It's easy to bang your head against a wall trying to find a decisive checking combination that ends in mate or winning major material. In order to solve this, you have to find the right plan which involves:

1) Finding the right weak spot to attack.
2) Finding the right way to attack it.

Ultimately, the right weak spot is the the pawn on g2, and the right way to attack it is with the right piece coordination. There's a huge difference between a piece that is attacked by the queen first and the rook secondarily, vs. the other way around. And there's a huge difference between a piece that is defended by the king first and the queen secondarily vs. the other way around.

Getting there involves a fair bit of calculation.

1. ... Qh2+ forces 2. Kf2, the only legal move.

Now white's weakness on g2 is clear. It's "protected" twice, but since the first one is the king, it's vulnerable to a double attack. So now black must have the patience to play a move that isn't a check:

2. ... Rg6! Black threatens to attack that weakness in the optimal setup: Rook leads, queen supports, while white is trying to defend it with the worst setup, king first.

White's only move now is Ke1 (Qf3 loses the major exchange to Rf6, as does Rg1 after Qg3+/Kf1 Rf6+ and the queen is forced to step in the way).

3. Ke1 Rxg2

Black takes the pawn and threatens all kinds of seventh rank hijinks, but he gives up the initiative. If you are going to give up the initiative in a Q and R endgame, you better be really sure you can't lose because of it. Now it's white's turn to threaten bad things. Amazingly, white has no way to directly attack black's back rank weakness. His only threatening square is Qf3, threatening Qxf7+ and Qf8#.

4. Qf3

Black has to prevent the immediate mate threat, and also not allow white any other major progresses. h6 creates a luft square, but allows white to equalize by taking the pawn with Qxf7+, then threatening a perpetual with Qf5+. Black must block the check with Rg6 and he's gained nothing, and now white would have a clearly superior position. h5 is a little better, but still needlessly gives up a pawn and allows white a solid attack.

f6 prevents Qxf7+, but white gets several clear lines with which to check. But ultimately, black escapes with


4. ... f6
5. Qd5+ (or Qa8+ leads to the same thing eventually) Kf8
6. Qd8+ Kf7
7. Qd5+ Kg6
8. Qf5+ Kh6

and now white is out of safe checks and black is about to maraud him on the seventh rank and can force a trade of queens if he ever wants to. White's best possibility now is to play Rf2, allowing the trade of all remaining heavy pieces. Black enters a 6v5 king and pawn endgame with connected outside passers, and should win trivially. (Another option, which a computer will like less but might have more practical chances, is what white played in the game: Qf4+, which keeps the rooks on the board but leaves white just as lost down two pawns and having the less active rook).

I just thought that was a fascinating position. It shows a lot of key maneuvers in R+Q v R+Q endgames, as well as the need for raw, ruthless calculation in both seizing the initiative and knowing when you can afford to give it up.

in move 4 white has to prevent the mate threat not black

The game ended (from the top position) 1.Qh6 Qd4 2.Rf1 e6 3.fe Rxe6 4.Kh1 Ne4 5.Be8? (Iphone Shredder can't decide on a move but seems to think White is better before this)


giving up the bishop for a tempo

...Ng6+ 6.Kh2 Nxf1 7.Qxf1 Qe5+ 8.g6 Qxg6+ 9.Kh1 Re8? (taking on e8 must give Black an edge but he was happy to get out alive) 10.Qb2+ f6 11.Qxf6 Kh6 1/2-1/2 with a perpetual.

The Qa8 and Ra1 are not participating so we need to get them back in the game. Once we manage to consolidate the game is over (should take like 5 moves from here?). Dark square control should automatically improve while doing this (too bad the other Q is not on c1, Qf8+ would work then). I'd start with 1.Qa6 or 1.Re1; gotta make sure there is no nasty surprises lurking somewhere, but something like 1.Qa6 Bc7 2.g3 e3 3.Re1 should win soon enough.
Summarizing,
What White wants to do:
- trade pieces
- activate the major pieces currently on the a-file
- consolidate
- get the dark squares under control
What Black wants to do:
- act quite fast
- get in a cheap shot
- create threats on the b8-h2 and a7-g1 diagonals

Qxh6 Re8 Qxe6

Re2! Seriously? I almost couldn't believe the computer when it suggested that.

If white takes the queen, or even if he trades them off and then wins a rook with Qh8+, black's remaining pieces come up with a stunning forced mate.

1. ... Re2
2. Rxf6 Ne3+
3. Kf3 (forced, only other legal moves lead to Re1#) Bg4+
4. Ke4 (only legal move) Nxc2+
5. Kd3 (only legal move) Ne1#

When I saw that, I was all ...

wow, that's beautiful.

In the second diagram, I think the Bc8 should be on d7

Aren't you in dire straits anyways? The position looks a lot like a Marshall Attack where somehow white ended up with the attack. After Bd7, i want to make a Rxh6 sacrifice work, maybe with 2. h4 so that at the end of the 3. Qh7+ Kf8 4. Rxh6 gxh6 5. Bxh6+ Ke7 sequence we have 6. Bg5, winning the Q. To me it's not at all obvious how black defends against that.

Yes to dire straits, yes to Marshall Attack where white ended up with the attack, yes black has some awkward defensive work ahead of him and yes to not being able to find it

White should make the trade. He'll enter a bad v. good bishop scenario, but his potentially outside passed pawn and his queen's access to b6 should be decisive.

The fact that white has a bunch of pawns on dark squares does not make white's bishop bad in itself.. There is an important distinction as to whether the bishop is "inside of the chain" or "outside" of it.

I was black, he declined the trade with Re1. It went:

1. Re1 Bg5
2. Qe2 Bh4

White should play g3 here and he didn't, instead playing...

3. Rf1

And now I think it's flipped and it's better for black to trade the rooks, because his queen is going to be the one with slightly quicker access to some key squares to attack the king and weak pawns.

3. ... Rxf1
4. Qxf1 =+ imo

Unfortunately, I made a move that I regretted almost immediately

4. ... Bg3
5. Bxg3 Qxg3

I was thinking that my queen on g3 would be in good position to try to win the Q+Ps endgame. Maybe not a sure win, but no real losing chances unless I blundered. But I almost immediately realized that trading bishops made it the drawiest draw that ever drawed. He got his queen to b6 and I perpetualed his king, gg.

I think every book I read has a different set of definitions for making the distinction between a bishop that is on the same color as central pawns (traditionally called bad) and a bishop that is not good in the position. The phrase "Many times, a bad bishop is good" encapsulates everything that is frustrating about chess books.

if anything white should be better because of his safer king and greater activity. I would expect it to be drawn though.