• Choose a category
• All
• Classic-Fiction

Such a cognitive theory, if sound, would justify the alienation of judgment from skilled professionals when things get too complex. But, in fact, it is often the case that when things get really hairy, you want an experienced human being in control. The preference for algorithms over intuitive judgments, when faced with causes that “lie in the very structure (i.e. pattern)” of a system, is precisely the wrong conclusion to draw if one gives due regard to the tacit dimension of knowledge.

The Tacit Knowledge of the Firefighter and the Chess Master

The basic idea of tacit knowledge is that we know more than we can say, and certainly more than we can specify in a formu laic way. Intuitive judgments of complex systems, especially those made by experts, such as an experienced firefighter, are sometimes richer than can be captured by any set of algorithms.

The psychologist Gary Klein has studied the decision making of firefighters and other experts who perform complex tasks in the real world. “In many dynamic, uncertain, and fast-paced environments, there is no single right way to make decisions,” Klein says. “Experts learn to perceive things that are invisible to novices, such as the characteristics of a typical situation.”7

The experienced mind can get good at integrating an extraordinarily large number of variables and detecting a coherent pattern. It is the pattern that is attended to, not the individual variables. Our ability to make good judgments is holistic in character, and arises from repeated confrontations with real things: comprehensive entities that are grasped all at once, in a manner that may be incapable of explicit articulation.8 This tacit dimension of knowledge puts limits on the reduction of jobs to rule following. It is not just the firefighter’s intervention that is inherently in situ (as the economist Alan Blinder would point out). His knowledge, too, arises in particular places: places where there are fires.

Algorithms can be made to simulate the kind of tacit knowledge that experts possess, as when IBM’s Deep Blue succeeded in playing chess at the highest level in 1997. Through brute computation of every possible move that adheres to the rules of chess (200 million board positions per second), the program was able to pick winning moves. To constrain the problem, the programmers made it their goal to beat one man in particular, Gary Kasparov, the reigning champion. Knowing his preferred opening moves and strategies made the problem tractable. But in beating Kasparov at his own game, Deep Blue was doing something very different than what a human chess player does. This is illustrated by an experiment in which an international chess master played speed chess with a limit of five seconds per move while also doing mental arithmetic. The arithmetic tied up his working memory and capacity for explicit analysis, yet he was still able to “more than hold his own” against “a slightly weaker, but master-level player.”9 Clearly, human chess players are doing something other than applying the rules of chess and comparing downstream board configurations along different decision trees, like a computer.

There is further evidence to suggest that what an expert human chess player does do is recognize patterns, like a firefighter. In a famous experiment, chess players of varying levels of competence viewed chess boards projected on a screen for a few seconds each.10 They then had to reproduce the configuration of pieces they had seen. When the projected configurations were ones that actually occur in the game of chess, grandmas ters were able to correctly reproduce the positions of twenty to twenty-five pieces, very good players about fifteen pieces, and beginners five or six. But when the pictures flashed before them showed random configurations of pieces, not corresponding to patterns they would have actually come across in playing chess, then there was no difference in the players’ ability to reproduce the positions from memory; players of all levels