rick@hanauma.Stanford.EDU (Richard Ottolini) (09/04/90)
[ Stimulated by discussion on college course on SciVi. ] In some respects I view SciVi following the evolution of Expert Systems five years ago. Expert systems were initially special case research projects in a few a academic'ish research labs. Then there was a transition period where companies developed generic toolkits called "shells" to make it easier for non-specialist to build expert systems. Now that many of the toolkits are mature, most work is in appplications. Both the special-case stage and toolkit development stage required deep knowledge of AI programing techniques particularly LISP. The mature stage no longer requires this. SciVi is entering the toolkit development stage. Users of the toolkits-- mainly scientists-- won't have to know the gory details of low level graphics programming. Therefore when one designs a course on SciVi, one must decide who the audience is and then what level of graphics programming detail is required. I suspect the current fad of SciVi won't suffer the negative fate of expert systems. ES were a solution without a clear match to problems to solved. Eugene Miya proposed this was possibly the case for SciVi in a comp.graphics debate a few months ago, but I believe otherwise. Also ES companies were essentially one-product companies--their shells. SciVi toolkits are auxilary products of graphics companies, workstation and supercomputer companies. If some fail, it won't kill the company.