Heuristic Reasoning about Uncertainty : An Artificial Intelligence

Negoita's review represents some aspects of my book quite well but misstates a few points and ignores some others. Since the book was published, my purported antipathy for numbers has been the object of contention so often that I nearly began to believe the strident claims people were making on my behalf. Negoita's representation of my position is sober by comparison: "Cohen 's stated purpose is to argue openly against numerical knowledge representation schemes that are currently in vogue in approximate reasoning." This is not the whole picture. Why do we argue for or against particular knowledge representations in artificial intelligence? Why would one criticize numbers when they have served science for centuries? I said: