Tuesday, September 07, 2010

Reading :: Ghost in the Machine

The Ghost in the Machine
By Arthur Koestler


A couple of years ago, J.P. Rangaswami visited Austin for the summer and graciously invited me to breakfast, where we discussed a number of things, including his recent acquisition of Ribbit for BT. We also discussed the increasingly open way of doing business in organizations and industries, and I mentioned the networked organization of the telecomm company I had described in Network. "Oh yes," he said, "we used to call them holons."

At that point I realized that I would have to look up holons and see how they squared with networked organizations. I have been sidetracked by many things, both large (getting some projects finished) and small (the UT library failed to list a book as missing). But I eventually used the researcher's best friend - Wikipedia - to track down the origination of the term "holon": Arthur Koestler's 1967 classic The Ghost in the Machine.

If that name rings a bell, yes, the book's title was borrowed by the Police for their 1981 album. According to Wikipedia, Koestler did not seem impressed by this borrowing. Maybe Sting had missed the references that Koestler had made to pop music in this book: "Even from the aesthetic point of view we have managed to contaminate the luminiferous ether as we have contaminated our air, rivers, and seashores; you fiddle with the dials of your radio and from all over the world, instead of celestial harmonies, the ether disgorges its musical latrine slush" (p.321), he argued. (Recall that the publication year, 1967, was the Summer of Love.)

As you may have gathered from the above quote, Koestler was a first-class polemicist as well as a pessimist and a bit of a classist. At one point, for instance, he complains that "the more crowded [humans] are in slums, ghettoes and poverty-stricken areas, the faster they breed" (p.330). Hmm. He was also at one point an enthusiastic member of the Communist Party and a Soviet propagandist, a fact we'll come back to later.

The thing about polemicists, of course, is that they are quite entertaining when they are attacking those with whom you disagree. Koestler starts out the book with some terrific broadsides against behavioralist psychology, pointing out its oversimplifications with a polemicist's flair. "Pavlov counted the number of drops which his dogs salivated through their artificial fistulae, and distilled them into a philosophy of man," he summed scornfully (p.10). Soon he moved to the impoverished way that behavioralist psychology treated language, contrasting it unfavorably with Noah Chomsky's work. (Some of you may know Chomsky best for his photo op with Rage Against the Machine, but in 1967, only one member of that band had been born.) Later, he turns his guns on Darwinist evolution, arguing that random mutation was insufficient to explain evolutionary patterns and especially "lietmotivs" such as strong parallels between mammals in Europe and independently evolving marsupials in Australia.

But Koestler is no creationist - and he's no socioculturalist either. To understand his point of view, you have to understand the "holon," literally, the "whole-part": a hierarchically organized unit that can be a part of a more complex unit, and is in turn composed of smaller and simpler holons. These units are hierarchically organized, but they are also interlocking, not necessarily fitting together in one single way. They are Janus-like intermediaries between levels of organization.

Holons apply in biological evolution, of course, and Koestler sees them as the solution to the puzzle of parallel evolution between mammals and marsupials. But Koestler also sees social holons (pp.50-51); developmental holons; language holons; he states that habits are behavioral holons (p.76); he states that "habits and skills are functional holons" (p.207). In fact, holons become a Theory of Everything.

Holons have both integrative tendencies (tendencies to function in a larger whole) and self-assertive tendencies (tendencies to assert their own patterns as a unit) (p.56). The interplay between these is both creative and destructive. Applied to evolution - and as we eventually see, to everything else - this interplay translates to "drawing back to leap," or undoing and redoing (p.167; Here, Koestler applies the principle to evolution, science, and art).

I'll talk more on holons in a minute, but see if they sound familiar to you.

As we get closer to the end of the book, Koestler's themes become more urgent. He postulates that we human beings have an evolutionary flaw, and makes the case that evolutionary flaws occur. For instance, in Chapter 16, "The Three Brains," he describes three brains with significant flaws. Arthopods' brains, for instance, are built around their gullets - something that creates a built-in limit to brain capacity, since greater capacity means less room to swallow (p.268). Marsupials, on the other hand, are generally arboreal animals, but their brains dedicate too much capacity to the relatively useless sense of smell and they lack a corpus callosum to link the evolutionarily "new" areas of the right and left hemispheres of the brain (p.272). The third flawed brain is ours: Koestler argues that we have insufficient coordination between our archicortex and neocortex, and consequently a dichotomy in function (p.273). Our reptilian ("lizard") brain, limbic or paleo-mammalian ("lower mammal") brain, and "late mammal" brain are "relatively autonomous holons" with parallel functions (p.278). Koestler sees this result as due to rapid evolution of the brain, which has "overshot" our immediate needs (p.299). He goes on to detail the results, including our fondness for interspecial violence (p.307).

Let's take a moment to recall what the world was like in 1967. World War II was a recent memory. So was the 1948 Berlin Airlift. The Bay of Pigs Invasion was a few years earlier, in 1961. Proxy wars such as Vietnam were raging. Paul Erlich was about to publish The Population Bomb, arguing that overpopulation was inevitable and would bring mass starvation. Indeed, in Chapter 18, The Age of Climax, Koestler warns that we will reach "7 billion people in 2000" - but concurs with a Ford Foundation report that the "four horsemen" would "take over" long before then (p.315). On top of that, Koestler expected nuclear war. To bring it back to Sting, Koestler didn't dare hope that the Russians' love for their children would forestall genosuicide (p.322). To put it simply, the evolutionary flaw was too powerful.

At this point, Koestler only saw one hope: that we could artificially induce change in human behavior through genetic modification (p.327). The old propagandist prescribes the top-down solution of developing, then convincing people to take pharmaceuticals that would help them coordinate their limbic, reptilian, and late-mammal brains. Here, he sounds more like A.E. van Vogt than anything. Today, this prescription seems bizarre.

Okay. So have you thought about holons? If they sound strangely familiar to you, think back to Koestler's Soviet past. Holons sound an awful lot like the universally applicable, totalizing, and ultimately mad dream of Engelsian dialectics. (Koestler, frankly, did not advance the ball very far here.) Thanks for playing!

13 comments:

  1. The '70's spawned a number of these nested constructs, it seems: Pike's "tagmemes", Colomb & Williams' "D-Units"

    They seem to me to be an attempt to have it all: a delimited unit of analysis with infinite interpretive richness.

    But this very notion belies the nature of inquiry, it seems to me, which begins with an kind of resignation to paying attention to something at the expense of everything else. This move never precludes a different decision the next time, by the researcher or someone else, and so it needn't be seen as a problem to be remedied by an infinitely scalable and, therefore, ultimately intractable figuration of the object of analysis.

    ReplyDelete
  2. Seems a bit shoot the messenger to me - give that a very large amount of work on networked informations systems explicitly uses the term holon and cites Koestler, and you weren't familiar with it, and you wrote a book on it.

    Perhaps your breakfast host was being kind to you by suggesting in such a gentle and considerate manner that you were so blatantly under-researched. Certainly looking in IEEE or ACM would have found a lot of peer-reviewed articles.

    You seem to have dedicated column inches to why Koestler wasn't necessarily a very nice man, but not a lot to why many many computer scientists have found the term holon (and the derived terms holarchy and holarchic) so useful, not to mention the affine term holistic (will you write a demolishing piece on Smuts as an apologist perhaps?).

    Holons are participants in holarchies. Holarchic systems have unique features and help explain a as a model a number of complex problems that arise in the situations you are writing about.

    If you read had the book through you would have noticed that he presented a paper on the ideas at the Alpbach Symposium in 1968, and that he derived his ideas from discussion with some respected General Systems theorists with whom he was acquainted.

    ReplyDelete
  3. Thanks for your thoughts, Derek.

    The question of relating different levels of complexity is one that comes up in various fields and disciplines. In field studies of sociotechnical systems, for instance, people tend to use frameworks such as Latour's actor-networks, Deleuze's rhizomes, and Engestrom's activity systems/networks. Others skin the cat differently (see Bill's comment above), and from what you say, designers of networked information systems have found holons to be a useful way to skin the cat - for their work.

    But holons, like other constructs (actor-networks, rhizomes, activity systems, etc.), are not infinitely adaptable or universally applicable. Although holons and holarchies may be great for designing complex information systems, they have sharp limitations when applied to the sort of work I do (and wrote my books about). When Koestler applies holons as a Theory of Everything, including a theory of human behavior, he oversteps the bounds of what holons can explain well. As I said in the review, "he's no socioculturalist either."

    Perhaps it wasn't clear in the review, but this is the point I was trying to make by discussing Koestler's classism and his propagandism for the Soviet Union - not that he "wasn't necessarily a nice man" (I couldn't care less), but that the former suggests why he sees human beings as fitting seamlessly into the holarchic system he describes, while the latter suggests why his explanation of holons seems to track so closely with Engelsian dialectics. See for instance my reviews of Ilyenkov's books (here and here): Ilyenkov's discussion of dialectics sounds quite similar to Koestler's holons, which shouldn't be at all surprising if Koestler took his time in the Soviet Union seriously.

    In contrast, of the three frameworks I discussed earlier (actor-networks, rhizomes, activity systems), two developed as oppositional reactions to Marxist dialectics. The third is grounded in dialectics, but has progressively incorporated non-dialectical thought such as Bakhtinian dialogism. So they sharply contrast with Koestler's holons. The first two were developed a couple of decades after Koestler's book; the third predates Koestler in its roots, but was intensely developed into its third stage starting in the late 1980s. All three focus specifically on sociocultural and sociotechnical problems, and that focus has meant that they have developed to answer specifically sociocultural problems - in contrast to holons as Koestler describes them in this book.

    It's great that holons remain a vital construct in other fields, Derek. But for the problems that interest me and the readership of this blog, Koestler's version of the holon unfortunately doesn't provide much insight. That is likely why I have rarely if ever seen holons cited in the work of the actor-network theorists, activity theorists, Deleuzeans, distributed cognitionists, situated cognitionists, media theorists, human-computer interaction researchers, computer-supported cooperative work researchers, or ethnographers that I read.

    ReplyDelete
  4. This comment has been removed by the author.

    ReplyDelete
  5. This comment has been removed by the author.

    ReplyDelete
  6. This comment has been removed by the author.

    ReplyDelete
  7. This comment has been removed by the author.

    ReplyDelete
  8. Thank you for replying to my comments

    By the time Koestler was writing these books (featuring holarchy) he had totally repudiated communist ideas and was engaged in a bitter war of words with them. He developed the theoretical background to the idea of a holarchy within the context of English and European General Systems Theory - he was fascinated by biological systems and how they worked as analogical structures. Righlty or wrongly, a lot of thought was then current about biological structures as the best model for self-organising systems.

    The holon per se was conceived of in opposition to the hierarchy, because of the single apex-resolving nature of hierarchies, to which he was also opposed.

    Regarding the role of holons in management information theory, I supposed I was trying to suggest that JP Rangaswami (having existed in the European mangement context, prior to that in the Indian one) would have cut his teeth on the Soft Systems Methodology of Checkland et al, which began with a reflection in the article 'The Case for the Holon' in Systems Practice in 1988 - quote 'We could improve the clarity of systems thinking at a stroke by conceding the word system to everyday language and using holon whenever we refer to the abstract concept of a whole or build a model of a holon (models being always descriptions of holons which might or might not map onto some bit of real-world complexity).' Checkland is talking about how Bertalanffy's term 'telon' hadn't taken off, and how the word 'system' was too polysemous to be of use.

    ReplyDelete
  9. Soft Systems Methodology is (for good or bad) ubiquitous in the UK, and very influential abroad, and learning about holons and holonic systems is the very beginning of it. So when someone of that background is talking about holons, it is very much the case that they are talking about holons in the context of SSM.

    Holons get a look-in in a lot of other UK-derived IS and management thought (for instance the PRoH Process-oriented holon systems of Clegg and Boardman, the Holon Framework group in education, McHugh and the Holonic Enterprise) though there were a number of theorists in the States - Hock, the chap who thought of Caord.

    I suppose that's why I was baffled by your blog entry in the first place. You were responding to the origin of the term in Koestler, rather that the ubiquitous term in organisational literature. And your interlocutor would not have been referring to Koeslter, but to the use of Checkland, McHugh, Clegg, Boardman et al.

    ReplyDelete
  10. Right, now this starts to make sense to me. I've heard of SSM, but only in passing - like holons, SSM just hasn't been cited much in the literature I've read. But perhaps that's because I haven't read much in org lit, having stuck to the sociocultural literature and related strands for the most part. The places where I've seen it mentioned have tended to be in European pubs that attempt to relate it to HCI. Interesting to realize that it's so ubiquitous in a different field/discipline/context.

    So now I face the dismaying prospect of absorbing another set of literature - and I suppose I should also be less flippant with my blog reviews in the meantime :/ Thanks again for the feedback.

    PS, I see that Blogger's misleading error messages got to you too.

    ReplyDelete
  11. Yes, the controls are confusing, and my broken up posts were in the reverse order

    Re what you said - yes, once again invisible walls separating what should be adjunct disciplines!

    Holons and holarchies as models have been very important in the designing wide area information systems in manufacturing and distribution (eg the Holonic Manufacturing Sytems Consortium and their standards), in multi-agent and unmanned vehicle systems systems (eg MAS/MABS), e-Logistics, CAST etc.

    There are some rehabilitating papers by Franz Pichler 2000 on the holon which are worth reading, wherein he attempts a purely mathematical justification of the model independent of Koestler's intuitive work. Unfortunately not many people cite him (as yet).

    I suppose people want a model that looks "like that" and have found other people citing Koestler, but not many people actually read him (given the number of citations and the number of copies in engineering and science libraries). They assume that other people have done the due diligence - which in fact what you did in your blog-review - and that the model as conceived is correct.

    ReplyDelete
  12. You are perfectly correct in picking it up as a Theory of Everything - it was intended to be - but GITM was a popularising book, the work being done elsewhere (though not necessarily more rigorously).

    There is a better discussion in "Janus: A summing up" which Koestler wrote later on, and in which he tried to do a better job of making it into a TOE (there was a bit of criticism at the time). But noone cites it.


    Koestler was trying to make for a unified approach to thinking about natural and social phenomena on naturalistic grounds. But he was in a double reaction to communism and therefore evolutionary theory on the one hand, and the analytic approach on the other.

    To this end he brought to the Alpbach Symposium an amazing array of thinkers (Beyond Reductionism: New Perspectives In the Life Sciences; the Alpbach Symposium) including Waddington, Piaget and Inhelder, Hayek, von Bertalanffy, Frankl, Weiss, Maclean, Kety, Thorpe, Smythies, McNeil.

    Drack and Apfalter suggest that Koestler was reverting to Viennese general systems school of thought here. Their paper (Is Paul Weiss' and Ludwig von Bertalanffy's System Thinking still valid today?) is a good place to start on that thought.

    ReplyDelete
  13. Great - thanks for the references here! I suspect Pilcher's mathematics will be over my head, but I'll take a look, and I'll certainly pick up the Alpbach collection as well as "The Case for the Holon" and the other references. An acquaintance has just sent me a summary Checkland wrote on SSM, so I'll work that angle as well. I'm especially interested in how these will compare and contrast with the frameworks with which I'm more conversant.

    My rule of thumb is to blog reviews/notes of every book I read, so you might see writeups popping up eventually. (Not too soon, though, as I have a backlog.) I know I don't have to say this, but please do feel free to challenge me in the comments! This has turned into a rewarding conversation.

    ReplyDelete

Note: Only a member of this blog may post a comment.