Joi Ito's Web

Joi Ito's conversation with the living web.

Simon has many very interesting models that he develops for thinking about and describing things. In his words, he describes interesting "states" and interesting "processes".

Simon has many very interesting models that he develops for thinking about and describing things. In his words, he describes interesting "states" and interesting "processes".

Simon seems to conclude that everything might seem very complex, but that actually if you design the way you think about and the way you view everything, it actually can be described. He describes a system of a hierarchy of chunks which are "near decomposable" which means that intra-component (intra-chunk) linkages are generally stronger than inter-component linkages.(1) Or… High frequency being independent from low frequency dynamics. This allows evolution to occur by level in the hierarchy making it easier and more likely. It also allows the system to be described more simply because the design of the function or the "efficiency" of each component can be separated from the internal design of the component. Thus, the fitness function of a DNA strand is a based at the end of a combination of organs which is not so concerned about the detail of the internal workings of the organs.

I have a problem with this model. It is possible that the internal design of a component may not affect the next layer up in the hierarchy, but may affect another layer. This may be very non-linear and not as easily quantifiable as "efficiency". For example, the color of an organ or skin may not have direct impact on certain variables, but may have significant impact in certain situations from the fitness point of view. I think that there is considerable "chaos" integrated in hierarchical complex systems, particularly when they have to do with information or culture. Simon talks about chaos, but classifies it as a different sort of complexity than his hierarchical complexity. This may be true in many of the models he presents, but I think that this model by itself is limited. I have a "hunch" that chaotic systems can be "described". Maybe not in words, but in some method that allows us to on the one hand deal with physical things in this sort of reductionist mode and deal with chaotic information based things in some other method and reconcile this in some other kind of output. I have a feeling that there are some answers in art, anthropology or sociology which is more familiar with the non-linear, but natural.

The way that he simplifies the complex is by collapsing similarities and creating new views that lessen the number of chunks by adding levels or by organizing a level in a certain way. This is very interesting and is probably very useful, but a risk is that there are various methods of making our storage and manipulation of a reality more efficient, but each of the methods brings with it a bias towards a specific direction or solution. (2) Simon talking about the folly of the model of optimized markets which maximize the utility function. One can find adequate solutions, but there is no optimum solution. I think that one can apply this idea to methods of describing complexity.

I very much like his style, his approach and his methods, but I think there is a risk of underestimating the impact of weak ties, culture and non-linear interactions. On the other hand, it allows a level of rigor in thinking about complex things that is very enlightening. His idea of the linearity of the brain is very useful (although he doesn't address illogical decision making processes, dreaming, emotions, etc.). His view that the utility function is too simple and aspiration and satisficing can help explain some of the problems with the idea of optimizing the utility function are useful. On the other hand, I think that trust networks, communities, SWT, common knowledge, culture, etc. are much more complex than just computing aspiration and satisfaction.

His description of the role of organization vs. markets was very interesting and I think will lead very well into Chandler, so I will reserve myself a little until then.

  1. This is really about the strength of weak ties. Simon describes some examples using scientific methods which are very useful in thinking about SWT. The idea of gravity being more important than electrical attraction at a macro level is a very good example of SWT. Talking about frequency is also very useful.
  2. Hall in Beyond Culture talks about the risks associated with thinking that your thinking is the only way to think. I think that a literal interpretation of Simon my cause greater culture gaps. For example, I think that GAAP is a very good way to organize components, but it makes certain types of assets invisible, and is a very specific viewpoint even though "the market" thinks it is reality.


OK. Below I was trying to organize a series of "chunks" that Simon was presenting as a sequence of quotes… After reading further and finally the last chapter, I realize that was presenting a series of sub-concepts that he would use to support his final theory on hierarchy, complexity and near decomposability. I'll leave the quotes to digest later and write a few thoughts about his conclusion.

We can view the matter quite symmetrically. An artifact can be thought of as a meeting point-an "interface" in today's terms-between an "inner" environment, the substance and organization of the artifact itself, and an "outer" environment, the surroundings in which it operates. p. 6

We might look toward a science of the artificial that would depend on the relative simplicity of the interface as its primary source of abstraction and generality. p. 9

In a benign environment we would learn from the motor only what it had been called upon to do; in a taxing environment we would learn something about its internal structure-specifically about those aspects of the internal structure that were chiefly instrumental in limiting performance. P. 12

The computer is a member of an important family of artifacts called symbol systems, or more explicitly, physical symbol systems. Another important member of the family is the human mind and brain. p. 21

Intelligence as Computation intelligence is the work of symbol systems…a physical symbol system…has the necessary and sufficient means for general intelligent action…p. 23

Chapter 2 Economic Rationality: Adaptive Artifice Economics illustrates well how outer and inner environment interact and, in particular, how an intelligent system's adjustment to its outer environment (its substantive rationality) is limited by its ability, through knowledge and computation, to discover appropriate adaptive behavior (its procedural rationality). p. 25

Today several branches of applied science assist the firm to achieve procedural rationality. One of them is operations research (OR); another is artificial intelligence (AI). p. 27

OR being linear and AI being heuristic

A large body of evidence shows that human choices are not consistent and transitive, as they would be if the utility function existed. p. 29

To deal with these phenomena, psychology employs the concept of aspiration level…A theory of choice employing these mechanisms acknowledges the limits of human computation and fits our empirical observations of human decision making far better than the utility maximization theory. p. 30

Roughly eighty percent of the human economic activity in the American economy, usually regarded as almost the epitome of a "market" economy, takes place in the internal environments of business and other organizations and not in the external, between-organization environments of markets. To avoid misunderstanding, it would be appropriate to call such a society an organization-&-market economy; for in order to give account of it we have to pay as much attention to organization as to markets. pp. 31-32

The key question here, one much discussed in "the new institutional economics" (NIE), is: what determines the boundary between organizations and markets; when will one be used, and when the other, to organize economic activity? p. 40