2 Dh5?

[lucasart]

The more material the more mobility. So they are dependent should I remove mobility or materialCount or both.

Being a smart-ass are we ?

If you read Tord's article, you will see that he talks about orthogonal projection. In the case of mobility, the way to minimize this is to center it, as I explained already in my prevoous post:

Code: Select all

weight(phase, piece) * [nb_squares - avg_nb_squares(piece)]

Now the mobility and material are more orthogonal.

But yes, if you don't center your mobility, tuning becomes hell, because:
* any change in the mobility weight indirectly affects the average piece values (even if the nominal values are unchanged).
* the nominal piece values can be far from the real piece average value, which is somewhat confusing and can lead to counter intuitive tuning results.

[Gerd Isenberg]

The more material the more mobility. So they are dependent. Should I remove mobility or materialCount or both.

Not necessarily, the more pieces the more likely they mutually block and hinder each other, decreasing average mobility. So mobilty covers something beside material - i.e. how pieces cooperate.

See also the discussion of Turing and Slater
https://chessprogramming.wikispaces.com/Mobility#Quotes