No chance

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Harvey Williamson, bob

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Post Reply
Lyudmil Tsvetkov
Posts: 6037
Joined: Tue Jun 12, 2012 10:41 am

No chance

Post by Lyudmil Tsvetkov » Thu Nov 27, 2014 10:07 pm

[d]r4rk1/1b2bp1p/1q2p1p1/3pPnP1/p1pP2Q1/PpP1PN1R/1P5P/R1BB3K w - - 0 17

Here is a position the likes of which I have commented many times in games of mine posted here. White has colossal advantage above, but I will not repeat once more why as it starts getting boring. In the past, engines seemed to struggle with similar setups, but maybe in the meantime they have improved a lot.

How many seconds until your favourite engine sees white mating?

zullil
Posts: 5133
Joined: Mon Jan 08, 2007 11:31 pm
Location: PA USA
Full name: Louis Zulli

Re: No chance

Post by zullil » Thu Nov 27, 2014 10:47 pm

Lyudmil Tsvetkov wrote:[d]r4rk1/1b2bp1p/1q2p1p1/3pPnP1/p1pP2Q1/PpP1PN1R/1P5P/R1BB3K w - - 0 17

Here is a position the likes of which I have commented many times in games of mine posted here. White has colossal advantage above, but I will not repeat once more why as it starts getting boring. In the past, engines seemed to struggle with similar setups, but maybe in the meantime they have improved a lot.

How many seconds until your favourite engine sees white mating?
How about "How many hours until any engine sees a positive score for White?" :lol: :lol: :lol:
(Stockfish favors Black here, at least through depth 40.)

Lyudmil Tsvetkov
Posts: 6037
Joined: Tue Jun 12, 2012 10:41 am

Re: No chance

Post by Lyudmil Tsvetkov » Fri Nov 28, 2014 12:20 am

zullil wrote:
Lyudmil Tsvetkov wrote:[d]r4rk1/1b2bp1p/1q2p1p1/3pPnP1/p1pP2Q1/PpP1PN1R/1P5P/R1BB3K w - - 0 17

Here is a position the likes of which I have commented many times in games of mine posted here. White has colossal advantage above, but I will not repeat once more why as it starts getting boring. In the past, engines seemed to struggle with similar setups, but maybe in the meantime they have improved a lot.

How many seconds until your favourite engine sees white mating?
How about "How many hours until any engine sees a positive score for White?" :lol: :lol: :lol:
(Stockfish favors Black here, at least through depth 40.)
I will be a bit busy for a while, so you might have 10, or even 24 hours, to check all relevant lines with SF. Then we might discuss.

But the mate is there. :D

Lyudmil Tsvetkov
Posts: 6037
Joined: Tue Jun 12, 2012 10:41 am

Re: No chance

Post by Lyudmil Tsvetkov » Fri Nov 28, 2014 12:03 pm

Btw., I forgot to ask: any engine evaluating the root position with less than 2 full pawns white advantage?

zullil
Posts: 5133
Joined: Mon Jan 08, 2007 11:31 pm
Location: PA USA
Full name: Louis Zulli

Re: No chance

Post by zullil » Fri Nov 28, 2014 12:55 pm

Lyudmil Tsvetkov wrote:Btw., I forgot to ask: any engine evaluating the root position with less than 2 full pawns white advantage?
Well, there's this (otherwise) very strong engine named Stockfish ...

Code: Select all

info depth 47 seldepth 75 multipv 1 score cp -28 nodes 431995179068 nps 20798947 time 20770050 pv c1d2 a8b8 g4f4 b7c6 d1e2 b6d8 d2e1 g8g7 e1f2 f8h8 h1g1 g7g8 g1h1 b8c8 h1g2 c6b7 g2g1 c8b8 g1g2 g8f8 g2h1 b7c6 f4g4 f8g7 g4f4 g7g8 f4g4 b8c8 g4f4 e7f8 f4g4 h7h6 g4g2 f8g7 a1g1 d8e7 g2g4 c8c7 g4g2 c7a7 g5h6 h8h6 h3h6 g7h6 f3g5 e7g5 g2g5 h6g5 g1g5
Didn't bother to follow the PV, and I must admit that White sure seems better than Black in this position. I wonder what Komodo thinks of this.

PK
Posts: 780
Joined: Mon Jan 15, 2007 10:23 am
Location: Warsza
Contact:

Re: No chance

Post by PK » Fri Nov 28, 2014 12:56 pm

IMHO this position is not about evaluation at all. Evaluation was important when black got into that mess, now it's about reductions and pruning. White has to find a long sequence of non-forcing moves, many of which temporarily decrease evaluation: Ng1 (lower centralization bonus), Rf3 (indirect attack on enemy king instead on direct one, assuming Nf5 still in place, blocking own knight in any case) Nh3 (knight on the rim) Nf2 (low mobility, as black pawns restrict it) and only now score starts to increase: Qh3 (king attack), Ng4 (king attack) Nf6 (lethal king attack).

Lyudmil Tsvetkov
Posts: 6037
Joined: Tue Jun 12, 2012 10:41 am

Re: No chance

Post by Lyudmil Tsvetkov » Fri Nov 28, 2014 1:39 pm

zullil wrote:
Lyudmil Tsvetkov wrote:Btw., I forgot to ask: any engine evaluating the root position with less than 2 full pawns white advantage?
Well, there's this (otherwise) very strong engine named Stockfish ...

Code: Select all

info depth 47 seldepth 75 multipv 1 score cp -28 nodes 431995179068 nps 20798947 time 20770050 pv c1d2 a8b8 g4f4 b7c6 d1e2 b6d8 d2e1 g8g7 e1f2 f8h8 h1g1 g7g8 g1h1 b8c8 h1g2 c6b7 g2g1 c8b8 g1g2 g8f8 g2h1 b7c6 f4g4 f8g7 g4f4 g7g8 f4g4 b8c8 g4f4 e7f8 f4g4 h7h6 g4g2 f8g7 a1g1 d8e7 g2g4 c8c7 g4g2 c7a7 g5h6 h8h6 h3h6 g7h6 f3g5 e7g5 g2g5 h6g5 g1g5
Didn't bother to follow the PV, and I must admit that White sure seems better than Black in this position. I wonder what Komodo thinks of this.
I do not know if we should play a bit of that, SF will lose anyhow, but the point is that black can do nothing here:
- can not play f7-f6
- can not play h6/h5, as files for attack will be opened
- can not transfer the king to the center or queen side, as the h7 pawn remains unprotected
- does not have any counterplay on the queen side or in the center, as the position there is closed

On the other hand, while black shuffles along, white can rearrange its pieces in a very efficient way:
- Bc1 could go to f4
- Ra1 could go to g1
- after h2-h4 is played or even before that Nf3 could go to g4, certainly possible with a long manoeuver, and then land on f6 with devastating effect, etc.

So that black can only wait, while white gradually builds up the attack. At a certain point, the attack will become too powerful, and SF will see mate. :)

I think the above plan is the most correct in terms of positional play, but the white position is so wonderful that other approaches are also possible:
- play Nd2, Bf3 and e4
- play Nh4, exchange the black knight and attack along the h file, etc.

Actually, I think any move should win for white, except for Qf5 and Ra2. Even transferring the white king to the center wins.

The problem is why SF does not recognize the tremendous white superiority in terms of eval, even eval at the root without any search? No kidding, white is better there by at least 2 full pawns, if all relevant factors are taken into consideration. For me it is obvious, but why SF does not see it?

Interesting how Komodo evaluates the above position. I would also join Louis in kindly asking someone to post some Komodo analysis of the position, maybe even its root evaluation.

But really no engine sees white is better with at least 2 full pawns here?

Lyudmil Tsvetkov
Posts: 6037
Joined: Tue Jun 12, 2012 10:41 am

Re: No chance

Post by Lyudmil Tsvetkov » Fri Nov 28, 2014 1:49 pm

PK wrote:IMHO this position is not about evaluation at all. Evaluation was important when black got into that mess, now it's about reductions and pruning. White has to find a long sequence of non-forcing moves, many of which temporarily decrease evaluation: Ng1 (lower centralization bonus), Rf3 (indirect attack on enemy king instead on direct one, assuming Nf5 still in place, blocking own knight in any case) Nh3 (knight on the rim) Nf2 (low mobility, as black pawns restrict it) and only now score starts to increase: Qh3 (king attack), Ng4 (king attack) Nf6 (lethal king attack).
Perfect assessment, Pawel!

That is how the game should be played.

But I do not agree about eval: all those moves do not change significantly the fact that white is much better. The point is why the engines think it is black edge or close to zero, when actually white has big advantage in terms of eval?

SF got into that mess because it thought this position is a promising one, but it actually is not. That is all about eval.

Is Rodent more successful than SF?

yanquis1972
Posts: 1762
Joined: Tue Jun 02, 2009 10:14 pm

Re: No chance

Post by yanquis1972 » Fri Nov 28, 2014 2:17 pm

Lyudmil Tsvetkov wrote:
PK wrote:IMHO this position is not about evaluation at all. Evaluation was important when black got into that mess, now it's about reductions and pruning. White has to find a long sequence of non-forcing moves, many of which temporarily decrease evaluation: Ng1 (lower centralization bonus), Rf3 (indirect attack on enemy king instead on direct one, assuming Nf5 still in place, blocking own knight in any case) Nh3 (knight on the rim) Nf2 (low mobility, as black pawns restrict it) and only now score starts to increase: Qh3 (king attack), Ng4 (king attack) Nf6 (lethal king attack).
Perfect assessment, Pawel!

That is how the game should be played.

But I do not agree about eval: all those moves do not change significantly the fact that white is much better. The point is why the engines think it is black edge or close to zero, when actually white has big advantage in terms of eval?

SF got into that mess because it thought this position is a promising one, but it actually is not. That is all about eval.

Is Rodent more successful than SF?
there is an extremely weak g pawn upon which all this hinges. id again like to see it actually demonstrated against a machine.

Lyudmil Tsvetkov
Posts: 6037
Joined: Tue Jun 12, 2012 10:41 am

Re: No chance

Post by Lyudmil Tsvetkov » Fri Nov 28, 2014 2:28 pm

yanquis1972 wrote:
Lyudmil Tsvetkov wrote:
PK wrote:IMHO this position is not about evaluation at all. Evaluation was important when black got into that mess, now it's about reductions and pruning. White has to find a long sequence of non-forcing moves, many of which temporarily decrease evaluation: Ng1 (lower centralization bonus), Rf3 (indirect attack on enemy king instead on direct one, assuming Nf5 still in place, blocking own knight in any case) Nh3 (knight on the rim) Nf2 (low mobility, as black pawns restrict it) and only now score starts to increase: Qh3 (king attack), Ng4 (king attack) Nf6 (lethal king attack).
Perfect assessment, Pawel!

That is how the game should be played.

But I do not agree about eval: all those moves do not change significantly the fact that white is much better. The point is why the engines think it is black edge or close to zero, when actually white has big advantage in terms of eval?

SF got into that mess because it thought this position is a promising one, but it actually is not. That is all about eval.

Is Rodent more successful than SF?
there is an extremely weak g pawn upon which all this hinges. id again like to see it actually demonstrated against a machine.
With pleasure, only if Louis' SF would like to collaborate, when Louis gets more free time.

Post Reply