One part of this is that computers are now so fast that you can get the answer to fairly complicated questions for large amounts of data without getting anywhere close to the optimal analytical solution.
The need to optimize and the associated knowledge is not quite as much of a problem as it was in the recent past.
This leads to a hidden form of bloat, programs written in a terribly inefficient way that still perform adequately in isolation but that fail to compose in a useful way because that quickly slows things down to the point where it no longer produces an answer in acceptable time.
That probably falls under ignorance in your enumeration but it may be a sub-category labelled 'appears unnecessary'.
There ought to be a an energy label for algorithms :)
Although perhaps not known to most people here, we do need pure mathematics to guarantee that numerical methods work. Although most of this is taken care of by most software packages available today, we needed pure mathematics to create those, and will probably need in in the future as well. However, this work is _hard_, and usually done by those who like math for math's sake.
The need to optimize and the associated knowledge is not quite as much of a problem as it was in the recent past.
This leads to a hidden form of bloat, programs written in a terribly inefficient way that still perform adequately in isolation but that fail to compose in a useful way because that quickly slows things down to the point where it no longer produces an answer in acceptable time.
That probably falls under ignorance in your enumeration but it may be a sub-category labelled 'appears unnecessary'.
There ought to be a an energy label for algorithms :)