In this draft I describe two different strategies for hiring professors in a given discipline. I take the particular case of a computer science department, but the two strategies are applicable to any discipline.
Assume that 10 research topics within computer science have been identified (for example, computational complexity, databases, image processing, cryptography, etc.) and 10 professors are to be hired, one for each topic. Only one professor can be recruited every year. How to proceed ?
One possibility is to select a different topic every year, make an open call and then hire the best candidate in this topic. After ten years one obtains a CS department with ten professors, and each professor was the best candidate in his field.
A second strategy is to simply make a general call, without pre-selecting any topic. Then every year the best candidate is hired, while making sure that he doesn't work in a topic already covered by the CS department.
The first approach seems widely used in European universities, who generally recruit through specialised profiles, whereas US universities seem to have general profiles. For example, the current CS faculty opening at Stanford says that applicants from all areas of computer science are sought, with high priority given to the candidate's work rather than his sub-area of specialisation within computer science.
This second strategy looks clearly superior, because professors are then recruited from a larger pool of candidates. Namely if for some year all candidates in "databases" are weak, then the first strategy will bring a weak candidate and miss a potentially stronger candidate in say "computational complexity", who could have been recruited by the second strategy. This might explain why US universities get so many Nobel Prizes compared to the European universities. US universities have an incentive to recruit the best professors, because part of their salariy is paid by research grants.