I do have the book and have checked out the references:

It says: The occurence of earthquakes above a given magnitude is measured by factor b... In most regions b varies between .7 and 1.0. In Northern California b equals about 0.8 if averaged over a few decades.

The smaller b is, the more quakes in a region.

Also another poster that read this sent some references - Here is more on b-value:
The equation log10N = a - bM (1) [Ishimoto and Iida, 1939; Gutenberg and Richter, 1944], where N is the cumulative number of earthquakes having magnitudes larger and equal to M, and a and b are constants, describes the power law decay of frequency of occurrence with increasing magnitude. The b-value is the slope of the straight line that usually fits observed data well, and is inversely proportional to the mean magnitude m, given the minimum magnitude, Mmin, of the data set m = Mmin + 1/2.3b (2) [Utsu, 1965; Aki, 1965]. Thus, mapping the b-value is equivalent to mapping the mean magnitude, i.e. the mean crack length. (http://www.aeic.alaska.edu/Input/stefan/volcanic.html)

It looks simply like the average number of quakes over a given magnitude in a region in a given amount of time or background seismicity. As to HOW it is calculated ??? I think we need a geophysicist !