A Personal Foreword
It is in the nature of conversation, indeed, it is (at a certain level) its purpose, that the unexpected happens and interrupts the normal and conventional flow. I had meant to write of conversation (in the sense of the great cybernetician Gordon Pask's seminal contribution to the field of Computer Aided Learning, "Conversation Theory") in this issue, as I had promised. But that old master of intervention, that old conversational deformer, that old conversation theorist Pask, himself, has interrupted my genteel progress.
Gordon Pask died at 9.00pm on Friday March 29th 1996, in the London Clinic. It was a relief, and it was a release, for he had been in pain and distress. But the actuality of his death, expected for a long time, was, nevertheless, a dreadful shock.
At the European Meeting for Cybernetics and Systems Research Conference in Vienna, April 1996, I was due to present a paper (NOAH: the Ark of Knowing in a Learning Environment) that was written by Gordon's long-time partner, Robin (Bobby) McKinnon-Wood and myself. It was an account of a new version ("NOAH") we had made of Pask's "ThoughtSticker"—the organisational "engine" of his computations for constructing learning environments—which took into account developments in programming techniques and styles, and a string of improvements and criticisms of ThoughtSticker and Conversation Theory which I had made over the years. Bobby and I worked on this project during Bobby's long struggle with death, and it was completed to the point that we could show the idea was viable just before he finally died last September. It was ironic, very pointed and poignant that, presenting a paper coauthored by one dead man, the other man who inspired it should also now suddenly be dead. The paper became a whole circle of tributes.
I decided to present the paper by talking about these two men, and their lifelong interaction in and through conversation. These were two giants amongst the mediocre. There is much to write about each of them as people, teachers, leaders, inventors, professionals, family men-but not here and not now. My story should not be taken as absolute historical truth (I simplify and dramatise in order to make points that I believe transcend pure chronology): it is not quotable as that sort of fact. But it is a true interpretation and it tells a story. I was lucky to know them, to be their student, to work and to play with them and be allowed to be an equal, to benefit from serious and equal discussion and mutual criticism of our ideas, and to have them as more than friends—one was like the elder brother I never had, the other like a father to replace mine, who died when I was very young.
This is a text version of that presentation in Vienna.
Rest in peace, Bobby and Gordon
Thomas Robin McKinnon-Wood: 23 April 1931 to 12 September 1995, the boffin.
Andrew Gordon Speedie Pask: 28 June 1928 to 29 March 1996, the prof.
____________
Robin McKinnon-Wood and Gordon Pask: a Lifelong Conversation
Just before 1950, two young men went up to Cambridge. Both had unusual backgrounds. One, the retiring one who was to become the boffin, had travelled the world with his family. He learned to speak in Switzerland, so that although his family was British, his first languages were French and German (and he never lost his French "r"). His teenage years were spent in Minneapolis where his father had been posted by the world bank. As a Cambridge physics scholar, he eventually came to inhabit Isaac Newton's old rooms in Trinity College. The other had had a lonely childhood. He was 20 years younger than his nearest sibling, a sickly child not expected to live. He spent much of his early years alone, ill in bed. His playmates were the companions of his mind, the fairies for whom throughout his life he left eight small piles of food on the rim of his plate. His education was as obviously strange and he had already studied mining at the local technical college before going up. Originally reading medicine, he was barred from the anatomy labs for impatiently cutting through a cadaver's limb with a fire axe, thus shattering the glass dissection table on which it lay. He was the dramatic one, destined to become the prof.
Somehow they met. The boffin was already, at this time, building computers (this before Alan Turing, father of the programmable computer, and builder of the world's first, committed suicide with a poisoned apple). A mutual taste was discovered for nightlife and the theatrical.
And they set to work on two extraordinary special purpose computers. The programs of these computers were truly astonishing, and it is amazing that neither their significance nor their authorship is currently widely recognised.
The first was called MusiColour. This was a machine that controlled lights so that they reacted to (live) music. But not like the trivial machines of today, which pulse with the beat and change colour with frequency. MusiColour (which erratically toured nightclubs and popular music venues) picked out patterns in the music, responded to these, and got bored. If the musicians did not provide enough variety, MusiColour would provide speculative change of its own accord, working itself up into a frenzy until the musicians changed how they were playing. This is truly interactive. And it is the first program that is an agent provocateur: its purpose is to interfere with the musicians. This computer, sadly long since dismantled, expects you to respond to it in interaction. It anticipates drama.
The second came to be known by the generic name "SAKI". SAKI stands for Self-Adaptive Keyboard Instructor, and covers a range of programs, of which the first was intended to aid the training of Hollerith punch card operators. Anyone who has learned to type using a computer program such as "Mavis Beacon Teaches Typing" will have used one of SAKI's descendents. The program measured the accuracy of performance of the punch card operator, in terms of accuracy (both of individual keystroke and of sequence) and speed (including rhythm) in carrying out the exercises it set, and it adjusted the exercise tasks to increase the operator's practise on (ie correction of) errors. Thus, the program adjusted to the performance of the punch card operator of its own accord, the first program to self-adapt, responding to you in the drama of interaction.
These two programs were quite revolutionary and remarkable, and are still serviceable today. Their central concern was the genuine interaction between the human user and the machine (in contrast to so much of today's so-called interactive material, which is merely push-button reactive: interaction as sales pitch rather than as meaningful).
We will leave the two undergraduates there, making their early interactive computers and founding their company "System Research", and move on about twenty years. During this time, they have remained in partnership, working together on occasional projects, but they also worked quite separately: the boffin, for instance, at Margaret Masterman's Language Research Laboratory in Cambridge, and the professor belatedly writing his PhD and actually becoming a "real" professor at Brunel University, just outside London, with actual students (including me).
The prof was now putting together what we would currently call a special purpose learning environment, CASTE (Course Assembly System and Tutorial Environment). This project, at the centre of a vast research enterprise funded by the UK's Social Science Research Council, derived initially from observations made during an analysis intended to assist the improvement of execution of a set of very complexly interlinked tasks.
And the boffin re-appeared, to collaborate. What they put together was a machine, a theory and a practice. The machine (product of the boffin) was an extraordinary special purpose (and very noisy) computer. Its contemporaries were, roughly, the early PDP mini-computers. This machine, however, was built on the cheap of ex-World War II relays. What it—and its successors—computed was relationships between topics of study (ie, things that we might want to learn) within a large heterarchy of such topics representing a field of study, and also the state of a learner's relationship with the topics (eg, had they been mastered, were they being studied). It looked like a strange and whimsical object from the Festival of Britain (which, in some respects, is exactly what it was).
The theory was Conversation Theory. Conversation Theory is a relativist, constructivist theory. It derives from teaching and learning: learning is the quintessentially constructivist activity. In even the lowest level of Computer Aided Learning, where the student can choose from many different alternative routes and thus compose, in effect, his own (view of the) subject, one important question is what has been learned. (That this is now recognised as an ever critical question is due, no doubt, in part to the development of CAL.) The matter of testing what is known is always hard, and it is harder still with a machine. One technique is called "Teach Back". In Teach Back the student tells the teacher, in his own words, what he believes (eg) a topic to be—how he understands it. (In CASTE this was done through a special device known as the Modelling Facility, as well as in interview.) Teach Back is interactive and conversational, and, when extended from simply testing what has been learned about some pre-defined topic, becomes an immensely powerful technique. In this view, conversation, its generalisation and abstraction, both as colloquially used and as a technical term in Conversation Theory, is a potential generator of novelty and of understandings (hence of worlds): and of communion. It is both medium and substance of drama. Conversation is an equivalent, an alternative statement (of the process) of design. It permits communication without coding and leaves meaning where I maintain it should be, with the communicators and not in their messages. Conversation is the basis. It is very powerful and it is very revolutionary. Rather like its author, the prof.
The practice was all that became associated with this theory. It included such matters as the consideration of learning styles and of their match and mismatch; how to define and represent what is learnable (and hence knowable); the structure required of topics so that they can be learned (they must be productive: any learnable topic must derive from at least two other topics (constituting its derivation or entailment), and the topics that form such a derivation must each be circularly derivable from the others; all topics in a subject matter must cohere to make a vast interconnected "cloud" of their mutual entailment; the learner must be able to begin anywhere (depending on his unique prior knowledge) and see how that allows the subject to unfold the topics to be learned; how to add new topics; etc); consequences in terms of performance of learners; the role and formalisation of explanation; and the connection between the topic as derived and the topic in its own right; etc. A vast body of elaboration of the theory into the practical (generally presented under the name "ThoughtSticker") in which the prof and the boffin were more than assisted (as they also were in the construction of both the machine and the theory) by their many co-workers at System Research and amongst the prof's students.
The body of what is collectively referred to as Conversation Theory is an astonishing achievement of imagination, nerve, rigour, ingenuity and adumbration (to use one of the prof's favourite words). It is, in many parts and particulars very hard to understand, because of a tendency to present it all, all the time, in its full complexity: which adds to the inherent difficulties of subtle and unfamiliar concepts (although they appear much less strange nowadays). It has not yet won the understanding it merits except amongst the cognescenti, in part because of these difficulties. But it is vastly imaginative, and has far more to offer in a far richer vein than so much of what is currently facilely trotted out. I hope that it will become better known, because it deserves to be and because it is so full of powerful and helpful insight and understanding of the drama of being human in communication. It is at least as amazing and astounding as were MusiColour and SAKI.
And now the two moved into more separate streams again. The prof took up residence in Montreal at Concordia University, and Old Dominion University at Norfolk, Virginia, USA. He becomes part of a team funded by the Dutch Government researching "Support, Survival and Culture", a professor at the University of Amsterdam. And he finds his last academic spiritual home at the Architectural Association School (AA) in London. The boffin hid himself away, as he was wont to do, until the prof brought him into the AA, where he introduced computing to the school (teaching me, too, so I could be his teaching colleague) and where they examined team decision systems—a natural extension of Conversation Theory, when you think of it. And he studied navigation: putting into practice the art of getting from A to B, which is just what a learner does in CASTE and ThoughtSticker. The boffin, always concerned with the practical, wanted to know how you got through a mesh, not what its special qualities were: all his life he was interested in "navigation".
During this phase, the prof was preoccupied with two matters. Both are rather more difficult than Conversation Theory, and I am not, myself, sure how resolved either is. I did, however, ask him about them a month before he died, and he was happy to agree what I describe here as being not completely wide of the mark.
The first grew directly out of Conversation Theory. It was always called Lp: L sub-p: the proto Language (or proto Logic) for Conversation Theory.
Lp is an attempt to construct a generative system for Conversation Theory. The need for it becomes clear when new topics are added to some collection (due to new discoveries or the addition of further authoring subject-matter experts, etc) which need to be fully linked into the richly connected mesh of entailments (a problem that arose in the work referred to in the program "EXTEND"). Adding a new topic is, of course, the way that a collection may be begun, although that was not always apparent. Lp is very complicated and recondite. I am not sure who (other than Pask) has understood its innermost subtleties. I do not claim to have. I suspect that its necessity is largely a product of the ideas in Conversation Theory being ahead of the conceptualisations of the instant of its invention, and that what it deals with may be done in another way, given the conceptual and technical developments (in computing, especially in the development of object oriented programming) of the last twenty years. But, equally, it might just be that I am not up to seeing the grand vision and the true significance of the undertaking—what is uniquely different about it.
Much the same might be argued for the Interaction of Actors Theory, which describes the space and the time within which conversation may happen. Later in his life, the prof became more concerned with what happened in the unending continuity that is interrupted so that a particular conversation may occur. His vocabulary includes such words as "immortality". The Interaction of Actors Theory is an attempt to describe the endless flux (the process of life) within which conversations can begin and end (as they all do) and how they can do so: the eternity of a conversational universe, in which each of us always acts and interacts and is always acting and interacting, bringing about the conversations of our experience and communication, and thus verifying that we are in communion, are in the drama of being alive. (For the prof, dramatic interaction was the basis of the experience of living). The availability of machines simply made this both easier to imagine and easier to do: we might become "immortal" through sharing with the (timeless) machine.
(The prof tried to bring all of these theories together in his final paper, "Heinz von Foerster's Self-Organisation, the Progenitor of Conversation and Interaction Theories", which will be published in Systems Research, volume 13 number 3—a Festschrift for his old friend and mentor, Heinz von Foerster.)
In all this remarkable work there are two themes: interaction, and drama. Interaction as in conversation. Drama as in conversation. MusiColour and SAKI are both involved in a conversation with the user. CASTE, ThoughtSticker, and EXTEND, set up conversational domains and Conversation Theory is just that. The other work contemporary with and complementary to Conversation Theory concerned engendering and measuring conversations. Lp and Interaction of Actors Theory define condiions in which conversations may start and stop. And so on. For this is a lifelong conversation, as is life itself.
Which is why I have told this story as I have.
Finale
Sometimes people ask me about the world that the boffin and the prof made. What is it like? they ask. I have tried giving various images, and the one I have come to prefer is this:
Imagine you go out at night in the depths of the countryside. There is a clear dark sky, no light spillage from civilization. It is a cold, damp night, and as your eyes become accustomed to the darkness you gradually see little points of light—the stars. These points of light shimmer (because of the dampness and the coldness in the air).
If you listen carefully, you can imagine them singing to eachother, forming into little choirs that sing together.
You, of course, are another of these stars, and you also sing. As you move, the other stars come into focus and form constellations: there are clusters that cohere for the moment, and that then dissolve as you move on. The songs continue, you hear the choirs of the constellations. The feeling is of overwhelming joy, beauty, wonder, oneness.
That is my picture of the world they made.
________
Afterword—Literature
MusiColour and SAKI are not very well covered in publications. But see:
"MusiColour" in Good IJ (ed) (1962): The Scientist Speculates, London, Heineman
"SAKI-Twenty Five Years of Adaptive Training into the Micro Processor Era" (1982) International Journal of Man Machine Studies, vol 17
Conversation Theory, CASTE, ThoughtSticker, EXTEND and Lp are covered in, for instance,
"Artificial Intelligence—a Preface and a Theory" in Negroponte N (1973) Machine Intelligence in Design, Cambridge, MIT Press
"CASTE: a System for Exhibiting Learning Strategies and Regulating Uncertainty) (with Scott BCE) (1973), International Journal of Man Machine Studies, vol 5
"A Theory of Conversations and Individuals (Exemplified by the Learning Process on CASTE" (with Scott BCE and Kallikourdis D) (1973), International Journal of Man Machine Studies, vol 5
"The Representation of Knowables" (with Scott BCE and Kallikourdis D) (1975), International Journal of Man Machine Studies, vol 7
Pask's impatience with much Computer Aided Learning (especially instruction) is expressed in
"Anti-Hodmanship: a Report on the State and Prospect of CAI" (1972), Programmed Learning and Educational Technology, vol 9 no 5
His delight in Art and the creation of intelligently interacting actors may be found in
"A Comment, a Case History and a Plan" in Reichardt J (ed) (1971): Cybernetics, Art and Ideas, London, Studio Vista
His own approach to Cybernetics may be found in the early, yet still apposite
An Approach to Cybernetics (1962) London, Hutchinson
The Interaction of Actors Theory awaits proper publication. There is a paper awaiting processing that was left on Pask's desk. Otherwise, consult the following
Interaction of Actors Theory (with de Zeeuw G, compiled and edited by Glanville R) (1991) OOC Program, University of Amsterdam
"Heinz von Foerster's Self-Organisation, the Progenitor of Conversation and Interaction Theories", Systems Research, vol 13 no 3
Appreciation and Evaluation
I edited a Festschrift for Gordon Pask (a special issue of the journal Systems Research vol 10 no 3, 1993). This contains appreciations, critical evaluations, personal notes, primer material, a list of publications. It is currently out of print, but I am optimistic it may be reprinted. A commemorative issue of the International Journal of Human Computer Interface (formerly the International Journal of Man Machine Studies, in which Pask so frequently published) will be edited by Bernard Scott and myself.
World Wide Web site
There is a commemorative web site: http://www.venus.co.uk/gordonpask/
You can send contributions to this site to: paskmemorial@venus.co.uk.
This piece is also published there.
Reader's Stories
Pask's daughter, Amanda Heitler, wrote a piece that I read for the family in Vienna. At the end she noted that everyone who had met Gordon had a Gordon story. She would like to collect these stories. If you have a Gordon story (or even several) you would like to share with the Pask family, please send it (or them) to: Amanda Heitler, 5 Cobham Road, Kingston-upon-Thames, Surrey KT1 3AE, UK.
Notes
1. I shall elaborate on this view of conversation and some of its consequences in a later paper.
Return to the content of this issue
Return to the Cybernetics and Human Knowing Homepage