Iterative Dialectic Engine for Automated Learning
Dialectic is a highly contentious subject, because it means different things to different people, and some of the people who derive meaning from it hold quite strange views. The following lines from Apocalypse Now are a clear expression of the more extreme mindset. Kurtz (Marlon Brando) is reading aloud Eliot's The Hollow Men, when Dennis Hopper's anonymous character chooses to enlighten Willard (Martin Sheen) on the meaning of the recitation:
"Do you know what the man's saying? Do you? This is dialectics. It's very simple dialectics, one through nine, no maybes, no supposes, no fractions, you can't travel in space, you can't go out into space, you know, without like, you know, with fractions, what are you going to land on, one quarter, three eighths, what are you going to do when you go from here to Venus or something? That's dialectic, physics, OK? Dialectic logic is, there's only love or hate, you either love somebody or you hate them..."
- at which point Hopper has to break off in order to evade a poetry book lobbed in his direction.
Like Kurtz, Karl Popper's reaction to dialectic was vigorous and expressed through the written word. Indeed, Popper's devastating attack on Hegelian dialectic remains unsurpassed as a refutation of woolly thinking. His main criticism is aimed at the acknowledged contradiction between the first two terms of Hegel's thesis/antithesis/synthesis triad:
...if a theory contains a contradiction, then it entails everything, and therefore, indeed, nothing. A theory which adds to every information which it asserts also the negation of this information can give us no information at all. A theory which involves a contradiction is therefore entirely useless as a theory. (1)
Popper concludes:
The whole development of dialectic should be a warning against the dangers inherent in philosophical system-building. It should remind us that philosophy must not be made a basis for any sort of scientific system and that philosophers should be much more modest in their claims. (1)
Why, then, does the 'D' in 'IDEAL' stand for 'Dialectic'? It is tempting to make reference back to the ancient Greek usage of the word, but even at that time 'dialectic' had several different definitions, and it would be foolhardy to imply that a modern reinterpretation has somehow unravelled the contradictions. For instance, a definition of dialectic as "the reconciliation of opposing opinions or facts, by means of logical disputation or synthesis" might satisfy the main ancient and modern exponents, Socrates and Hegel respectively, but would it satisfy Plato, Aristotle, Marx, Popper, Hopper, or any other commentator? We cannot say. The most we can say - and, in fact, the most we need to say - is that, since ancient times, it has been recognised that there is an aspect of rational enquiry that has persistently evaded description using formal methods of deduction and proof. Some call it 'dialectic'; Popper chooses to calls it 'trial and error' or 'scientific method'. It is instructive to quote Popper at length on this topic:
If the method of trial and error is developed more and more consciously, then it begins to take on the characteristic features of 'scientific method'. This 'method' can briefly be described as follows. Faced with a certain problem, the scientist offers, tentatively, some sort of solution - a theory. This theory science accepts only provisionally, if at all; and it is most characteristic of the scientific method that scientists will spare no pains to criticize and test the theory in question. Criticizing and testing go hand in hand; the theory is criticized from very many different sides in order to bring out those points which may be vulnerable. And the testing of the theory proceeds by exposing these vulnerable points to as severe an examination as possible. This, of course, is again a variant of the method of trial and error. Theories are put forward tentatively and tried out. If the outcome of a test shows that the theory is erroneous, then it is eliminated; the method of trial and error is essentially a method of elimination. Its success depends mainly on three conditions, namely, that sufficiently numerous (and ingenious) theories should be offered, that the theories offered should be sufficiently varied, and that sufficiently severe tests should be made. In this way we may, if we are lucky, secure the survival of the fittest theory by the elimination of those which are less fit. (1)
As Lakatos' concessions to Kuhn indicate, even Popper's most ardent supporters baulked at this uncompromising picture of science 'red in tooth and claw'. If Popper's 'method of elimination' were applied rigorously then no theory would survive, and science would go back to Year Zero. Every theory will fail if applied beyond its limits; indeed, failed tests are the best means of discovering the range of applicability of a theory. Theories are tools used for exploring the physical world. The successful scientist is the one who uses the right tool for the job, regardless of convention or ideology, and a theory proved 'erroneous' in one application may well be very useful in another. In practice, theories are rarely (if ever) eliminated in the manner described by Popper, whose failure to explain the nature of non-formal rational enquiry is almost as spectacular as that of Hegelian dialectic. This does not refute all of Popper's arguments, but it does allow us to take him less seriously, particularly with regard to his dire warnings against repeating the mistakes of dialectic.
'D for Dialectic' has been included in 'IDEAL' for a number of reasons:
An acknowledgement of the contribution of the ancient Greeks, who were well-aware that rational enquiry involves more than deduction and proof;
Popper's failure to establish 'trial and error' as the sole basis of 'scientific method' raises similar doubts that 'iteration' alone is sufficient to explain every aspect of rational enquiry;
'Dialectic' (as defined above, in the tentative manner advocated by Popper) implies that any design of an automated learning engine must be able to reconcile opposing opinions or facts, by whatever means;
An act of defiance to Popper's absolutism.
Click here to go back
Copyright © Roger Kingdon 2004