This classification 1)Scores are "harmonized", ie transformed in such a way that all tasters have the same average and the same standard deviation.
2) Le coefficient of each taster is computed (Chi-2 based), it measures his "agreement" with the panel of tasters. Tasters may be eliminated if this agreement is too weak.
3) Outliers scores of remaining tasters are whipped out.
4) The final score of a wine is obtained by weighting its scores by the tasters's coefficients.
is neither a eulogy on a particular wine nor an attack against a particular winetaster, but only an attempt to extract a trend from a mass of information (sometimes contradictory) i.e. 8666 scores for 1306 wines (to date).