I once had an argument about that. A very smart colleague of mine claimed to have specified a very complete natural language tense system in Mercury, sort of hard-coded into the language if I understood him correctly. He claimed that it was way better than what general semanticists had come up with and that the use of logic in semantics is totally superfluous.
My point was that it doesn't matter how good his system is, as long as he doesn't separate declaration from implementation, it will be of almost no use. Nobody will be able to run his Mercury program in 100 years from now. Then again, it's not possible to run my Higher Order Logic with Henkin models and Categorial Grammar out of the box either.
So I'm still not sure who had the better argument. Maybe if researchers could agree on a common programming language, we could indeed dispense with descriptive formalisms?
My point was that it doesn't matter how good his system is, as long as he doesn't separate declaration from implementation, it will be of almost no use. Nobody will be able to run his Mercury program in 100 years from now. Then again, it's not possible to run my Higher Order Logic with Henkin models and Categorial Grammar out of the box either.
So I'm still not sure who had the better argument. Maybe if researchers could agree on a common programming language, we could indeed dispense with descriptive formalisms?