This seems an interesting position, but just sticking to the question of whether it and similar positions are suitable for Computer Workout: right now, on the whole, they are not.
This reason is that for a position to be usable, two things must be true:
1. The computer knows how to win (or to draw when that is the aim). It doesn't necessarily have to see the winning line from the initial position right to the end, but it should be able to work its way to a win move-by-move. This is necessary so that it will react sensibly when the user hits "Best Move".
2. The computer defends against any winning (or drawing) plan in a sensible way (ie, it doesn't "randomly" give up a rook because it sees it will lose in any case; or, in the case of an endgame like this, it will try to counter whatever the other player is attempting to do). If it doesn't do this, then the lesson is lost.
At the moment, the engine used in Computer Workout is good enough for most tactical purposes but not for long-range plans, and so in the position you've posted the computer just aimlessly shuffles the pieces around. Perhaps in the future the engine will be stronger, I don't know; but given the nature of the position it will probably have to be quite a lot stronger before it becomes usable. Computers aren't terribly good at this sort of stuff.
Won/Drawn Endgame?
camembert, I'm not sure this is one of those positions you mention. As I said, I'm not sure if the simple plan of g5-g6 followed by just marching the king to h5 doesn't work. Rybka suggests this plan without too much thought and thinks it works, so maybe the Chess.com won't pointlessly move around pieces, but see the winning plan.
My warning about g5-g6 allowing Bh6xf4 seems mistaken as well. Sunday I did not calculate the complications through yet, but now I see that after playing g6 and putting the knight on d6 prevents any Bh6xf4 plans, since then simply g6-g7-g8 wins.
Maybe you can let the Computer Workout engine take a look at the position and see if it can spot a forced win for white?
Unfortunately the CW engine seems to have picked up a bug and I can't give it any new problems right now. As soon as I can, I'll get it to look at this position. I'm not optimistic, though, because my local copy of Fruit (the engine that CW uses) has no idea how to proceed here; as I say, it just shuffles the pieces around aimlessly.
I wonder if it can really be won, though. After looking at the position more closely I'm not sure about your claim that trading the g pawn for the e pawn will win. The trouble is that Black's bishop is really quite a good piece. If from your starting position we go Nc5 Bf8 g6+ Kxg6 Nxe6 then ...Bb4 seems very annoying as it threatens ...Bxc3, and this is something that White must constantly guard against. I suppose that to make progress, you have to defend against this with your knight and try to bring the king up on the kingside, but Black has ...Bd8 to keep it out, and I don't see how to break this down. Putting the knight on b7, covering both a5 and d8 might be an idea, but then if the White king gets too lively ...Bxd4 will draw at least (and anyway Black can use his king to keep the white king out, and revert back to ...Bxc3 ideas if the knight moves from b7). The line I've posted here illustrates the sort of thing I mean.
Of course, if Rybka sees a win, I must have overlooked something (I don't have the latest version). Wouldn't be the first time, nor the last :)
camembert, Ba3 instead of Bb4 is always an even bigger threat. That's why white needs to put his king on d1 or d2 first, as I did in the game. Then, trading the g-pawn for the e-pawn should give white a winning advantage, since the b2-c3-d4-e5 chain is safe (as long as white's king guards c1) while the white knight can pick off the whole e6-d5-c4-b3 chain then. All those pawns are on light squares so undefendable for black.
For example 1. Kd2 Bf8 2. Nc5 Be7? 3. g6+! Kxg6 4. Nxe6 Kh5 5. Nc7 Kg4 6. Nxd5 Bd8 7. e6 (the easiest way) and white wins easily.
Sorry, Phobetor, I didn't realise you were playing Kd2 first (in the line you posted you have Nc5 first).
After 1. Kd2 Bf8 may be a mistake, I don't know. But what about 1. Kd2 Bd8 2. Nc5 Ba5 and now if 3. g6+ Kxg6 4. Nxe6 then we have the same thing as I posted above. I still don't see how White wins here.
That's an interesting suggestion. Indeed, it seems that if black keeps his bishop on the d8-a5 diagonal after the trade of g5 for e6 (covering important squares like c7) white cannot make progress, even with the connected passed pawn. So probably the position is a draw for black after all!
God, endgames are amazing! So many fascinating plans and positions with so few pieces! :)
Yes, they're fantastic, aren't they :)
I posted this endgame in my blog also, but there I discussed the game where my opponent made a mistake and I won. If my opponent didn't make a mistake though, I'm not sure if I could have won or if it was a draw. What do you think? And if it's winnable, is it a good position for the Computer Workout?
Below's the position. Some things you should probably consider are: