Prof David West wrote: > A computer program, currently, is an attempt to mathematize; and the > goal of traditional computer science is to refine the process of > creating a computer program to the purely formal / mathematical. It is > still an attempt, because a huge gulf remains between what I want and > can say about what I want in a natural language and what the computer > can 'hear' with its mathematical 'ears.' > I think the challenge is not to just facilitate use of natural language. Natural language has the benefit of tolerating ambiguity and thus making it possible to communicate half of an idea while continuing a conversation. But I doubt that it is language issue per se. In describing a problem to, say, a skilled computer programmer, the programmer often models the intent in more operational and precise way than the speaker delivers it. She has the benefit of sharing a lot of common knowledge with the speaker, the challenge of acquiring domain relevant knowledge which she does not share, and also having to keep track of inconsistencies in the story, teasing apart things that are clearly unspecified from those that come from her lack of knowledge of the domain (things that are nailed down but not obviously so).
What we don't have now are programming systems that can take a vague set of propositions and instantiate possible candidate computer programs for evaluation. I think the answer is not magical automatic computer programming that can cope with the most muddled of thoughts, but many computer aided analysis and synthesis tools that help communicate back to the user what is not resolved in their idea and the consequences of that, i.e. I think humans have to change too. I would further argue that computer programs are a good way to move from coarse to refined science. A computer program is more expressive than the mathematical toolbox can handle, but with it can also move in that direction. It's a good staging ground for formalization. Marcus ============================================================ FRIAM Applied Complexity Group listserv Meets Fridays 9a-11:30 at cafe at St. John's College lectures, archives, unsubscribe, maps at http://www.friam.org
