Recently I published a post on the h-index, so, you see, I am concerned about the issues of science production evaluation. Today I was surfing on another publication listing service (google scholar) and I discovered a link to something called “g-index”. I investigated and discovered it is the following: you place all your publications in a list ordered by decreasing citations, then you calculate the cumulative number of citations. Your g index is the rank g of the publication for which this cumulative number is just smaller than g squared.
The theory behind this is in the article by Leo Egghe in Scientometrics. There is nothing metaphysical about the square of g, as Egghe shows, a researcher with a given h-index should receive at least h squared citations for the most quoted works. The author argues that this index corrects the inability of the h-index to evaluate the “power” of the highly cited works. Indeed, once a work entered the list of the most quoted, any new citations will not affect the h-index of that author. Egghe also showed, mathematically, that h <= g.
I calculated mine and it is, indeed, nine units larger than my h-index. I detected a problem, though, suppose a given researcher has many works piled up below the h threshold with exactly h citations (this situation happens with me), his g-index will most probably be smaller than the one of a similar researcher, who has many low cited works below the h threshold (and will accumulate citations slower).
Anyway, it is a new measure of research production. Probably, as always, the evaluation of a production quality will never depend on a single parameter. This one has the advantage to result in larger numbers (I know many people frustrated about low h values). This can power the researcher’s ego.