A bishop is good, if it is on an active place, a bishop is bad, if it is on a diagonal what is closed by pawns. Good bishop means active bishop, bad bishop means passive bishop. It is that simple.
Good vs bad bishop

As with a lot of things, the whole good and bad bishop thing is just a generalisation. Its only really meant as a reminder in the opening and middlegame the you don't really want to exchange a bishop on the opposite colour to your pawns. That said, cases like the above show that its not concrete. I think Nunn is just pointing out that mobility and development of a piece is more important than any long term benefits that may be gained from having a theoretically better piece.

In general Bishops are "gooder" with as few of your pawns as possible on their color and "badder" when their mobility is restricted by too many pawns on their color. If both players have the same color Bishop make yours as active (ie mobile) as possible by keeping as many of your pawns off its color as you can. If you have a Kt vs Bishop restrict the Bishops mobility by placing as many pawns as possible on its color. Of course these general rules of thumb need to be applied judiciously in every game and not just blindly followed whenever possible
http://www.thechesswebsite.com/good-bad-bishop/
http://www.dailymotion.com/video/xc92wi_chess-com-good-and-bad-bishops-ivan_school#.UVb4QKLNkTc
http://www.metacafe.com/watch/4219889/chess_com_good_and_bad_bishops_vs_im_tate/

As with a lot of things, the whole good and bad bishop thing is just a generalisation. Its only really meant as a reminder in the opening and middlegame the you don't really want to exchange a bishop on the opposite colour to your pawns. That said, cases like the above show that its not concrete. I think Nunn is just pointing out that mobility and development of a piece is more important than any long term benefits that may be gained from having a theoretically better piece.
Nunn wanted to show that in chess you can't apply "rules" blindfolded. He wrote the same also in the introduction of the book, writing that chess isn't maths and there aren't laws that control it.
I open this post to talk about good vs bad bishop. In his midgame book, John Nunn writes:
The black bishop is "bad" because c4 and d5 pawn are on white squares, but it is on a good diagonal. Instead, the theorically good f1 bishop is not active and at the moment is the worst piece.
Here the position
When I saw the position, I thought the same things of Nunn. Since I've started playing chess, I always judged stupid to judge when a bishop is good or bad, because you have to critically evaluate the position and the activity of the bishop, and I'm happy that one of the strongest GM wrote that. What do you think?