Welp, can't leave well enough alone:  I fixed Sylv's code to count true
duplicates.  I really tried!  Used ( >1  : != 1  :  >=2) in place of (
==2 ).  Here's full results:

paul2a	1.27
paul2	1.31
Sylv (>=2)	1.37
Sylv (>1)	1.38
Sylv (!=1)	1.43
martin	9.28
jegII	9.40
jeff	10.21
samk1	11.65
zach	12.53
park	12.80
markvh	18.95
paul	19.38
robert	19.84
paolo	21.91
simons	24.44
simonk	29.88
samk2	110.73
david	241.78

Strange that the '==' construct was measurably faster.  But at this
point, I have to agree with the person who said that at this level, the
speed differences between the top 5 are immaterial.