Sunday, May 19, 2013

Interacting Diagrams


I like to dabble in the stock market. And there is no end to the diagrammatic understandings of the market — from various rules of thumb to intricate and complex mathematical representations — offering me help. Except for occasional flashes, those of divine mathematical complexity and proprietary secrecy, seem to work well enough, but I have never found any within the range of my understanding that are all that helpful. What I need to know is just what all these diagrams seem to methodologically exclude; that is, the subjectivity of who is buying, who is selling, and why.

The single diagram is made up of indices and relationships, analogous to the subject matter, such that certain consequences follow and inferences can be made from them. Being an abstract reduction of the subject matter, the overall purpose of the diagram also has a bearing on its construction and viability. But when more than one diagram is involved the elements needed to represent this kind of situation change. The indices become actors, things to be diagrammatically understood in their own right, and the relationships between them are interactions that touch rather than connect them. It is no longer a question of the overall purpose of one diagram but rather of a number of individual purposes interacting through time.

This is not to say a diagram of the market as a whole, in a mechanistic sense of how it is set up and how it works, is not still not in play. However, much like geography at play in geo-politics, it is the board upon which the game is played. It can force outcomes now and then, and it definitely sets up currents and trends affecting the results, but it is not the game itself. The game is played between the subjects within the situation, subjects who have their own indices and relationships, drawn from their own analogies and based upon their own motives, such that certain consequences follow and they make certain inferences. Not only do these actors change, or evolve, in the course of these interactions; the board itself can change, evolve as it were, as result of what these actors do and how they try to do it. Interacting diagrams require a novelist's narrative that incorporates diagrammatic understandings of all involved. And, while working with a single diagram may be science, or capable of being a science, once more than one diagram are put in motion interacting with each other, understanding what is going to be an art.

Saturday, May 4, 2013

Man versus Machine

Nate Silver has a chapter dealing with the chess match between Gary Kasparov and IBM's chess-playing program Deep Blue in his book The Signal and the Noise: Why So Many Predictions Fail — but Some Don't.  The match was billed as a man-versus-machine confrontation.  The machine working mechanistically, diagrammatically as it were, was matched against a human whose thought is more general and all-encompassing in a, perhaps, non-diagrammatic kind of way.
But computer chess programs can't always see the bigger picture and think strategically.  They are very good at calculating the tactics to achieve some near-term objective but not very strong at determining which of these objectives are most important in the grander scheme of the game. [p. 247]
Kasparov himself felt, after he won Game 1, that Deep Blue had seen only the near-term tactical considerations not longer-term strategic ones.
"Typical computer weakness," Kasparov later said. "I'm sure it was very pleased with the position, but the consequences were too deep for it to judge the position correctly." [p. 249]
But is this really two different kinds of thinking, one diagrammatic and the other not?
 

Or, is this one kind of diagram, a tactical understanding of atomistic indices in precise relationships, versus another kind of diagram, a strategic understanding with vague indices and less specific kinds of relationships?  We could even see the strategic diagram as encompassing and moving beyond the tactical (à la Hegel's aufheben or Kuhn's paradigm shift).  But maybe even this is too much.  Kasparov began to doubt his assessment of Deep Blue's capabilities during this game when Deep Blue made what was for him and his second an inexplicable move:
To see twenty moves ahead in a game as complex as chess was once thought to be impossible for both human beings and computers.  Kasparov’s proudest moment, he once claimed, had come in a match in the Netherlands in 1999, when he had visualized a winning position some fifteen moves in advance.  Deep Blue was thought to be limited to a range of six to eight moves ahead in most cases.  Kasparov and Friedel were not exactly sure what was going on, but what had seemed to casual observers like a random and inexplicable blunder instead seemed to them to reveal great wisdom. [p. 251]
It was his undoing — "Kasparov would never defeat Deep Blue again" [p. 251] — but the distinction between "tactics" and "strategy" seems to amount to no more than the number of moves ahead being considered.  Limits in calculative abilities may require different diagrams, or approaches, but at bottom the diagrams have a quantitative basis.  Silver, in fact, refers to what have to be strategic considerations in Game 2 offering a "three-tenths of a pawn" [fn 36, p. 252] advantage.
 

One difference between man and machine, however, does stand out in Silver's account.  Kasparov was clearly assessing Deep Blue's capabilities, the diagram it was employing, and trying to take advantage of any shortcomings he could find.  Deep Blue, on the other hand, was not attempting to do anything like that.  The programmers, being the humans they are, did design Deep Blue with Kasparov in mind [p. 256], but so far as we know Deep Blue, as a computer program, was not making those kinds of assessments.  Considering how an opponent sees the situation, how it draws its inferences, and what purposes might be guiding it does seem to be a peculiarly human characteristic.  And, it also seems classically human that such an advantage be ironically rich enough to be their undoing as well.