The generalized Hamming distance is commonly used in determining a clo
sest crisp relation to a given fuzzy relation (which, in turn, is cruc
ial in studying exact choices originating from fuzzy preferences). Thi
s note illustrates a difficulty attached to this practice, and propose
s an alternative metric which is intuitively more appropriate in the c
ase of fuzzy preferences. Some implications of this new metric on find
ing the nearest exact relation to a fuzzy preference ordering and on t
he understanding of the max-min transitivity property are also discuss
ed.