...I find it troubling that their definition of mathematically talented is 95th percentile (or even 99th percentile). Dr. Stanley's work focused on children who were really at the extreme tail (even well beyond the 99.9 subtest standard used by Davidson). At the nationally competitive level, that's who we are dealing with.
Even my DD, who is not mathematically talented, is consistently at above 99th percentile. She is just not at the extreme like her brother or even her mother.
Performance on a standardized test or in a math competition seems like a narrow definition of math talent to me. Both of these things require speed; what about people who prefer to approach problems s-l-o-w-l-y and consider many different aspects of an idea?
Many things in our society (including chess competitions, tenure decisions, and so) reward speed (e.g. how fast can you make your next move?? How many publications can you get out in the next 5 years??). IMO, when we fail to let people move slowly, we fail as a society in a very important area.
