SWISSTEST: 1.RobboLito - 2.Rybka - 3.Stockfish

[playjunior]

There are different kinds of "lost". If you can prove against a win with analysis spending a lot of time etc. it does make that move equivalent to a mate-in-1.

[Albert Silver]

I'm not a fan of these types of positions, as if your program is strong enough, it may well see that everything loses, and then choosing the best move is really irrelevant, if you can "see".

For me, choosing the best move is one where this move and no other, achieves a significant or winning advantage. Avoiding a move, in a test suite, is one where the move to avoid leads to an undesired result (a draw or a loss), and other moves achieve a different result.

This was the whole point of my satiric position. Who cares if I don't play Kc8, which is mate in one, when the alternative leads to the exact same result? This for me is an exercise in futility. It is no different than complaining the engine found one forced mate, instead of another. A forced mate is a forced mate.

I take issue with those positions because they prove quite literally nothing. All moves lose forcibly, so who cares which losing move the engine chose?

[Albert Silver]

Some analysis with Stockfish 1.5.1:

3: SwissTest4_03 - Verteidigung, Aristarch45-TheBaron15, CompGame 2005
r1b2rk1/2p2ppp/p2b4/1p6/6Pq/1BP5/PP1P1PP1/RNBQR1K1 w - - 0 1

The different between 1st and 2nd move is almost 1 pawn, and Re4 is NOT the move to avoided. Is so, the Pos. ist solved.

This is the point that Walter means.

Just because Stockfish is incapable of seeing a win for Black, changes nothing. If I played that position against you, let's say via e-mail or whatnot, I will win 100% of the games even if you play Re4. How is this an improvement over other losing moves?

[Glarean]

Actually, I was hoping you'd explain the purpose of testing for moves when all moves have the same end result.
Still, the bishop endgame will make a good class subject. I give a weekly class at my club, usually on tactics and analysis, but this is a good one too, as I'm sure many of the students (the highest rating is only ~1850) will have similar difficulties in evaluating it.

Hmm... Albert the great chess teacher...

But still you don't get the point - I'll explain again:

You have criticized my little study B vs B (SwissTest No.20)

5k2/5ppp/p7/3p1b2/5P2/5P2/PPBK3P/8 b - - 0 1

After the "avoid move" Bxc2?, Black is lost immediately; after only a few moves the pawn endgame is over,
and it's a very easy play for white.

But see the position after Bd7 or Bc8):
- How many moves does it take until the definitive defeat of black? 40? 60? 100?
- And on this long long way, white could make a little mistake...
- BTW: Black has a very difficult position, but is he really lost? When you think so,
then show us a forced win for white after Bd7 or Bc8, please...

- So the only conclusion is: "Don't exchange the bishops!"

As you can see, in reality it's a big difference between the "candidates" and the "avoid-move".
That's chess, my friend, and that's the stuff that you need to teach
your ~1850-kiddys: "Always search for the maximum defense!"
(And engines have to do so too!)

Avoiding a move, in a test suite, is one where the move to avoid leads to an undesired result (a draw or a loss), and other moves achieve a different result.

Yeah, you're thinking in black&white - but chess is in color...

[jarkkop]

After the "avoid move" Bxc2?, Black is lost immediately; after only a few moves the pawn endgame is over,

You mean few, a few is much more than few. i.e

After the "avoid move" Bxc2?, Black is lost immediately; after only few moves the pawn endgame is over,

[Glarean]

After the "avoid move" Bxc2?, Black is lost immediately; after only a few moves the pawn endgame is over,

You mean few, a few is much more than few. i.e

After the "avoid move" Bxc2?, Black is lost immediately; after only few moves the pawn endgame is over,

Yes, thanks.

[michiguel]

Actually, I was hoping you'd explain the purpose of testing for moves when all moves have the same end result.
Still, the bishop endgame will make a good class subject. I give a weekly class at my club, usually on tactics and analysis, but this is a good one too, as I'm sure many of the students (the highest rating is only ~1850) will have similar difficulties in evaluating it.

Hmm... Albert the great chess teacher...

But still you don't get the point - I'll explain again:

You have criticized my little study B vs B (SwissTest No.20)

5k2/5ppp/p7/3p1b2/5P2/5P2/PPBK3P/8 b - - 0 1

After the "avoid move" Bxc2?, Black is lost immediately; after only a few moves the pawn endgame is over,
and it's a very easy play for white.

But see the position after Bd7 or Bc8):
- How many moves does it take until the definitive defeat of black? 40? 60? 100?
- And on this long long way, white could make a little mistake...
- BTW: Black has a very difficult position, but is he really lost? When you think so,
then show us a forced win for white after Bd7 or Bc8, please...

- So the only conclusion is: "Don't exchange the bishops!"

I do not believe this is that simple. With the bishops the victory is straightforward, with the pawn endgame could be tricky because of the pawn structure in the King side. You still have to choose your moves very carefully or could end up drawing.

Miguel

As you can see, in reality it's a big difference between the "candidates" and the "avoid-move".
That's chess, my friend, and that's the stuff that you need to teach
your ~1850-kiddys: "Always search for the maximum defense!"
(And engines have to do so too!)

Avoiding a move, in a test suite, is one where the move to avoid leads to an undesired result (a draw or a loss), and other moves achieve a different result.

Yeah, you're thinking in black&white - but chess is in color...

[Uri Blass]

I'm not a fan of these types of positions, as if your program is strong enough, it may well see that everything loses, and then choosing the best move is really irrelevant, if you can "see".

For me, choosing the best move is one where this move and no other, achieves a significant or winning advantage. Avoiding a move, in a test suite, is one where the move to avoid leads to an undesired result (a draw or a loss), and other moves achieve a different result.

Without discussing about the test suite(did not look at the positions) I think that a move to avoid does not have to be a move that leads to a different result.

You may consider a move as a mistake even if it does not change the theoretic result of the game.

If my opponent have mate in 50 in KQPKQ endgame and I play a move that allow him to trade queens to a simple KPK winning endgame then I consider my move as a blunder.

The optimal move clearly give me better practical chances.

We did not solve chess but even if 1.e4 a5 does not change the theoretical draw result of the game then I will still consider 1...a5 as a positional mistake.

Uri

[Paloma]

> Just because Stockfish is incapable of seeing a win for Black

(-1.29): 15.Re4 Qh2+ shows stockfish knows that white stands bad, but losing later as for example Qf3.

From engine view - and we test engines here - Re4 is still to be the best move prevent the lost immediately. Black wins also against Re4 that's no question, but later such as Qf3 or g3.

That's was the engines should be detected.

[Albert Silver]

We did not solve chess but even if 1.e4 a5 does not change the theoretical draw result of the game then I will still consider 1...a5 as a positional mistake.

Uri

Only if the opponent can make a move that changes the result or evaluation. Let us suppose that chess has been calculated to a draw, and that 1...a5 also leads to the draw. Let us also suppose we are both capable of playing perfect chess. Perfect chess means that we are now reduced to 3 evaluations: white wins, black wins, or a draw.

If I play 1...a5, and am capable of forcing the draw, and there is no way you will play moves that allow you to lose, then 1...a5 is not a positional mistake.

Albert