ABSTRACT
We attempt to construct a theory of decision support system design, based on the three independent concepts of reliability, power and speed borrowed from epistemology. We say that a system is reliable if a large part of its performance is useful or correct, that it is powerful if it performs in a useful way in a variety of situations of interest, and that it is fast if its behaviour is consistently dynamic. An arithmetic calculator, for example, is more reliable than a mathematician, but the latter is more powerful. A programming language is as reliable as a calculator, but the calculator is faster.
