Keirsey is useless with a lot of his explanations and his NF/NT/SP/SJ temperaments and his stereotypical descriptions has a more black/white aspect and clearly biases itself towards Ns over Ss.
Unfortunately the MBTI culture in general has used keirsey for a lot of the profiles formed and temperaments. There are more sensible temperements that one can use to distinguish between types, and socionics has already adopted a lot of these with things like descriptions of specific patterns in connection.
Temperamental differences by perceptive disposition:
SPs = Se users
SJs = Si users
NPs = Ne users
NJs = Ni users
Temperamental differences by judgemental disposition:
TPs = Ti users
TJs = Te users
FPs = Fi users
FJs = Fe users
Temperamental differences by same functions used (quadras theory in socionics):
NTJs and SFPs
NFJs and STPs
STJs and NFPs
SFJs and NTPs
Temperamental differences by same theme (clubs in socionics):
Temperamental differences by information disposition (temperaments in socionics):
IPs = Ji-doms
IJs = Pi-doms
EJs = Je-doms
EPs = Pe-doms
If MBTI could expand and implement similar temperamental dispositions MBTI would have far stronger face over socionics as it's psychological understandings work far more intricately in specific thought processes over the function definitions in socionics.