about differential productivity for racial/gender/other groups that was full of miscommunication, illustrating Aretae's point above.

February 25, 2010 at 3:05 PM  
Anonymous Anonymous said...


I hope you weren't implying that I think prediction markets are historically accurate. My point, in the article on futarchy, is that prediction markets are notoriously inaccurate for anything other than very short term predictions (which are, of course, useless for futarchy-type decisions).

Yes, we did have difficulty communicating, but I don't believe it was the result of miscommunication. Your understanding of signaling theory was incomplete, which made it impossible to understand the arguments. To correct this, I have an exercise for you.

Recently, on Robin's blog, there was a topic in which someone mentioned that, if "bad" prayers are answered by God with catastrophes, such as the Great Flood, perhaps prayers should be regulated. Robin, "agreed", but did qualify this because it would be impossible to regulate prayer.

The question for you is, even if it were possible to regulate prayers and assuming God is the vindictive sort, does it make any sense, at all, to regulate prayer?

Hint: use signaling theory.

Let me know when you have your answer.

March 6, 2010 at 6:35 PM  

Post a Comment

Subscribe to Post Comments [Atom]

<< Home