Sign In

Communications of the ACM


Please Explain

Bertrand Meyer

Dear Master:

I have been told that some institutions judge computer scientists by their H-index in the Web of Science or Scopus databases. I have even seen scientists defined by their scores, like baseball players by their batting average,  as in  "Professor Jill Smith (Scopus 36)". These numbers determine hires, promotions and careers. I am unschooled and inexperienced so I am sure you will easily answer my naïve worries:

1. An article with one author brings that author the same recognition as another with five, 10, or 100 co-authors. Is that good?

2. Jean Brown has 20 articles with 20 citations each. Jane Jones has 19 articles with 20 citations each and one with 10,000; she seems to have had a significant influence. They have the same H-index.  Is that good?

3. These databases do not count books, so Knuth's multi-volume Art of Computer Programming, probably the most influential CS publication of all time, cited by zillions of other works (in fact almost 40,000 for one of the volumes from another source), does not contribute to his H-index. Nor do other milestone contributions such as the Design Patterns book. Is that good?

4. The main argument I have seen (other than "others use them") is that the numbers correlate with Nobel Prizes. Is that good? There is no Nobel in CS. Besides, age too would correlate (few 19-year-olds get the prize).

5. Age may be too simple, so I am thinking of suggesting the S-index instead: shoe size. Is that good?

Please, Master, enlighten me!

Your humble student.


Daniel Mackinlay

A superior method would be the C-index, the Chocolate Consumption Index, based on Messerli's pioneering work in the New England Journal of Medicine showing correlation between Nobel Prizes and chocolate consumption. (preprint)

Displaying 1 comment

Sign In for Full Access
» Forgot Password? » Create an ACM Web Account
ACM Resources