From: Robert Hettinga <rah@shipwright.com>
Date: Mon, 7 Apr 1997 22:53:42 -0700 (PDT)
To: cypherpunks@toad.com
Subject: Re: "...markets are fundamentally chaotic, not efficient"? Really?
In-Reply-To: <v0302092eaf6ea75718f6@[139.167.130.246]>
Message-ID: <v03020981af6f7e81385e@[139.167.130.246]>
MIME-Version: 1.0
Content-Type: text/plain


At 12:36 pm -0400 on 4/7/97, Kent Crispin wrote:

> Nope.  Not any of those things (Gee, there's that "statist" word
> again.  There must be a playbook somewhere).

Yes. It's from the same playbook which says I now get to call you a "twit".

> More a "security
> analyst" notion.  See the cute little book "Chaos and the Capital
> Markets" (I don't have it handy so I can't give you the author.)
> Anyway, he makes an empirical of security prices, and demonstrates
> that they are chaotic.  It was more oriented to practitioners, not
> theorists...

Bark. Chaos theory has nothing to say about the capital markets practically
by definition, but that's a "security analyst" notion, so you probably
wouldn't understand it.

> Chaotic is not the same as stochastic.

Actually, chaotic behavior is a subset of stochastic behavior. I stand by
what I said. Just because a random variable is predictable within certain
parameters doesn't mean that you can call say, every flip of a coin 100
times. When you get to the limit of predictability, you are as "efficient"
as you can get.

> There are obviously stochastic
> factors in markets.  Equally obviously, they are not the only factors.

Sure. What? The devine right of kings? The inherent good of the
surveillance state? I say that stochasticity, including chaos, is the only
factor that matters.

> The fact that markets demonstrate a chaotic element is potentially
> exploitable as a trading strategy.

That statement, is, of course, an another oxymoron, just like "market control".

> I heard that's why trading houses
> were hiring physicists, incidentally -- most of the expertise in
> Chaos Theory was developed in the context of physics.

Rediculous. Most of the "expertise" in chaos "theory" is in the hands of
dillitantes who like to draw pretty pictures. Most of the physicists hired
by Wall Street were people who discovered market analogs to physical
processes. A friend of mine, for example, who used Monte Carlo simulations
to analyze sonar returns, and now uses Green's functions to get faster
results on interest rates and total return scenarios on bonds. Or another,
who did 2D magnetohydrodynamic code once, and now does fun stuff in the
currency markets.

The problem with simulating a market with emergent systems is that you
can't say anything about a given market when the simulation is over. Just
about the simulation. Otherwise (duh?) it's not chaotic enough. Which, by
the way, was my point. When you get to "chaos", you're as "efficient" as
you can go.

That's not to say that using emergent systems technology to create markets
*themselves* isn't useful. Far from it. That's the kind of stuff several
people around here hope to see arise from autonomous entities operating in
cash-settled internet markets, for example. Cash settlement technologies,
by the way, which need strong unescrowed cryptography even to function.

> That's "Crispin", Mr. Applethwaite.  I hate being mislabeled.

Ah. Another ad hominem. See "twit", above. You're ugly, and your mother
dresses *you* funny, too. By the way, I'm sorry I misspelled your name. For
some reason, it seems I didn't respect you enough to get it right...

> I never said anything about planned economies being efficient.  In
> fact, I never mentioned planned economies at all.

You said that "chaotic" markets aren't efficient. I said exactly the
opposite. That they are the ultimate in efficiency, and that if you believe
that chaotic markets are not efficient, you must favor planned economies,
because they're the only alternative. I figure this tendency towards
stringent control must be from where you work or something. Oops. Another
ad hominem. So sorry. They must be in the air this evening.

> Obviously I can't "prove" markets are not efficient -- that's an
> empirical matter, not a mathematical matter.  However, no one can
> prove they are efficient, either (that's why it's the "Efficient
> Market *Hypothesis*").  There are many other examples of persistent
> inefficiencies in markets,  the presidential election year cycle
> being the first to come to mind.

Right. And the increasing American importation of bananas throughout the
20th century caused an increase in suicide. Another informal fallacy.
You're nine for nine tonight, Mr. Ch^hispin. Of course there are actual
exceptions to the efficient market hypothesis, fools' markets being the
most famous example. However, the crash of any given fool's market is
completely unpredictable, and, as such, is as efficient a price as you're
going to get, paradoxically. Which was my point.

> [interesting but besides the point argument deleted]

Actually, it was the *whole* point, but you didn't get it. Another one of
those "security analyst" notions, which kind of blew by you, in what seems
to be a rather breezy evening in your neck of the woods...

> > This inability of a hierarchy to handle as much information or resources as
> > a geodesic is, of course, a major problem with key escrow,
>
> No, it's a theoretical problem that has no impact on practical key recovery
> systems.

Okay. I'll tell you this one, for fun, too, since you don't get *it*
either. In a geodesic network, a single node can not possibly process all
the information in the network. It chokes, and the network routes around
it. You can bet that any key escrow agent would be innundated with
surviellance requests and would eventually become either useless under the
load no matter how big a processor it had. (Hint: it's the same kind of
problem with certificate revocation.) Or, it will be forced to cache its
keys in other locations, making them more insecure. Soon, either everyone
can't get access to the keys because they can't handle the load, or
everyone can because the keys aren't secret anymore. An interesting
corrolary is that even if the keys were available, the surveilling entity
would choke on *its* information load. This is FinCEN's eventual problem.
Maybe we could call it the Freeh paradox, or something.

Anyway, the best you can hope for is some kind of distributed caching
model, based on the distributed trust models we talk about here. But, if
some kind of centralized entity tries to access *those* keys for
surviellance, it chokes even worse than if it had all the keys in the same
place.

> "Whip me! Beat me! Savage me in Cypherpunks!"

Just lay there and take it, slave. You know you love it, or you wouldn't be
here.

Cheers,
Bob Hettinga

-----------------
Robert Hettinga (rah@shipwright.com), Philodox
e$, 44 Farquhar Street, Boston, MA 02131 USA
Lesley Stahl: "You mean *anyone* can set up a web site and compete
               with the New York Times?"
Andrew Kantor: "Yes."  Stahl:  "Isn't that dangerous?"
The e$ Home Page: http://www.shipwright.com/