ABSTRACT
I start with a review of some general properties of the collection of procedures (the methods of generating causal links between our terms and the things they refer to).
They’re an open-ended collection, for we’re always developing new ones. Sometimes a “new” procedure is a fine-tuning of an old one, so it gives different answers in certain cases. But technological developments also involve inventing new mechanisms without obvious antecedents.
Procedures, generally, are trivalent and vague. By “trivalent” I mean that there are cases where we apply them, and get the answer “Yes”; cases where we get the answer “No”; and cases where we can’t get any answer at all. These last can arise one way, at least: we may be unable to apply the procedure in a particular situation (try biting into something located in the sun). By “vague,” I mean that in certain cases a procedure may give no clear and unequivocal answer, or the answer may not be stable (repeating the procedure in the same situation results in a different answer). 1
Procedures overlap in their domains of applicability. For example, recognizing gold by means of one’s senses gives the same answer on many items that more sophisticated approaches do. But it doesn’t always agree with them.
With these facts clearly in mind, let’s provisionally adopt the revisionist project: We drop our current language, and instead associate a unique term Pp with each procedure p, where Pp applies to any item p gives the answer “yes” to. 2