mobility = # of legal moves or mobility = average # of moves per piece or some combinatory weighting of both those factors (m_1)^3/2 + (m_2)^3/2 + ... etc. putting a little more weight on less total moves but large amounts of moves per piece. potential impact on overall playing strength and style. more valuable than certain other aspects of knowledge that could be removed so as to maintain a constant search/eval/MC cpu distribution? thoughts, opinions, criticisms, complications, problems, perks, banes, boons, likes, dislikes. share them all and don't forget to rate me down

You'll have to try several things and see what works for your program. I don't think the average per piece is very relevant.

You should probably only consider moves that don't result in an immediate loss of material. This requires that you implement Static Exchange Evaluation (Google it).

In my program the most important part was giving a large penalty for having pieces with very little mobility.

Notice that unless you get the values just right so the average piece gets a mobility score of 0, you'll have to adjust the material scores for the pieces.

Why would I rate you down?