Fists and irreducible complexity

Update: This post is a partial review of Malcolm Gladwell’s Blink: The Power of Thinking Without Thinking.

A key question in rhetoric and communication studies is how people are persuaded to act. Sometimes the act in question is overt in that it is the completion of some action; other times, the action could be implicit, in that it is the acceptance of some idea or line of reasoning as being true (or false). This latter group of actions are variously referred to as decisions or making up one’s mind. (I don’t consider these categories to be all that rigid. Consider them convenient shorthand for some temporary ideas—an argumentative place to hang your hat.) In Blink: The Power of Thinking Without Thinking (2005), Malcolm Gladwell argues that many of our decisions are made without our conscious input, that they are the result of unconscious processes that occur independently of our considered, conscious thought.

Many of Galdwell’s proofs for this idea come from the ideas of complexity theory and have interesting applications to rhetoric. In the first case, Gladwell references Paul Ekman and W. V. Friesen’s Facial Action Coding System (FACS) (here is a link to a brief explanation of FACS, and a FACS chart is pictured below). Ekman and Friesen documented over 10,000 configurations of the facial muscles, of which they found about 3,000 that were meaningful. This result comes from the layering of actions in the facial muscles, actions which grow exponentially as more muscles are worked together (or in sequence). The result seems very much like a strange-attractor type problem, where out of many countless meaningless results—what Gladwell calls ‘the kind of nonsense faces that children make’ (201)—a few stable meaningful results arise.


In Gladwell’s discussion of ‘thin-slicing’, his name for the ability to find emergent (my word, not his) properties of a system in very small samples of that system—say, ten seconds of a couple’s conversation is sufficient to make a highly-probably determination of that couple’s future, or a similarly small sample of Morse code is enough for a trained listener to be able to identify the operator transmitting that code. According to Galdwell, this code pattern from the second example, called a fist, ‘reveals itself in even the smallest sample of Morse code’ and ‘doesn’t change or disappear for stretches or show up only in certain words or phrases’ (29). This fact would seem to indicate that the fist is not irreducibly complex, that is, that the message is not the shortest possible way of describing the fist, for the fist shows up even in very small samples of the message.

In complexity theory, the irreducibly complex is equivalent to the random. Take the example of a random string of numbers. This string is the prototype of an irreducibly complex message because it cannot be expressed in a reduced form. The shortest method of reproducing a string of random numbers is the string itself. Language theorists like Jacques Derrida seem to argue that all symbol messages are irreducibly complex in this way, that they cannot be expressed in any shorter form than what they are, for to shorten or summarize them would be to make a different message by leaving out key information.

The fist example seems to indicate, however, that some significant portion of symbol messages, those parts that are roughly equivalent to style, are able to be reduced and maintain their identity. I’m not quite sure what the implication of this result is, but I find it interesting, especially in the context of analog and digital communication. Though Morse code is essentially a digital medium, the fist only appears as an analog aftereffect of the digital message. Similarly, Nicholson Baker’s advocation for the preservation of library card catalogs is an example of a digital message that is willing to discard analog aftereffects that are deemed unimportant.

Now, it is obvious that the digital portion of a message is also not irreducibly complex. New methods of compression might make it possible to transit the same message in a shorter form. The counterpart in communication theory, I suppose (and someone feel free to correct me if I’m getting all of these theories in a muddle), is that the iterable nature of symbol systems allows for shorthand communication of messages.

If messages are made up of both digital and analog components, neither of which are, by themselves, irreducibly complex, what then, in the Derridian sense, is the irreducible part of the message? I wonder if it is the interplay between these two elements, the connection between the analog and the digital, that is random and irreducible.

One of Gladwell’s arguments in Blink is that it is in our interest to discover where our unconscious decisions arise from as a means of determining whether or not they are to be trusted. On that note, a final thought: the analog portion of the message is much more difficult to counterfeit than the digital, though such imitation is not impossible. When all information is digitized, it is able to be copied—falsified—endlessly, much more easily than analog messages. This is because digital messages lack the global, emergent features of the reducibly complex, like the Morse code operator’s fist. As digital information is still carried via analog devices (analog telephone lines, for instance) it is possible that this portion of the signal can be analyzed for identifying features. One solution to our current concerns with digital security might be finding a way to reconnect the analog and the digital, making individual messages more difficult (perhaps impossible?) to counterfeit.

Confections (s)ugare

Update: This post is a partial review of Edward Tufte’s Visual Explanations: Images and Quantities, Evidence and Narrative.

Edward R. Tufte describes a confection as a group of visual elements assembled to describe or enhance a written argument in his 1997 bookVisual Explanations. A confection, he argues, is distinct from both collages, which are intended to convey messages not associated with written arguments, and diagrams, which convey messages but whose elements lack the disparate nature of a confection. A straightforward photograph is, according to Tufte, not a confection; but two photographic images superimposed upon one another to create a fantastical mélange that cannot be photographed would be a confection. In this sense, a confection is an arrangement of disparate elements so as to make an argument. (One example is the title page of Hobbes’s Leviathan, pictured below.) It is this arrangement, the confection’s fantastical placement—either in space or in time—of otherwise unrelated visual elements, that makes them theoretically interesting.


Let me explain why. Accepting the above definition, I ask: what does it mean to say that a confection makes—either implicitly or explicitly—an argument? That is, is the argument made by the confection independent of its fantastical arrangement, or is the argument dependent on this arrangement?

I’m thinking two things: 1) since the characteristic of confections that makes them so—the arrangement of disparate elements to make an argument—is true of all graphical arguments, be they diagrams, graphs, straight photographs, and drawings, does it not also follow that all graphical output is confectionary in some degree? And doesn’t this realization of the fantastical in all graphics lead to certain conclusions about their trustworthiness and veracity? More on this after: 2) Isn’t the very construction of a causal argument confectionary in its selection and arrangement of elements that are not adjacent in space and time?

This, I think is a key point, for what makes a graphic a confection makes it an argument. Realizing this fact—that what argues is confectionary—provides a graphical explanation that undermines claims of objectivity or absolutism on the parts of even the most clever conclusions.

Decentralized Systems

Update: This post is a partial review of John Holland’s Hidden Order: How Adaptation Builds Complexity and Mitchel Resnick’s Turtles, Termites, and Traffic Jams: Explorations in Massively Parallel Microworlds.

Both John Holland (Hidden Order, 1995) and Mitchel Resnick (Turtles, Termites, and Traffic Jams, 1994) argue that it is difficult to discern the behavior of a system from the behavior of its parts. Through the use of computer modeling—cellular automata in Holland’s case, the StarLogo programming environment in Resnick’s—both attempt to begin to understand the nature of these complex systems.

Holland defines complex systems as a product of “the interactions” between their relatively simple parts (3). The result of these interactions, which are often relatively simple themselves, is that the “aggregate behavior of a diverse array of agents”, or the “parts” of the system, “is much more than the sum of the individual actions” of those parts (31). That is, the ordered behavior comes as a result of the particular way in which objects interact, rather than from any kind of centralized oversight, which is presumed to be the source of most ordered behavior. Holland gives the example of a city as a complex system that “retain[s]” its “coherence despite continual disruptions and a lack of central planning” (1). Now most cities obviously have central planning architectures in the form of governments, but those centralized authorities often find it their job to combat or enforce city behavior that does not originate directly (at least in appearance) from their decisions. Where do city-level features like traffic jams, ethnically- or economically-segregated neighborhoods, and homelessness come from? Rarely can they be directly attributed to central planning. Rather, the interactions between residents—which are often dictated by central planning organizations—and other structures in the environment help to form and maintain the city and its “personality.” This aggregate behavior results in “an emergent identity” that, though continuously changing, is remarkably stable (3).

Similarly, Resnick explicitly focuses on these “decentralized interactions” and the systems that result from them (13). He provides five “Guiding Heuristics for Decentralized Thinking”: 1) “Positive Feedback Isn’t Always Negative”, that is some kinds of positive feedback, in the economy for instance, can lead to increases, rather than decreases, in order; 2) “Randomness Can Help Create Order”; 3) “A Flock Isn’t a Big Bird”—systems do not behave like a larger version of their components; 4) “A Traffic Jam Isn’t Just a Collection of Cars”, or decentralized systems are more than the sum of their parts; and finally 5) “The Hills Are Alive”—the environment and context of a decentralized system are key components of its behavior (134).

Resnick sees the decentered view of the world as necessary to changing deeply-entrenched centralized ways of thinking. According to him, this decentered view became apparent in the work of Freud and his description of the unconscious, and other decentered metaphors have been slowly gaining ground in other fields since then.

This view of the world as being primarily the product of decentered behavior has interesting applications for rhetoric, for persuasive situations are as decentered and interaction-dependent as the systems studied by Resnick and Holland. Resnick notes that as decentered thinking—in the form of chaos and complexity theory—has gained traction, “scientists have shifted metaphors, viewing things less as clocklike mechanisms and more as complex ecosystems” (13). Rhetoricians since the sophists, however, have known the complex, dependent nature of communication, where “decentralized interactions” like the complex interactions of rhetorical appeals and “feedback loops” of self- and community-reinforcement (13) are well known.

These connections imply that rhetoric is well-suited for application of decentralized thinking. Certainly there is a tendency even in rhetoric to over-emphasize centralized behaviors to the detriment of decentralized ones—see the work of Peter Ramus. As rhetoric continues to move closer to a sophistic understanding of the power of persuasion, the models of Resnick and Holland have the possibility of shedding light on rhetorical situations, providing a language to explain behaviors that, though recognized, might have been previously unexplainable.