December 5, 2017

Herbert Simon and Gilbert Simondon, Part I

One of the most interesting 'new' (technically speaking, as a new translation was published in 2017) ideas about technology I have come across is the 'mechanological' theory of French philosopher Gilbert Simondon. Simondon has an interesting background, comprising both the usual humanistic education you would expect of a French Big Thinker – a graduate degree in pre-Socratic philosophy – and more unusual educational interests – certificates in mineralogy and psychophysiology as well as broader intellectual fascination with zoology, mathematics, and the arts. Simondon's ideas are not easily characterized. Why? Continental philosophy, for one, uses scientific and technical terminology in a much different manner than established science and technology (sometimes incorrectly). But the main reason why it is difficult to characterize Simondon is perhaps best explained by the fact that he is principally known by his philosophy of technology, even if his ideas are not necessarily limited to the technological. Simondon is not necessarily a philosopher of technology, but he sees philosophy from the perspective of technological problems. In this respect, Simondon's similarity to the American 'scientist of the artificial' Herbert Simon is obvious and goes far beyond merely the recurrence of the string 'Simon' in both of their last names. Both have similar, if nonetheless divergent, views of the world that are reflected through the question of the natural and the artificial and the scientific debates that occurred before, during, and after World War II about cybernetics, feedback, and information. Simondon is a kind of Gaullic version of Simon, and Simon the Yankee version of Simondon. This does not mean, however, that they converged on the same insights or that their ideas can be fused together as a kind of intellectual Frankenstein creature. For Simondon to be a Gaullic variant of Simon and vice versa implies the exact opposite!

The similarities and differences between Simon and Simondon are, for the most part, not an object of significant discussion among devotees of either. This has little to do with the much-lamented gap between the 'two cultures', something that Simon and Simondon both – in their own ways – demonstrated to be fallacious. Simon was well-read in philosophy and was a student of the philosopher Rudolf Carnap. Simondon, for his part, engaged substantially with the ideas of foundational figures in cybernetics such as Norbert Wiener. Both saw themselves as broadly responding to the moral challenges of modernity presented by science and technology. Simon sought to find a middle path between tyrannical control and unrestricted choice, a moral aim that guided his obsession with bounded decision-making and the role of hierarchy. Simondon fiercely critiqued the sterile conception of social control advocated by Wiener and others and rejected both humanist disdain for technics as well as the idolization of technics often espoused by those with disdain for humanists. Finally, both tended to straddle the line between various disciplines and topics, maintaining consistency only really in worldview and ideas rather than necessarily subject matter. The delta between the two Simons begins from the question, really, of why anyone should care about the philosophical dimensions of the artificial to begin with – a question with very different framing and answers in the US and France. Neither Simon or Simondon are purely creatures of their intellectual environment, but the variant environments of their intellectual formation also cannot be overlooked.

The 'discovery of the artificial' in the Anglo-American scientific and engineering communities is generally linked to the mechanization of mentality. This story did not begin with the 20th century; it began centuries ago with the quest to ban the assignment of agency to natural phenomena. We do not say that a rock falls because it seeks the center of the earth. Nor do we necessarily say that plants pursue sunlight. Cybernetics and the assorted set of disciplines that it comprised and influenced found, via an convenient trick, a way of using broadly teleological explanations without buying into the idea of a final, motivating cause. This was the product of ambiguous but nonetheless critical relationships with the design of analog and digital control and computation systems, which became either sources of analogy or one-to-one correspondence for the behavior of humans and other animals. The tradition of 'mechanology' in France and later Canada had a much different origin and focus. Beginning with Jacques Lafitte, mechanology often took the form of descriptive typologies of machines. Instead of using machines as a way of explaining human behavior – though they were not averse to sometimes finding places of commonality and analogy – mechanologists sought a philosophical description with the role of machines in society. It was often critical of the cyberneticists but respectfully so. The reaction of the French intellectuals to cybernetics was not solely to critique it but to further systematize it in ways that had an enormous impact on French philosophy in the latter half of the 20th century.

This is suitable background for a comparison of Simon and Simondon by way of their respective magnum opuses, The sciences of the artificial (SOTA) and On the mode of existence of technical objects (MEOT). I have accumulated some notes on both, which I will add here as I make them legible. One thing I hope to convince readers of is that Simon and Simondon offer complementary but distinct perspectives on the particular problem of how to think about the world through the lens of technology. These perspectives are rooted in a common viewpoint that technical things are distinct from natural things in ways that complicate the presumption that biological ~ natural. Simon, for example, notes that farming and agriculture involves biological entities but owes its origin to deliberate human artificing and manipulation. Likewise, Simondon uses the example of a plant in a greenhouse. This too is biological but it is also dependent on human intervention for its sustenance and continued existence. This example, however, illustrates their subtle difference. For Simon, artificial things are primarily to be understood as a function of their adaptations, procedural mechanisms, and other features of their interfaces and behaviors in an environment. Environments will mold artifacts to adapt to the requirements of the environment with respect to a goal. For Simondon, technical objects are characterized by 'individualization,' processes by which the functions of a technical artifact recursively converge while making use of the surrounding physical environment as a component of its functionality. This is one way in which, as Yuk Hui observes, the two differ fundamentally. As the conclusion of this post, I will explicate this difference tentatively.

As Simon Mills notes in summarizing Simondon, "[t]he abstract form of a technical object describes a technical object that has ‘an intrinsic perfection of its own that needs to be constituted as a closed system in order to function.' Simondon uses the example of the move from water-cooling to air-cooling systems for combustion engines to demonstrate this shift. Although a water-cooled engine consists of two systems which are perfectly suited to carrying out their specific functions, when linked together a degree of disparity emerges between them. Simondon refers to the joining of these conflicting technical individuals as creating a ‘series of problems to be resolved.' The concretization process is one where such problems are resolved. Thus, the development of air-cooled engines, by the addition of gills to the cylinder, is seen as a measure of concretization because the engine's cooling function is no longer provided by a separate closed water-cooling system, which requires its own conditions for operation that conflict with the operation of the engine, but as part of the normal operation of the single technical system." Therefore, "as part of the organization of functional sub-systems into a total functioning, an associated milieu is invented and maintained. It is important that the specific meaning Simondon gives to the term invention is grasped here. Invention does not refer to the traditional hylemorphic notion in which a person has an idea and then builds something that corresponds to that idea; rather it is the ‘birth’ of a new environment or ‘regime of functioning’ brought about by the operation of recurrent causality involving the actual operation of the technical individual itself." For [Simondon]( a designed artifact takes on the features of a natural entity as it grows more independent of the original conditions of its design, in short. How does Simon approach the same problem? A review of Simon's ideas is sufficient to basically establish the divergence in worldview.

For Simon, the progress of a designed artifact have to do very much with the "thin interface between the natural laws within it and the natural laws without." The natural laws without are only partially under the control of the designer, e.g, the designer of an airplane cannot directly control the laws of aerodynamics. The challenge is to evaluate the alternative airplane designs, with the knowledge that doing so involves optimizing with respect to both material constraints/trade-offs and organizational constraints/trade-offs. Simon is often depicted as a rote, rationalistic positivist that ignored ambiguity about goals. This is not really the case. Simon recognized that the process of design actually involved evolution of the goals of design. Simon also more narrowly was particularly taken by the idea that decision-making in the present changes one's future environment of decision. In fact, as Hunter Heyck reflects, this is critical to Simon's articulation of means-end decision-making. "Choices are selections of certain actions or sets of actions from all the vast set of possible behaviors, selections which are intended by the chooser to bring about one (or a set) of possible future states. Drawing on Wittgenstein (and using chess as a recurrent example), Simon describes a decision as being like a move in a game, where a specific action is taken (one out of a vast set of possible moves is chosen) in order to bring about a new configuration of the gameboard that is the world." Heyck notes further that "Simon was impressed by what he termed "the time-binding” nature of decision-the fact that decisions of the present moment can determine not only one’s present actions, but also one’s future choices. Because individuals (and societies) can construct their future environments of decision through their present choices, decisions are integrated with each other not only in a hierarchy of values, but in a sequence of decisions – in courses of action or strategies, to use the terms that became popular with the postwar rise of game theory."

It is tempting to draw an equivalence between Simon and Simondon based on their mutual interest in the particular purposeful structures of artifices. Simon predicts that, within some bounds, artificing will reflect the molding of the artifce to a hierarchical scheme that best organizes goals and subgoals to guide the course of present and future action, something that Simon refers to as the erection of an "environment of choice." Likewise, Simondon discusses how technical individuals must inherently utilize an milieu – or physical environment – associated with them, and that a particular efficient and well-organized arrangement of the technical object is definable as a particular mechanistic composition of objectives and functions. But, in a purely technical sense, Simon and Simondon would likely disagree about what makes complex artifacts work. Given his theory of 'near-decomponsability', Simon would likely find Simondon's insistence on technical evolution as the erosion of modular and hierarchical functionality to be ridiculous and redolent of the fashionable 'catastrophe theories' that competed with complexity theory. Simondon, who – unlike Simon – took his technical inspiration from engineering topics such as crystallization processes instead of computer architecture – would likely retort that Simon took the existence of certain structural forms for granted without explaining the particular operations that produced such entities and continuously operates on them. Simon would be somewhat sympathetic to Simondon's point that the 'perfect automaton' is mythological, perhaps nodding along as Simondon discussed the absurdity of the idea of an entity with little margin of indeterminate operation in complex environments. However, it is likely that both would disagree about the nature of this indeterminate operation, as Simon declares that "ontogeny recapitulates phylogeny," a meaning of ontogenesis that Simondon is certain to disagree with. As I read more of the commentary on Simondon and Simondon's translated works, I may reconsider some of these characterizations and comparisons. But for now, I make this the first of my series on the Two Simons.

Tags: concept