What Is Wrong with Cae and Can It Be Xed?

1 Early CAE promises \Contemplation of the similarity between the brain and computers promises to shed light on some of the mysterious workings of the mind and also suggest fruitful directions for computer design and programming," announced Dalkey (1958, p. 3) at the ASCE First Conference on Electronic Computation, in 1958. While acknowledging that the above similarity is only an analogy, he did not nd it diicult to foresee in 100 years a remarkable era, \not because of the practical beneets which will ow from those remarkable machines, great as those beneets doubtless will be, but because the electronic computer will eliminate our last claim to special status in the universe." (p. 3) The details of realizing these promises that were based on the early work on AI included: a machine that will assist in its own programming (that is, a machine that learns), a machine that will be creative by evolving its programs randomly through mutations (that is, achieving creativity by genetic programing), and the improvement in hardware. Thirty six years later, and in contrast to occasional statements by CAE researchers, the forecast is far from reality. I argue that this contrast is rooted in CAE research methodology. 2 What is wrong with CAE methodology? The answer to the question in the section title may be as simple as employing the wrong interpretation of the term CAE. Instead of studying computers facilities for aiding engineers, CAE researchers study computer-aided engineering, often in isolation from engineers. This isolation allows researchers to employ whatever interpretation of engineering they wish. Thus, not only could the research issues be irrelevant to engineers but the research strategy employed could easily prevent from addressing the stated issues. For example, the divide and conquer strategy that is often employed to decompose large issues into smaller issues, focuses on simple \manageable" problems