From: Dan Fabulich (daniel.fabulich@yale.edu)
Date: Fri Mar 10 2000 - 10:21:25 MST
Eliezer Yudkowsky did say:
> The reason I disagree with this theory is that it's static. Alice
> already has the Apples and Bob already has the Bananas. The theory
> doesn't talk about production, or about where the Apples and Bananas
> come from. The static analysis presents a clear "potential energy
> surface" in which a set of local quantitative valuations is enough to
> easily compute the space of beneficial global transactions, and it's
> likewise easy to navigate to a point in that space using a set of global
> quantitative valuations determined by supply and demand. There's no
> point in complex analysis because the utility of Apples or Bananas is
> simply the joy experienced by the person consuming that Apple or Banana,
> and there's no functional difference between valuing that joy in Cattle
> or valuing it in Diamonds.
> The utility of trade is *specialization*. A smith can produce iron
> weapons more cheaply than a hunter, and an experienced hunter can
> produce meat more cheaply than a smith.
You seem to be arguing two points at once:
First, you're arguing against the "static" analysis which I used when I
presented my argument as to why trade is good, arguing that it fails when
it faces a dynamic situation. But, of course, in my example, there
*aren't* any dynamic effects. In my example, Alice wants Bananas today
just as much as she wanted them yesterday, and just as much as she will
want them tomorrow. The fact that I didn't have changes over time in my
example doesn't mean that I can't take them into account, and doesn't
imply anything at all about *how* I'd take them into account.
Specialization is good, it's true, but I can show that in the standard
analysis with a single currency just as easily as you can with a complex
barter analysis. (This is sufficiently trivial that I won't bother unless
you ask.)
Second, you're arguing against the "utility" model of economic efficiency,
which has some problems, but not the ones you're raising. You seem to
think that, for some reason or other, the utility model can't take into
account the "physical interaction" of the goods in question. This is just
false. The physical interaction is and can be taken into account through
a utility-wise calculation, by specifying, as you suggest, the value of a
Banana to me today and tomorrow (or the value to me of a certain bundle of
goods, as in the case of shoes, where a right shoe is almost no good
without the matching left shoe). So it's not obvious where your objection
lies with regards to utility calculations. If your objection holds, it's
supposed to yield wrong answers somewhere, but you haven't shown where.
Here's your first attempt:
> The way that a single-currency system handles the dynamic problem is to
> talk about interest rates; the utility of the Banana to Alice is 2
> Apples now, or 5 Apples in one year. But that assumes that Alice and
> Bob have already computed the barter; the currency valuation simply
> labels it, *it doesn't produce the transaction*. If Alice and Bob are
> in the Marketplace, trying to buy things with Diamonds, they might
> starve to death where a complex-barter system would keep them alive.
This example doesn't show what you're arguing. Remember, the
single-currency is one of the goods being traded on the market. Let it be
Bananas. In a static situation, if Alice and Bob declare the number of
Bananas they're willing to trade for an Apple, they've just declared all
the information they need. If you're with me there, then notice that
Alice also has enough information to declare the number of Bananas she'll
accept today in trade for Apples tomorrow (or next year, or whatever your
time scale is), namely, she'll trade as many as 5 Apples tomorrow for 3
Bananas today. With that declaration, of course, Alice and Bob are ready
to make a deal. As expected, simply by declaring what they know in terms
of a standard good, the deal happens.
Of course, when there's only two goods, there's no difference between
barter and single-currency, so the point here is moot.
You make a second attempt to give faulty results under the standard
analysis, but qualify it by saying:
> And yes, in this particular scenario, you can always ask why they don't
> sell their existing Apples or Bananas at the reduced price and use them
> to buy Bananas or Apples from Eric or Drexler. But that's just an
> artifact of the way this particular scenario happens to work. The
> problem, in essence, is that Alice, Bob, and Carol's Carrots have been
> priced out of the Marketplace.
Actually, this isn't an artifact of the scenario at all. It's just an
instance of Ricardo's "comparative advantage" theory. It will hold no
matter HOW you figure the numbers. An interesting fact about this theory
is that there's no way to get "priced out of the Marketplace," since
you'll always get some advantage from specialization and trading with the
market. The main factors that get in its way are transaction costs and
starvation-like problems: in some cases, even the added advantage of trade
isn't sufficient to sustain life.
Unfortunately, complex barter doesn't solve starvation in any meaningful
way, nor does it make trades possible which are impossible on the Diamond
system you describe.
> Of course, none of this deals with the *real* advantage of complex
> barter: Complex barter futures, which protect virtuous cycles from
> pricing shocks.
Now we get to the real issue. Price shocks. Yes, I discussed the issue
of local price stability briefly at the end of my long post on barter.
There I pointed out that, of course, we *do* want more than one currency
so as to be able to sustain instability in one currency without affecting
the rest of the economy. In a multi-currency system, like ours, the
currencies are traded via complex barter. An entire economy run on
complex barter would be the extreme case where *all* of the goods would be
treated as currencies; this economy would have maximal price stability in
this domain.
I argued in my last post that the computational cost of adding new
"currencies" (that is, of treating more and more goods via complex barter,
and determining the value of the rest of the goods on the market in terms
of all of those currencies,) is on the order of n^2, while the added
stability diminishes as the number of currencies increases. So imagine
two curves which intercept the y-axis at 0: one, the stability advantage
curve, concave down and starting off with a large initial slope,
eventually reaching a horizontal asymptote, and another, the computational
cost curve, concave up and starting off with a small initial slope. On
the y-axis is cost/advantage. On the x-axis is the number of currencies.
The point where these curves intersect determines the efficient number of
currencies. (Draw this. No, really. It'll take two seconds.) Let "x"
be the efficient number of currencies.
A pure complex barter system (where all goods are traded via complex
barter) is only efficient when x is larger than the number of goods on the
market, though we might consider an economy to be largely organized by
complex barter if x is somewhere on the order of the number of goods on
the market. Right now, with computational & info-transfer costs as they
are, this number is probably no greater than a few dozens worldwide, (I'm
guessing, but if I'm off by an order of magnitude, I'll be Really
Surprised,) whereas the number of goods on the market is a hell of a lot
larger than that.
So, you're right, there are stability advantages to complex barter.
However, by the time computational costs get *that* cheap, I'm expecting a
Singularity of some form or another.
-Dan
-unless you love someone-
-nothing else makes any sense-
e.e. cummings
This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 15:27:17 MST