Main problems with evaluation functions:

Superlatively, evaluation functions should be quick calculates. Wherever is chance they take a long time to estimate, so after then less of the space will be such a searched in a given time limit. Exceptionally evaluation functions should also match the genuine score in goal states. Noticeably there is, isn't true for our weighted linear function in chess, because goal states only score 1 for a win and 0 for a loss.

If ever, we don't could do with the match to be accurate - we can require any values for an evaluation function, as long it scores more for better board states.

A bad evaluation function can be disastrous for a game playing agent. Because there are two main problems with evaluation functions. After that very firstly, same like certain evaluation functions only make sense for game states that are quiet. So that a board state is quiescent for an evaluation function, f , if the value of f is suspect to exhibit wild swings in the near future. We noticed in this type of example that in chess, board states like as one whether a queen is threatened by a pawn, where one piece can take another without a similar valued piece being engaged back in the just next to move are not been quiescent for evaluation functions such as for the weighted linear evaluation function mentioned above. Well now to get around this problem, we also expand that game state until a quiescent state is reached, which type of the value of the function for that state, we can compose an agent's search so more sophisticated by implementing a quiescence search, wherever by, given a non-quiescent state that just we want to evaluate the function for. If quiescent positions are very much more likely has done so then non-quiescent positions in a search, so after then such any extension to the search will not slow things down too very much more. Well in chess, a search strategy may choose to delve any further into the space moreover a queen is threatened to try to avoid the quiescent problem.