H. G. Muller wrote on Tue, Sep 27, 2016 02:13 PM UTC:
Well, computer programs usually do not rank pieces, the programmers usually do that for them. The quality of play also turns out to have surprisingly little effect on piece values.
Fairy-Max self-play showed the Falcon value to be slightly above that of a Rook, perhaps equal. (Pieces close in value tend to pull their values towards each other, complicating the measurement.)
Well, computer programs usually do not rank pieces, the programmers usually do that for them. The quality of play also turns out to have surprisingly little effect on piece values.
Fairy-Max self-play showed the Falcon value to be slightly above that of a Rook, perhaps equal. (Pieces close in value tend to pull their values towards each other, complicating the measurement.)