it's been several years since stephen wolfram published his new kind of science. my reaction to the work had been mixed. on one hand, it offered a clear dissemination of the computational implications of cellular automata.
on the other hand, it argued for a black box physical science, where rational understanding of the mechanisms and principles of the natural world is subsumed by heuristics. the fundamental problem is that these heuristics may or may not explain nature; they simply compute it. of course, such computability has its uses.
now, wolfram is about to release wolframalpha, a new kind of science based processing of the www (and all that it contains), accessed via a familiar one line query. this is of course very exciting and important.
the fundamental problem of new kind of science remains, and it will be fascinating to see the answers that wolframalpha will compute in terms of their relevance, accuracy, and rationality.