No it is arbitrary. The science shows that cyclists have a range of optimal cadences, indeed a number of different studies have found wide variations in the determined optimum cadence - Coast & Welch, 1985 , 90–105 rpm; Eckermann & Millahn, 1967, 30–60 rpm; Hagberg, Mullin, Giese, & Spitznagel, 1981, 80–90 rpm; Wildrick, Freedson, & Hamill, 1992, 35–57 rpm for example.
The optimal cadence for any particular rider varies depending on a number of biomechanical differences, and also on what optimisation you are looking for. Foss & Halen found in 2004 that the optimum cadence for the same person increases the greater the amount of work required by the cyclist, therefore a touring cyclist on that measure could well expect to have a lower optimum cadence than a pro racer, simply because cycle touring is normally a lower power pursuit.
I do find it rather annoying to see people blindly quote values like this without understanding the mechanisms behind the calculations, and what the experiments they were derived from were trying to test.