I have been moving various files from Queen Mary to St Andrews. After the recent discussion about what constitutes publication, I started wondering about who counts as a co-author.
I believe I have 152 coauthors of published or accepted papers. But MathSciNet puts the number at 140, and the Erdős Number Project homepage only gives 129. Interesting to compare?
The first thing to notice is that both MathSciNet and ENP overcount by 1, since they don’t realise that Alejandro, Priscila P. is the same person as Kazanidis, Priscila A.
The MathSciNet rules (or what I guess are the rules, I couldn’t find this stated anywhere) are that to count as a coauthor you must have a joint paper listed in the MathSciNet database. This rules out Budd, Cooter and Spiegelhalter (with whom I wrote an article for Mathematics Today about the BBC Horizon programme on infinity), Beineke and Wilson (with whom I am jointly credited with the introductory article in Topics in Algebraic Graph Theory), Hirchfeld (joint editor of the proceedings of the first Isle of Thorns conference), and six of my nine coauthors on the paper on intricacy (see here), published under the nom de plume W. E. Opencomb. They also miss Joshua Browning, presumably because the paper is too recent (it went up on the JCT website today).
The ENP do have my coauthors from W. E. Opencomb, but miss out Robert Bailey, whom they initially confused with Rosemary Bailey; and many other people, since it is some years now since the lists were updated.
But in none of these cases do they count the arXiv as publication. Google Scholar does, however, as Robert Bailey pointed out; this would increase the number by four. However, they only found 25 of my co-authors, and offered me about the same number of suggestions of people I might wish I’d written a paper with but haven’t (such as Laszlo Lovász).
So Google Scholar may overestimate publication counts but doesn’t do so well on co-authors!