Ted Nelson on Gordon Pask
from "Computer Lib/Dream Machines" 1974
©1974 Theodore H.Nelson
In a field full of brilliant eccentrics Pask has no difficulty standing
out.
Pask is one of the Artificial Intelligencers who is working on teaching by computer, about which more will be said; but the original core of his interest is perhaps the process of conceptualisation and abstraction.
Pask has done a good deal on the mathematics of self-contemplating systems, that is, symbolic representations of what it means for a creature (or entity omega) to look at things, see that they are alike, and divine abstract conceptions of them. A crowning moment is when Omega beholds itself and recognises the continuity and selfhood. (Pask says several others - scholars from Argentina, Russia and elsewhere - have hit on the same formulation.)
Models and abstraction, then, are what we may call the first half of Pask's work.
Gordon Pask will be continued on p.DM47
I will now try to describe Pask's work as he has explained it to me. Perhaps this will be of some help to those who may have been dumbfounded by contact with this fabulous man.
Gordon Pask's concern is abstraction and how concepts are formed, whether in a creature of nature or a robot or a computer program. Abstraction is of interest primordially (as life evolved thinking capacity), psychogenetically (as the mind acquires new facilities, described most peculiarly by Piaget), and epistemologically (how do we know? Like how do we know, man?), and methodologically (how can we most effectively find out more?).
His interest, then, is in teaching by allowing students to discover exact relations in a specific subject matter by the very process of abstraction that is of so much interest.
What he does, then, is prepare given fields of learning so that they can be studied by students using abstractive methods, without guidance.
This preparation basically has two steps. First he sets up the whole field. This is done in collaboration with a "subject matter expert", who names the important topics in the field and states what interconnections they have. The result is a complex graph structure which Pask calls a conversational domain. It comes out to huge diagrams of labels and lines between them.
Then Pask processes this structure to make a more usable map of the field that he calls an entailment structure. The processing basically involves removing "cycles" in the graph, thus making the structure hierarchical in a slightly artificial way justified by what the subject-matter-expert has said is the structure of the field.
(This processing is carried out by a program called EXTEND)
The resulting Entailment Structure is then presented to the student as a great map os the field which he may explore.
Pask intends that the student's explorations will consist of testing analogies, or what PAsk calls morphisms, to find the exact exact structures of knowledge he is supposed to be acquiring. This knowledge will be in the form of isomorphisms, or exact analogies, i.e. laws.
Pask's overall system, examples of which he has running in his laboratory in England, he calls CASTE (Course Assemnbly System and Testing Environment). A further development which is to be put on a PDP-11/45 computer at the Brooklyn Children's Museum, is called THOUGHT-STICKER. This program is intended to allow the demonstration and testing of analogies directly, by children.
Essentially Pask is reducing a field to an extremely formal structure of relations which may then be studied by the student, at the student's initiative.
(What I don't quite understand is how the analogies are to be explored and tested.)
Anyway, a principal point is that the student is in control and may use his initiative dynamically; the subject is not artificially processed into a presentational sequence. Moreover, the arbitrary interconnections of the subject, which are no respecters of the printed page, are recognised as the fundamental structures the student must deal with and understand. On all these points Pask and I are in total agreement.
Indeed, his explorable sytems - (I don't know if they will be what I elsewhere call hypergrams or responding resources) - will be fascinating, fun and terrifically educational. Because he is.
Now it turns out that this exactly complements the notion of hypertext as I have been promulgating it lo these many years.
Hypertext is non-sequential text. If we write a hypertext on something, it will be most appropriate if we give it the interconective structure of the field. In other words, the interconnective structures chosen for the textual parts are likely to have the same connective structure (in general) as Pask's Entailment Structure.
For another kind of hypertext, the anthological hypertext built up of lots of other writings, it is also reasonable to expect the connective structures to cluster to the same general form as Pask's entailment structure.
In other words, the very same field of knowledge Pask is out to represent as an explorable, formalised whole, I am out to represent as an explorable informalised whole, with anecdotes, jokes, cartoons, "enrichment materials", and anything else people might dig.
In still other words, let's have both and call it a party.