Wednesday, November 17, 2010

Abstract Geography

I've never really considered myself much of a history buff, which is why I've spent some time since returning from the Middle East pondering just what it is I went over there to see.  It's been several years now that the beginnings of urbanization have interested me, even though I haven't put much effort into reading about it and have only picked up a few things here and there.  So, if it's not some general cultural history I'm keen on ... if I don't much care who conquered the Hittites, what exactly am I looking for here?

I think the fundamental thing that interests me is the transition from a nomadic hunting and gathering economy to an urban civilization.  Once settlement has gotten going and labor specialized and divided, the rest is just a list of Empires won and lost, gods and priests and pharaohs served and betrayed, weapons and pots gradually improved.  The coming together is what I find fascinating.  How did a bunch of roving neo-monkeys launch this revolution?

The first time I can remember thinking about this stuff was when we read Lewis Mumford's The City in History for book club.  I remember this as a great sprawling account of urbanization down through the millenia, written from the perspective of someone who insisted that technology (and what is the city but a bit of social technology?) should be made to serve human ends rather than its own.  It was more than anything, a history of the way our technology gets the upper hand.  Since book club languished ten years ago already (alas), I decided to go back and reread some of the early chapters where Mumford speculates about the precursors to urbanization.  True to form for thinking with serious firepower, it's lost nothing of its shine in the intervening decade (or for that matter since the 60's, when it was written). 

Right off the bat, Mumford observes that settled life may have begun with the dead.  After all, they don't move around much.  A peripatetic tribe can at least be counted on to return to wherever they buried their ancestors.  The first city was the city of the dead, and it remained at the center of urbanization for years to come.  His comment resonated with the simple observation that the little bit of the fertile crescent I saw was littered with tombs.  And while Herr Doctor pointed out to me that the necropolis never the literal physical center of the city, it is also never far off; it seems plausible to attribute a central psychological significance to the city of the dead.  It holds our first dearly guarded but rarely accessed links to the past, though it's more important, perhaps, for the gatherings of the living it facilitates than for anything the dead themselves per se.  These gatherings may have existed as periodic rituals and pilgrimages for a long time before the echo of death's knell resonated with permanent habitation. 

Fast forward a bit.

Last week I went to an investor day for Equinix, a company that builds giant data centers to house internet servers (otherwise known as colocation, aka, cities for computers).  With more and more stuff going to "the cloud", the explosion in virtualization, and the monster bandwidth requirements and latency intolerance of video over IP, it's boom time for the data center industry.  As an investor though, you have to be careful, especially when you understand all this techno-babble as poorly as I do.  The demand for colocation space may be there, but if the idea is just to build a concrete bunker with a long power cord and fat pipe, it seems like it should be pretty easy to bring new supply to market.  Indeed, essentially the entire industry went bankrupt in 2002 despite the fact that there was only a mild slowdown in the demand for space and bandwidth. 

Equinix is only interesting as an investment because some portion of their business is related to customers who use their facilities to directly connect their traffic to other customers without having to route it through all the various hops that a typical packet can experience in an increasingly overloaded best effort network.  The more direct routing reduces latency and improves reliability, and is thus worth big money to a guy running a snappy Web 2.0 app, or, you guessed it, to an algorithmic trader who makes money every microsecond.  This sets up a network effect where new customers show up simply because the people they want to connect with are already there, not because of anything intrinsic to the place where they are meeting.  Metcalf's Law, owning the most liquid marketplace, eating at strip clubs, blah blah blah, you get the idea.  Equinix essentially runs a digital souq. Put a tollbooth on something like this and hire yourself a marketing department and it's easy money. 

Which and so but this brings me right back to my interest in transitions.  When you see network effects and feedback loops like Equinix's data centers or some Assyrian urban agglomeration, the typical concept of cause and effect doesn't really do them justice.  It's not history any more. These phenomena don't have proper necessary and sufficient causes.  It would be better to say that they are seeded by some prior state of affairs.  Or to complicate the crystal analogy, we could call these, a la Cymatics, patterns created by the resonant coupling of some driver with the characteristics of a particular medium (if you haven't already heard me wax cosmological about Cymatics, you should go check out the videos for yourself). 

I swear I'm being difficult for a reason here.  Because it's easy to fall into the trap of thinking of these things too causally and linearly.  First they buried a few people here.  Then they moved in permanently.  Or first  they found a well or an oasis or a defensible hill, then they built a citadel.  Or in the case of Equinix, it turns out that these data centers originally became valuable because they were neutral, third-party-owned places where the competing Tier 1 internet backbone providers could trust one another to come together and exchange traffic.  One network effect leads to another, and now they are popular with financial exchanges and streaming media providers.

I don't want to dispute the importance of these initial facts to what happened later.  Simple geography clearly has a great deal to do with where ancient cities are situated.  And you can hear the echo of New York's perfect port in the data centers built for the stock exchange only miles away from where it has stood since the 1700's.  I just think its wrong to think of these factors as causes when they are more Archduke Ferdinand style triggers.  Of course, I don't mean to abandon determinism with this, just to complicate the simple stories we often fabricate to convince ourselves that one link in the chain leads inevitably to the next.  Perhaps this is why I find these moments of transition so interesting.  In retrospect, everything looks obvious.  This is where two rivers came together.  This was on caravan crossroads.  Washington DC was unused swamp precisely halfway between important places. This data center was near enough to a big urban population.  And yet if everything is so obvious, why are these things are so difficult to predict in advance? 

There is a bit of magic in the resonance between one level of organization and the next.  There's something creative there.  There are new needs that pick up the detritus of the past and adapt it to the new purposes in unpredictable ways.  Maybe we should plant some of those felafel trees next to where we buried granddad?  Maybe we should put a data center in Siberia instead of right next to all those people?  If history didn't have some creative element in it, it would all be determined by geography, all from the beginning, like some giant billiard shot.  And there would be no such thing as qualitative change.  Once some crystal had formed it would expand forever uniformly.  I am suddenly reminded of The Grid Book, which described how the very innovation and productivity that every standard fosters actually creates new forces that end up undermining or at least decoding its own logic.  History isn't unpredictable only because some chance external force meddles with its causal crystalline perfection.  The mechanism actively generates randomness, which is what makes growth and decay, coming together and falling apart, such interesting phenomena.

Having said all that, I must admit that I started writing this because of a simple analogy which suggests that even history's creativity my have a sort of language to its patterns.  Today, another wave of change is sweeping over Equinix's data centers.  Some of the big customers like Google and Facebook are moving a lot of their servers out to their own data centers because this is much cheaper over the long run.  They still leave some machines with Equinix though, because of the latency issues; the bulk of a popular Youtube video might live in a Siberian data center built on a cheap geothermal power source, but chunks of it are still cached closer to the people likely to request it, to make sure it plays smoothly.  Facebook's monster collection of photos apparently works the same way. 

Data center ROM becomes the mirror of our own long-term memories, and these stack up like so many ancestors.  Eventually we push the necropolis aside to make way for the interactions we want to have every day.  These companies bring only their freshest produce to market, and leave our digital dead buried like memories in an abstract geography.

No comments: