Tuesday, June 22, 2010

Tarde just in time

I'm going to take another crack at this.  Every time I've tried to articulate my theory that the stock market is a machine for ever more closely approximating total randomness, I have fallen short of capturing what I mean by this. 

This time though, I've got a running start.  I'm in the middle of one of the recent Prickly Paradigm Press pamphlets -- Bruno Latour's introduction to the economic theories of Gabriel Tarde: The Science of Pasionate Interests.  I love these little pamphlets.  This one covers some late writings of the brilliant but long forgotten turn of century French sociologist.  Tarde got briefly famous for writing The Laws of Imitation, which turns out to be one the first versions of the theory of mimetics, and he went on to turn the "society as organism" metaphor inside out and the push it into new territory before he was unceremoniously run out of academic favor by Emile Durkheim.  100 years later Durkheim looks like a bond company stooge, and the upper hand, to my mind at least, is on the other foot ... intellectually speaking.  So while he's impossible to find in English, Tarde is worth tracking down if you read French, or even Spanish, and he sounds bizarrely modern to my ears.

Enough background though; vamos al grano.  Tarde's pamphlet contends that economics is not a real science.  Yeah, I know, yawn.  File under overpaid underworked Marxist professorial twit and move on.  But wait, what's interesting about Tarde is why he feels that economics falls short of being a science.  Unlike what you would expect, Tarde is upset that economics is not quantitative enough.  He doesn't complain about how it reduces all of human values to merely monetary calculations and thus fails to take into account the unmeasurable qualities of the human heart.  No, he wants to measure everything.  And in 1902 no less.  He thinks that all value is measurable, it's just not all (directly) measurable in money.  Money is really just a poor indirect proxy for value.  It's like trying to diagnose an illness exclusively by taking someone's temperature.  So to construct a decent theory of production and consumption we need more direct measures of the things that influence these, and cannot just settle for looking at the price and inferring everything else according to some pitiful black box type theory of how individuals operate.  We need to measure how individuals and institutions actually operate, you know, empirically, like we were, um, scientists.  He shoulda worked for Google.

The argument runs something like this.  Science is supposed to be quantitative and empirical investigation.  You take a system and you try to figure out how to measure something about it in a way that allows you to 1) predict the future values of this measurement and 2) have the measurement be useful for something you want to do.  Some scientist types might disagree with (2).  They may claim that true science shouldn't be concerned with practical and useful questions.  They may defend "pure" science.  What they have in mind though is really a political, rather than a philosophical objection.   There is no such thing as "pure" science, because there is no obvious measurement.  When you approach a new system, it's not at all clear what you should be looking at, what you should be measuring.  Maybe you should be measuring how fast the ball you dropped is going when it hits the ground.  Maybe you should be counting how much wind it produces.  Maybe you should care what shape it is.  Maybe you should ask how much heat you can get out of this ball dropping gig.  There's a  million things you can measure.  Some will be fruitful and multiply, and others will be dead ends.  Science progresses like that.  It invents new things to measure and new theories that these quantities would correspond to.  And it invents them, always, according to the two conditions I mentioned. 

If this sounds like a critique of science, you're right in a sense.  The idea that science is Objective and Universal strikes me as ludicrous.  Bunch'a apes running around with slide rules, congratulating themselves on not creating a black hole?  You call that Objective? This stuff is as contingent as it gets.  Which doesn't make it any less objective and real in a simpler sense of course.  Drop shit and it falls.  Regardless of your political position.  I'm not trying to suggest that science is wrong or completely arbitrary, just that it is partial and always fragmentary.  It is an animal behavior pattern.  What else would you expect?

Yes, I aware this could be a long discussion.

Anyhow, enough philosophy of science.  Tarde argues that economists chose the wrong thing to measure.  Or at least, they chose to place too exclusive a value on the one variable they decided to measure.  They did because they based their choice on faulty analogies with physics and chemistry.  Again, it's important to understand that he's not saying that human behavior is inherently less quantifiable than the behavior of balls and atoms.  His point is that physicists chose to measure simple thing about balls, like their mass and position because they could not get inside them to see how these machines actually worked. As these sciences have progressed and their tools for looking inside smaller and smaller balls improved, they discovered a completely different set of forces operating on the inside.  In other words they started off by measuring what they could get their hands on.

Economics, imitating physics, pretended it was on the outside of its object, even though we are all in fact right in the middle of the economy.  The system economics wants to study is so big that it's all around us.  You don't need a microscope to see exchanges of value taking place.  You don't need a particle accelerator to break the economy apart and see how it ticks.  We already have a particle accelerator for the social.  It's called the internet.  We live inside this explosion of data that measures the system at the micro level.  Why not use all this data to create a science of economics which would then be dramatically more empirical and quantifiable and less dogmatic?  Why settle for pretending that humans are nice tractable little black box utility maximizing agents when that doesn't convince anyone?

Tarde's point is that the real economy is filled with real people and that -- unlike what happens in a proton -- we know all kinda of things about what real people want and how they interact.  We can measure these things in clicks or surveys or traffic patterns or church attendance or whatever.  All we have to do is look around. 

Now that I'm so off topic that you quit reading ages ago, I feel like I should back up a bit and bring in the central thesis of  the other Tarde I have read.  It will give us a good analogy for thinking about this stuff. 

The basic idea of Monadology and Sociology is that everything is a society.  I'm still partial to Deleuze and Guattarri's formulation of this as, "politics precedes being", but whatever.  Everything is a society because everything is composed of interacting parts, all the way down to Leibniz's monads, which are less the tiniest subjects than just raw sparks of belief and desire (hence the two parts of the title).  Ignoring the indivisible unity of the monad question for the moment, this thinking makes sociology the queen of the sciences, a startling reversal of the usual way of looking at things.  If everything is a society, then sociology is the study of things.  And one of the images that Tarde goes back to repeatedly in that book is of a brain so large that we could walk around inside of it and observe the society of neurons, as it were, with the naked eye.  If we could do that, we could start to understand the brain as a society, and, running the analogy backwards, to understand society as a sort of brain, or at least a sort of organism. 

You can see how this fits perfectly with his ideas about economics.  Economics, or "psychological economics"  as Tarde later book is entitled, should be the study of the societal brain from the inside.  We walk through the organism we wish to study everyday.  We observe at first hand how the parts of it function.  We only lack the ability to integrate all of our observations into a coherent whole.  That quantitative integration is the science we want to build.

We can keep using the brain analogy to illustrate why just keeping track of money will be insufficient to study the economy.  If you studying a brain, it surely might be useful to know what parts are active at what time.  And you might measure this by blood flow and take pictures of it with an fMRI machine.  But you can't tell me that even the fanciest theory of of predicting the hue of a bunch of red, blue , and green pixels on a map of the brain is really going to fully explain how the system works.  You're going to need to measure all sorts of other things too.  "Activity" is fine, and so is money, but these are tremendously crude measures of the inner workings of value, or information processing.  The economy is like a brain is like a society is like an ecosystem-- and it's a jungle out there.   We lack the science that would enable us to understand systems like this, because the sciences we have all started very very far outside their objects, at a distance where the complexity was impossible to see.  What we need are new measures of all different kinda of things.  We need to measure more than just money, or activity.  We need to get in there and drop some mushrooms and do some electrophysiology.


We actually have a great data set that simultaneously measures all kinds of things about the economy, but economists have traditionally ignored it almost completely.  The market (a misnomer really, as what we have is a set of numbers rather than a unified anything, a fact which dark pools and algorithmic trading is making increasingly obvious) is connected to everything.  It measures everything.  It is indirect squared though.  No number in the market measures any one thing in particular.  All the wires are crossed and everything is jumbled together.  But each number measures a little bit of everything.  Taken together, these numbers are the best data repository we have so far.  It's all in there, somewhere.  Economics in the usual sense, but also psychology, culture, politics, technology, everything.  And it's all perfectly quantitative.  Little numbers, little meters, moving up and down all the time.

There's a book I haven't read, but which sounds interesting (though is reportedly awful, unfortunately) called Investing: The Last Liberal Art.  The idea being that investing is the last refuge of the renaissance man who needs to know a little bit about everything.  But maybe it should be titled "Investing: The First Social Science".  Investing is, after all, the art, or science, take your pick, of predicting these little numbers.  They, in turn, incorporate the whole world, though in a giant jumble (which incidentally is what makes the big casino so much fun)*

Incorporate and reflect it, and of course feed back into it as well.  Which is naturally where your real nasty randomness gets started.  If blowing the the system up to "life-size" proportions helps to measure it easily from the inside, it also means that we are, um, inside the system.  New meters will build are also part of the system, and act on the system, which is how is grows and incorporates ever new vistas of measurement. 

Consider, for example, the "Flash crash" we had a few weeks back.  For a second there, the little numbers went berserk.  P&G spent 7 seconds trading at a nickel.  The weather vane was caught in a hurricane of its own making.  Think how much richer our data set got though.  That was a beautiful way to incorporate the latency of servers, the new programming languages, the growing importance of our computational infrastructure.  It's in the record now.  As is the regulatory response that precisely measures our desire to control the range and time scale of change.  All in the form of numbers, if only we knew how to read them.  Someday we will improve at this.  Though by then the machine will have moved on.  It has to if it is going to continue to incorporate everything.  If has to if it is going to be the increasingly rich raw data for our science of human desire, if it is going to incorporate and perform wider and wilder variations, to support the proliferating diversity of value.  Like the universe, it has to get more and more random every day, just to keep up with us.

I don't know whether I did my original thought justice this time either.  In the hopes that it helps summarize, I leave you with a quote from Freeman Dyson's Infinite in All Directions.

As a working hypothesis to explain the riddle of our existence, I propose that our universe is the most interesting of all possible universes, and our fate as human beings is to make it so.

I Pangloss on acid.

* I'm not claiming to have a science of the markets.  I do, however, know about some swampland in Florida ...

No comments: