Bishops of Opposite Colors with Queens

It's commonly understood that in endgames without other pieces, bishops of opposite colors are drawish:  one, two, sometimes even three extra pawns may not win.

Add a bit of additional material, even in a semi-endgame, and this is no longer true.  Instead, the player with the initiative can often win because the opponent's bishop is unable to defend effectively.  This is particularly true when queens are involved.  In this article I'll look at three game positions with opposite-colored bishops that turned out to be anything but drawish.

Game 1 is from a chess.com match, and arose from an Anglo-Dutch.  Both players are apparently attacking:  which one will succeed?

I found it hard to develop any general principles for handling this position.  White often had two or three plausible moves, all but one of which would lead to chasing Black's king endlessly around his bishop.  I noticed that when my queen and bishop were on the same diagonal they were most able to threaten mate, but progress almost always involved breaking up this configuration so as to control more squares.  Otherwise, I had no generalizations at all:  I spent many hours pushing my queen and bishop around the analysis board, missed several wins later spotted by the computer, and and finally found one.  All I really had besides calculation was a short list of objectives:  win the bishop with check, bring my other bishop into play with check, or give mate with queen and bishop on the edge of the board (there's an obvious focal point on g7 but it's also possible to chase him to the queenside and mate him there in some lines).

I feel that there must be some principles governing positions like these, but I don't know them.  I suspect that in a face to face game I would have had great trouble winning this.

Game 2 is from the Washington Class Championships:

A much shorter combination:  White is unable to handle the battery of queen and bishop on the long diagonal, as his queen and rook are ill-suited to protect both g2 and h1.  During the game I was shocked at how rapidly the reduced material can generate mate threats (not all for me, either).

This one has more material on the board, but the family resemblance will become apparent in a few moves:

Many players have learned the rule that opposite-colored bishops mean a draw; this misconception can be fatal with queens around.  The ability of queen and bishop to line up on a focal point (close-up mating square) that the defending bishop cannot possibly cover is the basis of many nasty attacks.  This leads to two rules of thumb.  If your opponent has the initiative in such a position it is worth sacking a pawn, maybe even two, to get rid of the queens and enter a potentially drawn endgame.  If you have the initiative, you should keep the queens on and scour the board for a win.  I find it helpful to make a mental list of ways I could win, so that each candidate move can be judged on its potential to lead to any of them. 

The more ways to win you can identify, the better.  When I have ended up playing ring-around-the-rosie with Q+B it's generally been because I can find only one objective, and the opponent can handle just one.  In the first game above I had three, and he was not able to cover them all.