Consider a class of computing problem for which all sufficiently short programs are too slow and sufficiently fast programs are too large. Most non-standard problems of this kind were strictly alone for the first twenty-years or so of the computing era. There were two good reasons. First, the above definition rules out both the algorithmic and the database type of solution. Second, in a pinch, a human expert could usually be found who was able at least to compute acceptable approximations -- for transport scheduling, job-shop allocation, inventory optimization, or whatever large combinatorial domain might happen to be involved.