John Pollock has died

John Pollock, a founding father of the area of defeasible argumentation, has died. He combined theoretical, computational and practical considerations in his design of an ‘artificial person’, OSCAR (see, e.g., his Cognitive carpentry). In this high ambition, he has had no followers.

See, e.g., his influential paper Defeasible reasoning (in Cognitive Science, 11:481–518, 1987; also available in full text). A – too brief – one page introduction to his influential ideas on undercutting and rebutting defeaters is in this text (p. 229) on that other founding father of defeasible argumentation Stephen Toulmin.

A few dozen millions of items of knowledge

I came across Marvin Minsky’s 2006 book _The Emotion Machine_ (draft available at his site). I briefly looked at what he had to say about commensense, the biggest hurdle for AI. Here is how he estimates how much a typical person knows, ‘a few dozen millions of items of knowledge’:

Everyone knows a good deal about many objects, topics, words, and ideas—and one might suppose that a typical person knows an enormous amount. However, the following argument seems to suggest that the total extent of a person’s commonsense knowledge might not be so vast. Of course, it is hard to measure this, but we can start by observing that every person knows thousands of words, and that each of those must be linked in our minds to as many as a thousand other such items. Also a typical person knows hundreds of uses and properties of thousands of different common objects. Similarly, in the social realm, one may know thousands of things about tens of people, hundreds of things about hundreds of people, and tens of useful items about as many as a thousand people.

This suggests that in each important realm, one might know perhaps a million things. But while it is easy to think of a dozen such realms, it is hard to think of a hundred of them. This suggests that a machine that does humanlike reasoning might only need a few dozen millions of items of knowledge.

I found it worthwhile to read a bit in this work by one of the big defenders of  ‘knowledge-intensive AI’; the old-fashioned kind that I am a child of. (I only recently learnt that the pun GOFAI for good old-fashioned AI should probably be pronounced with ‘o’ as in ‘goof’.) I saw that (in the same chapter on commonsense) Minsky briefly comments on the currently fashionable ideas that we need to copy the brain and that we need to harvest the web.

Here are some points in a review by the neurologist Richard Restak that make me curious:

Actually, the loss of cells results from passive disuse — use it or lose it — rather than active deletion.

Of the 1.1 trillion cells in the human brain, only 100 billion are neurons.

[A]natomical interaction of neurons highlights only one aspect of brain functioning. Equally important are alterations of the brain’s chemical messengers, the neurotransmitters, along with changes in local and distributed electrical fields.

I looked at Minsky’s web site after I Wolfram|Alpha’ed him together with Norbert Wiener, about whom Lambert spoke to me yesterday. (Beware: Wolfram|Alpha is the latest geek tool; innovative, useful, addictive.)

Prediction in AI & law

In AI, prediction is associated with people such as Norbert Wiener (who “worked with McCulloch and Pitts and influenced a number of young researchers including Marvin Minsky”, Russel & Norvig 2003, 757). In law, prediction is associated with Oliver Wendell Holmes’ stance towards law: the law should be regarded as a prediction of what brings punishment or other consequences from a court (see Wikipedia; see there how Hart took a different view).

Communication with little grounding

Pieter de Bie is currently in Edinburgh (with Simon Kirby and Thom Scott-Phillips) working on a game in which players must learn to communicate with little grounding. In a recent discussion about the progress of his work I was reminded of one of Léon Vié’s elegant puzzles that appeared in the 1980s in the NRC Handelsblad: a message from space. It is in Dutch, but here is all you need to know to do the puzzle: what is the meaning of the 14th group of signs received by Ipokrypa, an inhabitant of the planet Jupiter, and where does the message come from? If you really need the solution, it is posted at http://bartver.files.wordpress.com/2009/04/vie.pdf.

Peer review and justice

Something to think about every now and then (for me at least):

The peer review process is a form of pure procedural justice (Rawls): there is no criterion for the fairness of its outcome other than the correct application of the procedure itself.

This perspective is enlightening (and sobering) both when receiving a review’s outcome (pos or neg) and when writing a review.

Coincidence by association

I came across a message that I sent to Ron Loui a while ago (November 2007):

In Saint Louis, Missouri, (home of the renowned dialectician RPLoui) there is an apparently renowned restaurant The Seventh Inn owned by Else Barth, who shares the name of another renowned dialectician. And the association net is densified further since this was discovered by Bart Verheij (another dialectician who visited RPL) who lives in Paterswolde (near Groningen) where also Else Barth used to live.

The web says there was a fire in the restaurant so perhaps it no longer exists.

A typical blog entry to be tagged ‘leisure, argumentation’.

Argumentation, mathematics, Lakatos

I have been reading in a special issue of Foundations of Science on the connections between argumentation and mathematics (Vol. 14, Nos. 1-2, 2009); guest editors Andrew Aberdein and Ian J. Dove. There is Andrew’s useful introduction to the issue, and also a paper by Alison Pease and colleagues on the computational representation of Lakatos’s proofs and refutations (see also Mandy Haggith’s ‘A meta-level argumentation framework for representing and reasoning about disagreement‘). Nice! Andrew gives pointers to Wilfrid Hodges’ text ‘An editor recalls some hopeless papers‘ on journal submissions disproving Cantor’s diagonal argument and to recent work by Erik Krabbe connecting pragmadialectics, argumentation and mathematics (‘Strategic Maneuvering in Mathematical Proofs‘; with a comment by Sally Jackson). Hodges refers to psychological work by Rips and also by Johnson-Laird and Byrne – about whose Deduction he is interestingly critical (“I know I am not alone in finding its accounts of logical theory almost incomprehensible”). Krabbe quotes Goethe:

Die Mathematiker sind eine Art Franzosen: redet man zu ihnen, so übersetzen sie es in ihre Sprache, und dann ist es alsobald ganz etwas anders.

Krabbe also discusses a ‘proof’ by mathematical induction that all horses have the same color.

Interesting, and simply: good fun.

Empirical logic, the principle of distributivity and quantum phenomena

In his paper Is logic empirical? Putnam has defended (Wikipedia:  “at one point in his career”) that quantum logic is the ‘right’ logic from an empirical point of view. Quantum logic fails the principle of distributivity: p & (q v r) <=> (p & q) v (p & r).

Knowledge representation

Erik Sandewall reviewed the recent Handbook of Knowledge Representation in the Artificial Intelligence journal. He mentions two contemporary incarnations of Cyc: Freebase and Dbpedia. Here is his final sentence:

[O]ur field is missing a very important part of its mission when we focus so much on logic, algorithms and other methods for the representation and the computational use of knowledge, and when we disregard so regularly the actual bodies of knowledge that our methods are supposed to operate on.

Modeling the behavior of large groups of individuals

Bruce Edmonds has reviewed Philip Ball’s popular science book Critical Mass. As before (in the discussion with Dignum and Liz Sonenberg on the use of logic in agent-based social simulation), he is insightful and skeptical. He quotes Ball’s book:

The skill lies in knowing where a mechanistic, quantitative model is appropriate for describing human behaviour, and where it is likely to produce nothing but a grotesque caricature.

Edmonds lists the following lessons that the social sciences can learn from physics:

• The effort that is put into getting data and information about processes, for example, by developing new methods of measurement and new experimental techniques.
• The relative importance that is given to artefacts ‘lower down’ the chain of abstraction, thus evidence over models; and concrete models over frameworks and paradigms.
• The relative willingness to develop new modelling techniques when existing ones turn out to be inadequate.
• The acceptance of evidence as the ‘court of last resort’ for choosing between competing theories.