ETAI Newsletter Actions and Change

ETAI Newsletter on
Reasoning about Actions and Change


Issue 97004 Editor: Erik Sandewall 29.9.1997

The ETAI is organized and published under the auspices of the
European Coordinating Committee for Artificial Intelligence (ECCAI).

Debate

Discussion with Wolfgang Bibel about his IJCAI lecture

From: Wolfgang Bibel (answer to Marc Friedman)

Oh. Maybe I said it wrong. I meant that if there are synchronic rules, and transition rules, represented in your talk by two different kinds of implication arrows, then there are two different mechanisms -- one which limits each proposition to use in a single connection, and one which does not.

But then - what should be the problem with this technical change in the syntactic characterization of valid formulas which reflects our semantic use of (two different) implications?

From: Patrick Hayes

I'm surprised to read that any theorem-prover has solved the frame problem, since the FP is a problem in representation, not in theorem-proving, and has nothing particularly to do with how deductions are processed. It is also rather surprising to read that some specialised logic has solved the FP, since to do so its semantics would have to embody all known and future causal laws. Could someone briefly explain how a better deductive search engine, or an exotic logic, can solve a problem in representation?

On what might be a related matter, Bibel claims that ' deduction provides a generic problem solving mechanism ' (response to Friedman, ETAI Newsletter on Actions and Change, 26.9.1997). Taken literally, this is clearly false, since deduction itself provides no mechanism whatever: one only gets a mechanism when one chooses a strategy for performing deductions. For example, unification is not imposed by deduction; other strategies for instantiating universal variables are possible, computationally ridiculous but deductively perfectly valid. So Bibel must be understood as referring not to 'deduction' per se, but to a particular deductive strategy, or class of deductive strategies. Perhaps in his original lecture (which I havn't yet got access to) he tells us which ones they are, but a brief summary would be helpful.