Getting acquainted with cybernetics

About 10 minutes from me is a local bookstore that I like to make a point of visiting when I feel like my office has too much space. It was on one such occasion in January I happened to find a copy of Norbert Wiener's The Human Use of Human Beings on the shelf.

At the time, I had some passing familiarity with feedback systems and information theory. I've worked with Kubernetes, a system for scheduling networked computing workloads that takes its name from the Greek term for governor. I've also implemented some closed-loop controllers in code, so the idea of what looked at first glance like a study of these systems in a broader societal context was at least worth $15 and some of my time. It was also worth the experience of having someone ring up a purchase on a homegrown POS system running on an Ubuntu desktop terminal, which is reason enough for me to keep going to said store. As I do, I bought The Human Use of Human Beings along with a few others and headed on my way.

Before getting into the content of the book itself, I'll note first that the Da Capo Press copy I purchased is not particularly great. It's a paperback reprint of the 1954 edition, which is generally fine, but the cover is blurry and the paper is not of great quality. I'm not picky about that sort of thing when picking books off a shelf, but in this case it stood out to me.

Now, on to the book. The Human Use of Human Beings is written as an introduction to and commentary on cybernetics as a field. Borrowing concepts from statistical mechanics and information theory, it is here where Wiener provides his central thesis of cybernetics: society and life can only be understood through the messages passed between systems^[1]. These messages contain information, a quantity describing a message's bearing on the present. Information is subject to the same tendency towards entropy as physical systems^[2], and thus a system's survival depends on its capacity to process information^[3]. Systems reduce local entropy and increase their level of organization through feedback, defined as control based on observed system performance. Those systems that can learn within their environments necessarily possess some form of memory and patterned behavior, where such systems maintain their organization by altering the latter in response to the former through feedback. In cybernetics, this same principle applies to both living organisms and machines; the difference between the two is largely a matter of degree rather than kind^[4].

Cybernetics, however, is about more than just the mere mathematical study of information and feedback. It is a response to a belief in a universe where life itself is "an island here and now on a dying world"^[5]. The only force that matters is the increasing and totalizing entropy that will eventually obliterate all possibility of being. Self-organization stands in opposition to decay, and feedback is the central mechanism for maintaining this organization in an ever more disorganized reality. While all life will end and the universe with it, cybernetics makes the claim that prolonging self-organization for as long as possible is good, with Wiener going so far as to posit that this local reduction of entropy is what gives rise to a sense of purpose in human society^[6]. It is not clear why it is good to prolong self-organization in such an environment, though any alternative would make it hard to justify continued existence and presumably the writing of the book in the first place.

Wiener lays out his case for the universality of feedback systems in nature by applying cybernetic principles to a variety of domains, such as natural language, neuroscience, law, and the broader climate of American scientific research in the 1950s. The theme tying these together is as follows. All living systems exist because of their capacity to process information^[7], a task for which humans are uniquely suited. Despite this outsize capacity in human individuals, human society regularly fails to process information effectively: Renaissance scholars destroyed Latin by refusing to adapt it to then-current needs^[8], fascists squander human potential by classifying all people into insect-like roles^[9], valuable art becomes worthless because of derivative works^[10], and sprawling laboratories keep their researchers isolated from each other to the detriment of progress^[11]. The ultimate consequence of this in an era of machines of ever-increasing information processing capacity is an apocalyptic arms race that ends in humanity losing its control over the systems it has created and subsequently destroying itself^[12].

It's an interesting book, and Wiener makes a reasonably strong argument for the thesis that information forms the foundation of not just societies, but life itself. What is left unsaid in The Human Use of Human Beings is why, or whether, increased organization is a necessary response to this state of affairs. Tainter argues differently in The Collapse of Complex Societies, claiming that human societies exist to solve problems (i.e., process information), and those solutions require energy. The processing of energy and information yields declining returns, requiring new solutions to new problems. This recursive growth in complexity repeats itself until it exceeds the processing capacity of the society, at which point it reverts to a level of lower complexity appropriate for its environment; this is what marks societal collapse^[13].

Cybernetics, Wiener argues, does not need an answer for these concerns. Information is a natural phenomenon which we will use whether consciously or otherwise, and history shows that in our ignorance we have done a poor job of this so far. Understanding information, then, at least grants us the chance to use it effectively, and it is this effective use of information that the field promises. Whether this represents a true escape from our current situation or merely another rung on Tainter's ladder of complexity is left as an exercise for the reader.

As a coda: One may be tempted to believe cybernetics is or was a niche 20th century fad that came and went, like so many other postwar intellectual trends. This is not so. At its core it is an intellectual project to reframe other disciplines in terms of feedback, and in this respect it has largely succeeded. In the modern era, the difference between insulin response and TCP congestion control is merely a matter of the parameters you choose. This is the world of the data center, Project Cybersyn, and a very good album from 1979. It is a world that cybernetics demands we modify to suit its needs, and we have modified ourselves with it.