This is where top Engines fail compared to human............

[peter]

Problem is, what is this one:

[d]6k1/6p1/8/6R1/3q1P2/6P1/6K1/8 w - - 0 1

and this one:

[d]8/5k2/4p3/3q4/4R3/5P2/4PK2/8 w - - 0 1

I suppose they are both draws.

If lomonomonorosow is reliable, you'r right, Lyudmil, btw.

http://tb7.chessok.com/

costed about 20$ per year when I bought an account for the 7men there back in April.

Oops, maybe that's not allowed to write here, is it?
If not, sorry, wouldn't mind it being deleted, dear Mods

[BeyondCritics]

Hey Lyudmil,

i think i have to support you here...

Do you have an android tablet or smartphone (iphone works too)?
If so you should definitely install the lomonosov access app:
https://play.google.com/store/apps/deta ... omonosovtb
Gives you free access to the lomonosov table bases.

To resolve the fortress of the game Kortchnoi-Salov please install Finalgen:
http://finalgenchess.ovh/home_ing.php

It will solve the position in question in "draw mode" in just under 5 hours or so.

Then rework your fortress algorithms and you could make a difference.
Note that such an algorithm is _not_ required to resolve any tactics. If there is forced win in a few moves, but your static algorithm says otherwise this would be ok, since stockfish will overcome this. But if your algorithm says draw in a position which is basically won, this would be a problem.

[Lyudmil Tsvetkov]

another draw:

[d]8/8/4q1k1/5Rp1/6P1/5PK1/8/8 w - - 0 1

this is true only for defended pawns on files g and b, as otherwise the enemy king penetrates

- pawn defending other pawn that defends rook
- rook defending pawn defending other pawn
- enemy pawn blocking pawn defending rook (in this case the enemy queen pawn is not on adjacent file to the file the pawn defending rook is, but on the same file, which complicates things)

=draw

Did not we saw the above position in last TCEC, and neither Komodo, nor SF had a clue this is a simple draw?

well, I think if nothing else, at least this rule for pawn defending pawn that defends rook which rook is on ranks 4 or 5 could easily be implemented.

[Lyudmil Tsvetkov]

I am not certain what this is:

[d]8/8/8/3q1k2/4Rp2/5P2/4PK2/8 w - - 0 1
if this is draw, then if the defending pawn is on its 3rd rank, it ,might also be in the center, and not only files g and b

For Q vs R+2 pawns, whenever pawn defends other pawn that defends rook, that will be a simple draw:

[d]8/8/3q1k2/2R5/3P4/2PK4/8/8 w - - 0 1

why give here 250-300cps advantage?

Seeing all of the above, if you do not want to be extremely accurate, but just gain something, considering a draw any position where pawn defends pawns which defends rook on rank 4 or 5, and enemy only pawn is not a passer for draw might be true in some 80% of cases.

[Lyudmil Tsvetkov]

Hey Lyudmil,

i think i have to support you here...

Do you have an android tablet or smartphone (iphone works too)?
If so you should definitely install the lomonosov access app:
https://play.google.com/store/apps/deta ... omonosovtb
Gives you free access to the lomonosov table bases.

To resolve the fortress of the game Kortchnoi-Salov please install Finalgen:
http://finalgenchess.ovh/home_ing.php

It will solve the position in question in "draw mode" in just under 5 hours or so.

Then rework your fortress algorithms and you could make a difference.
Note that such an algorithm is _not_ required to resolve any tactics. If there is forced win in a few moves, but your static algorithm says otherwise this would be ok, since stockfish will overcome this. But if your algorithm says draw in a position which is basically won, this would be a problem.

Thanks Oliver, I supposed so.

I do not really like those things with small displays, if I use them, it will be just occasionally.

Besides, what are Milos and Peter for?

SF will overcome nothing, as their programmers are too lazy.

[Lyudmil Tsvetkov]

Problem is, what is this one:

[d]6k1/6p1/8/6R1/3q1P2/6P1/6K1/8 w - - 0 1

and this one:

[d]8/5k2/4p3/3q4/4R3/5P2/4PK2/8 w - - 0 1

I suppose they are both draws.

If lomonomonorosow is reliable, you'r right, Lyudmil, btw.

http://tb7.chessok.com/

costed about 20$ per year when I bought an account for the 7men there back in April.

Oops, maybe that's not allowed to write here, is it?
If not, sorry, wouldn't mind it being deleted, dear Mods

thanks Peter.

what are 20 bucks, when are helping the community effort?

[Arpad Rusz]

I suppose this one is also a draw:

[d]8/3pk3/2q5/3R4/4P3/3PK3/8/8 w - - 0 1

if this is the case, which I am almost certain, then indeed this is a very good rule, which could easily be specified, but will unfortunately reflect just a small portion of all rook fortresses:

-pawn defending pawn which in turn defends rook, which rook is on rank 4 or 5
- enemy pawn on a file adjacent to the file where the pawn defending rook is

= simple draw

This is a Lomonosov draw:

[d]8/3pk3/2q5/3R4/4P3/3PK3/8/8 w - - 0 1

but 1.Kf4? already loses! The following is a win:

[d]8/3pk3/2q5/3R4/4PK2/3P4/8/8 b - - 0 1

1...Qh6+ 2.Kf3 Qd2 zugzwang.

I think that the number of self-defences is an important factor in fortess-like positions. In the first position we had 5 self-defences (R->d3, e4->R, d3->e4, K->d3, K->e4), in the final position there are only 4.
Paradoxally, sometimes another important factor for the defender is to have only a few king moves (e.g. near the board's edge). That means that the king cannot be attacked from many directions (e.g. by queen checks).

For example, this is only a pseudo-fortress:
[d]8/8/3k1q2/4R3/3P4/2K5/8/8 w - - 0 1

but this is a true one:
[d]8/8/8/8/1k1q4/2R5/1P6/K7 w - - 0 1

[Lyudmil Tsvetkov]

I suppose this one is also a draw:

[d]8/3pk3/2q5/3R4/4P3/3PK3/8/8 w - - 0 1

if this is the case, which I am almost certain, then indeed this is a very good rule, which could easily be specified, but will unfortunately reflect just a small portion of all rook fortresses:

-pawn defending pawn which in turn defends rook, which rook is on rank 4 or 5
- enemy pawn on a file adjacent to the file where the pawn defending rook is

= simple draw

This is a Lomonosov draw:

[d]8/3pk3/2q5/3R4/4P3/3PK3/8/8 w - - 0 1

but 1.Kf4? already loses! The following is a win:

[d]8/3pk3/2q5/3R4/4PK2/3P4/8/8 b - - 0 1

1...Qh6+ 2.Kf3 Qd2 zugzwang.

I think that the number of self-defences is an important factor in fortess-like positions. In the first position we had 5 self-defences (R->d3, e4->R, d3->e4, K->d3, K->e4), in the final position there are only 4.
Paradoxally, sometimes another important factor for the defender is to have only a few king moves (e.g. near the board's edge). That means that the king cannot be attacked from many directions (e.g. by queen checks).

For example, this is only a pseudo-fortress:
[d]8/8/3k1q2/4R3/3P4/2K5/8/8 w - - 0 1

but this is a true one:
[d]8/8/8/8/1k1q4/2R5/1P6/K7 w - - 0 1

Excellent, Arpad, excellent!

Chess is all about captures, captures mean attacks, and defence is the reverse of attacks.
Engines pay much attention to captures and attacks, but almost no attention to defence, and it is essential.

Btw., going to R+minor vs Q:

[d]8/5p2/2qpk3/R7/3B4/3KP3/8/8 w - - 0 1

I almost fully certain this is a draw, although the queen side even has a pawn more.

So, when you have R+ minor vs Q, it is indeed much easier, all you have to specify is that queen does not have passers or candidate passers and also no significant other relative advantages, i.e. score not bigger than Q value - (R value + minor value).

Why give above 250-300cps advantage and deceive yourself to enter such position, when it is actually a draw, and much better line alternatives have been available?

[Joerg Oster]

OK, this is the critical position. With the King on e4 the Rook has no safe square left on the 5th rank, as on f5 it is skewered and traded. White therefore cannot afford to allow the King to be driven into the open.

What makes this not a fortress is that the Rook is not safe against a Queen sac on f5. The black plan is to bring white in zugzwang by confining the King on g3, to force a Rook move away from h5, and then fork Rf5 and g1.

The fortress-recognition algorithm would have recognized that f5 is not a safe Rook square with the black King on g6, and that the fortress candidate thus collapses under zugzwang.

Very hard to find for humans, I guess. Especially OTB.
I was really almost sure this is a fortress.

Now the interesting thing is that SF without TBs doesn't consider the queen sac at lower depths, because the resulting pawn endgame is evaluated as almost equal. Analyzing from the root position this means, it will take very, very long to find the win.

Obviously, TBs help a lot here, as the pawn endgame will get a much higher static eval.

[Vinvin]

On Wikipedia :
http://en.wikipedia.org/wiki/Fortress_% ... rsus_queen
http://en.wikipedia.org/wiki/Fortress_% ... more_pawns

Problem is, what is this one:

6k1/6p1/8/6R1/3q1P2/6P1/6K1/8 w - - 0 1

and this one:

8/5k2/4p3/3q4/4R3/5P2/4PK2/8 w - - 0 1

I suppose they are both draws.

If lomonomonorosow is reliable, you'r right, Lyudmil, btw.

http://tb7.chessok.com/

costed about 20$ per year when I bought an account for the 7men there back in April.

Oops, maybe that's not allowed to write here, is it?
If not, sorry, wouldn't mind it being deleted, dear Mods

thanks Peter.

what are 20 bucks, when are helping the community effort?