PDA

View Full Version : Study-like victory



Rhubarb
21-05-2004, 05:06 PM
I've seen studies where Q&N vs. Q is a win with a long series of 20 or more moves. It's typical of Shirov that he should do something similar in a tournament game. It's worth trying to visualise the two main lines after 66...Qf2+.

1.Nf3 d5 2.c4 c6 3.e3 Nf6 4.Nc3 e6 5.d4 Nbd7 6.Bd3 dxc4 7.Bxc4 b5 8.Bd3 Bb7 9.a3 b4 10.axb4 Bxb4 11.0-0 c5 12.Qe2 0-0 13.Rd1 a5 14.Ne5 Qc7 15.Nxd7 Nxd7 16.Qh5 f5 17.Qe2 cxd4 18.exd4 Rae8 19.Bb5 Re7 20.Qc4 Qxc4 21.Bxc4 Rc8 22.Bb5 Nf6 23.Na2 Bd6 24.Bd3 Be4 25.Bf1 Ng4 26.Be2 Bxh2+ 27.Kf1 Bc2 28.Re1 Bb8 29.Nc3 Ba7 30.Nb5 Bb6 31.Bg5 Rd7 32.Rec1 h6 33.Bd2 Rdd8 34.Bxa5 Bxa5 35.Rxa5 Bb3 36.Rc3 Nf6 37.f3 Bd5 38.Nc7 Rb8 39.Rb5 f4 40.Ke1 Kf7 41.Bd3 g5 42.Rcc5 Ke7 43.b4 Rxb5 44.Bxb5 Bb7 45.Bc4 Rxd4 46.Nxe6 Rxc4 47.Rxc4 Kxe6 48.Rc7 Bd5 49.Rg7 g4 50.b5 gxf3 51.gxf3 Bxf3 52.b6 Nd7 53.Rh7 Be4 54.Rxh6+ Ke5 55.Kd2 Bd5 56.Ke2 Ke4 57.Rd6 Nc5 58.Rd8 f3+ 59.Kf1 Bb7 60.Kf2 Nd3+ 61.Kg3 f2 62.Re8+ Kd4 63.Rd8+ Bd5 64.Rxd5+ Kxd5 65.b7 f1=Q 66.b8=Q Qf2+ 67.Kh3 Nf4+ 68.Kg4 Qg2+ 69.Kf5 Qh3+ 70.Kf6 Qh4+ 71.Kf5 Qh5+ 72.Kf6 Qg6+ 73.Ke7 Qg7+ 74.Kd8 Ne6+ 75.Kc8 Qc3+ 76.Kb7 Qc6+ 77.Ka7 Qa4+ 78.Kb6 Qb4+ 79.Ka7 Qa5+ 80.Kb7 Nc5+ 81.Kc8 Qa6+ 0-1

Kevin Bonham
21-05-2004, 11:45 PM
I've seen studies where Q&N vs. Q is a win with a long series of 20 or more moves. It's typical of Shirov that he should do something similar in a tournament game. It's worth trying to visualise the two main lines after 66...Qf2+.

Indeed. The other one is shorter. It's impressive how well the Q+N work together - the starting configuration (which of course he would have seen in advance) is very good for the winning side. It's also study-like because White seems to play the best possible defences but all to no avail.

I've never seen QNvQ first-hand in a serious game. BCO has an amusing practical example Ljubojevic-Hjartarson 1991 in which Black walked into a mate in an otherwise drawn position - but escaped with a draw because the blunder was 47 moves into the endgame and White needed five more moves for the mate. :rolleyes:

Sutek
22-05-2004, 06:25 AM
Indeed it is study like and a nice way to finish the game.
But how does black win after 61.Kf1!?

Regards
Sutek

Kevin Bonham
22-05-2004, 12:07 PM
Indeed it is study like and a nice way to finish the game.
But how does black win after 61.Kf1!?

I'm guessing that he can't, but feel free to disabuse me of that notion. :) A valiant attempt is 61...Ne5 62.Kf2 Bc6 63.Rc8 Nd3+ 64.Kf1 Bb5 but now White has 65.b7! and if Black "wins" the pawn by ...Nc5+ it all goes nowhere after 66.Kf2 Nxb7 67.Rf8. If Black saves the pawn by ...Be2 there's 68.Re8+ Kd3 69.Ke1! with the stalemate defence Re3+ coming, a sorry tale indeed. :( 61...Bc6 just seems to lead to the rook pestering things forever.

My comment about White finding the best defences related only to the QN vs Q ending; it looks like Shirov was a bit lucky to get that ending in the first place and that 61.Kg3 allowing it is a mistake.

Sutek
22-05-2004, 12:29 PM
Hi Kevin,

I couldn't find a win either after 61.Kf1.
Maybe Greg can find one.

Regards
Sutek

PHAT
22-05-2004, 06:58 PM
The pgn stops working at at 3. ... Nf6 for me :?:

Bill Gletsos
22-05-2004, 07:23 PM
Works fine for me.

Rhubarb
22-05-2004, 08:14 PM
The pgn stops working at at 3. ... Nf6 for me :?:
Matt, I think the pgn's OK, but takes a long time to fully load every piece every time you open the page if you have a slowish connection like me (plus Explorer has this annoying habit of saying 'done' at the bottom of the page, even though it still has lots of things left to load).
[Note to gareth: don't quote games in your replies - if you must quote (as opposed to 'post reply') then edit out the actual pgn before posting ;) ]

Steve, Kevin, I haven't looked properly but I can't find a win against 61.Kf1 either (most impressive stalemate defence there Kevin!) although I haven't yet engined it.

I did, however, look at some of the other positions Kozul could have allowed. Before both sides queen, he could have had the positions with (W)Rd8, (B)Bd5, Kd4; or with (W)Re8, (B)Be4,Ke3; or he could have played Q&N vs. Q with Black's King on e4, rather than on d5; and finally there was the variation 64.Rf8 Ke3 65.Re8 Be4 66.Rxe4 Kxe4 67.Kg2 Ke3 68.b7 Ne1+! 69.Kf1 Nf3 70.b8=Q Nd2+ etc. In all cases, Shirov is winning, sometimes in quite different fashion, but all of the variations are relatively easy to calculate. Which makes me wonder: how much did Shirov see? How much time did he have? Perhaps he intuitively realised how strongly positioned his minor pieces were in all of these endings, knew that similar endings were wins, and felt there would be a win? Maybe he'd completely stuffed up a winning ending earlier and Kozul should have drawn with 61.Kf1? Maybe Shirov accurately calculates at a million miles an hour and had seen everything?

I hope he annotates the game, because as far as I know he's intellectually
honest.

Kevin Bonham
22-05-2004, 09:26 PM
and finally there was the variation 64.Rf8 Ke3 65.Re8 Be4 66.Rxe4 Kxe4 67.Kg2 Ke3 68.b7 Ne1+! 69.Kf1 Nf3 70.b8=Q Nd2+ etc.

I've been looking at this one with silicon help and haven't found a forced win yet. May be missing something.


Which makes me wonder: how much did Shirov see? How much time did he have? Perhaps he intuitively realised how strongly positioned his minor pieces were in all of these endings, knew that similar endings were wins, and felt there would be a win? Maybe he'd completely stuffed up a winning ending earlier and Kozul should have drawn with 61.Kf1? Maybe Shirov accurately calculates at a million miles an hour and had seen everything?

I reckon it was a big fat world class swindle. :eek: (I'm not sure if that's true, it just might be.)


I hope he annotates the game, because as far as I know he's intellectually honest.

And also because he won't shirk from exhaustive detail, and if he feels the analysis is especially tricky he'll feel like he has to do it.

Rhubarb
22-05-2004, 11:44 PM
I've been looking at this one with silicon help and haven't found a forced win yet. May be missing something.
No, you're not missing much :) - I hadn't analysed this 4th and final variation properly. I believe you're right; Kozul could still have drawed.



I reckon it was a big fat world class swindle. :eek: (I'm not sure if that's true, it just might be.)
Not a swindle; a demonstration of power/knowledge-over-the-board.



And also because he won't shirk from exhaustive detail, and if he feels the analysis is especially tricky he'll feel like he has to do it
Let's hope so.

eclectic
22-05-2004, 11:56 PM
No, you're not missing much :) - I hadn't analysed this 4th and final variation properly. I believe you're right; Kozul could still have drawed.



Not a swindle; a demonstration of power/knowledge-over-the-board.



Let's hope so.
Is it possible that this (end)game might merit perusal by Karsten Muller in his next Endgame Corner instalment at Chesscafe ?

eclectic

Kevin Bonham
23-05-2004, 03:20 AM
No, you're not missing much :) - I hadn't analysed this 4th and final variation properly. I believe you're right; Kozul could still have drawed.

Yes I'm still not 100% certain but it looks like Kozul always has a line he can walk to survive.


Not a swindle; a demonstration of power/knowledge-over-the-board.

What I meant was that poor White, having played his heart out and apparently had a drawn position right up til the end, falls into the trap of allowing QN vs Q in a lost position rather than a drawn one. (Not really a "swindle" as such but it does look like it was drawn if a different QN vs Q ending was allowed.)

skip to my lou
23-05-2004, 04:19 AM
3R4/1b6/1P6/8/4k3/3n1p2/5K2/8 w - -

Rhubarb
27-05-2004, 02:10 AM
Thanks Jeo, an excellent innovation! I tried the various human/engine vs. engine/human options and didn't find a win for Shirov after Kf1 (although as Kevin pointed out it looks like Kozul could've drawn later anyway).

Myself, I'm from the "silicon as a last resort" school of thought, and am trying to improve my currently weak analytical/calculation ability with a mixture of old-fashioned and (only occasionally) high-tech techniques.

Which brings me to my point: Is chess in its Indian Summer? Will chess die in 20 years when the human capacity for judgement and 'understanding' has been completely overtaken by the brutes? Will humans turn to games like Go, where the larger matrix means that machines will take several more generations to surpass human mastery? And would any such game be likely to replace chess anyway, with its thousand-year pedigree and war-like allegory?

Finally, will any programmer actually try to beat humans by modelling such 'intangibles' as intuition, synergy, planning? Or will they just kill chess by brute force? (btw, whatever the hell happened to neural networks?) I suppose it's a measure of chess's standing, as the intellectual game of choice, that the amount of effort spent on programming a machine to play chess is many hundreds of times the effort spent on all other games combined.

Regards,
Greg

P.S. Shirov just won again in Sarajevo to go to 7/8. He eats 2600-type players for breakfast.

Sutek
27-05-2004, 06:47 AM
Thanks Jeo, an excellent innovation! I tried the various human/engine vs. engine/human options and didn't find a win for Shirov after Kf1 (although as Kevin pointed out it looks like Kozul could've drawn later anyway).

Myself, I'm from the "silicon as a last resort" school of thought, and am trying to improve my currently weak analytical/calculation ability with a mixture of old-fashioned and (only occasionally) high-tech techniques.

Which brings me to my point: Is chess in its Indian Summer? Will chess die in 20 years when the human capacity for judgement and 'understanding' has been completely overtaken by the brutes? Will humans turn to games like Go, where the larger matrix means that machines will take several more generations to surpass human mastery? And would any such game be likely to replace chess anyway, with its thousand-year pedigree and war-like allegory?

Finally, will any programmer actually try to beat humans by modelling such 'intangibles' as intuition, synergy, planning? Or will they just kill chess by brute force? (btw, whatever the hell happened to neural networks?) I suppose it's a measure of chess's standing, as the intellectual game of choice, that the amount of effort spent on programming a machine to play chess is many hundreds of times the effort spent on all other games combined.

Regards,
Greg

P.S. Shirov just won again in Sarajevo to go to 7/8. He eats 2600-type players for breakfast.

Hi Greg,

You could always take up punting?
The silicon beast is excellent at compiling, displaying and number crunching but I don't see many Deep Blue's watching the horses parade at Randwick :)

Regards
Sutek

Ps. They are racing there this Saturday :)

skip to my lou
27-05-2004, 12:38 PM
Which brings me to my point: Is chess in its Indian Summer? Will chess die in 20 years when the human capacity for judgement and 'understanding' has been completely overtaken by the brutes? Will humans turn to games like Go, where the larger matrix means that machines will take several more generations to surpass human mastery? And would any such game be likely to replace chess anyway, with its thousand-year pedigree and war-like allegory?

:lol:


Finally, will any programmer actually try to beat humans by modelling such 'intangibles' as intuition, synergy, planning? Or will they just kill chess by brute force? (btw, whatever the hell happened to neural networks?) I suppose it's a measure of chess's standing, as the intellectual game of choice, that the amount of effort spent on programming a machine to play chess is many hundreds of times the effort spent on all other games combined.

Many people think chess engines use brute force. They are right, they do. They calculate at millions of nodes per second. Deep Fritz 8 on my computer can easily reach 2,000kN/s.

But since chess is so complicated, this brute force is nothing. To even 'survive' against grandmasters, they need an opening book (which the human doesn't get), several (thousand) strategies and plans built in, and then endgame tablebases. If the computer is winning significantly after it loads tablebases, the human is finished. If no tablebases are loaded, the computer has a high chance of losing in the endgame, because there are more possibilities here.

So who is really playing here..? the engine? nope! It is all the programmers and grandmasters that make the program and have access to a (super duper tactical and artificial) silicon brain (that never gets tired) and a library of chess books during the game :p

Rhubarb
31-05-2004, 01:05 AM
Many people think chess engines use brute force. They are right, they do. They calculate at millions of nodes per second. Deep Fritz 8 on my computer can easily reach 2,000kN/s.

But since chess is so complicated, this brute force is nothing. To even 'survive' against grandmasters, they need an opening book (which the human doesn't get), several (thousand) strategies and plans built in, and then endgame tablebases. If the computer is winning significantly after it loads tablebases, the human is finished. If no tablebases are loaded, the computer has a high chance of losing in the endgame, because there are more possibilities here.

So who is really playing here..? the engine? nope! It is all the programmers and grandmasters that make the program and have access to a (super duper tactical and artificial) silicon brain (that never gets tired) and a library of chess books during the game :p

OK, after you've added all the libraries and table bases and some typical strategies, what you're left with is incremental advances in chess software versus exponential advances in hardware. Sure, the software needs to be more sophisticated to take advantage of the increased processing power, but the hardware generally seems to be where most of the improvement comes from.

To say that brute force is nothing is going way too far in the other direction...

skip to my lou
31-05-2004, 08:03 AM
Put it this way,

I have a 3.2Ghz P4 / 2GB RAM / G8KNXP / DSP2 etc etc..

Without books/tb's/strategies it wouldn't stand a chance against an IM. This hasn't changed since even my 1.2 Ghz... Nevermind this..

Do you know how many kN Hiarcs calculates at? Hardly anything compared to Fritz. It hardly uses brute force, yet it is one of the top engines. :hmm:

Rincewind
31-05-2004, 09:58 AM
I think Jeo has a point here. Although you might get exponentional increases in processing power, this is off-set by exponentional increase in tree search for each half-ply. It's not as simple as that but I think the two counter each other and leaves you with something linear or at best polynomial - although I'm leaning towards the linear. Also will processor clock speed continue to double every 18 months? I feel limits much be reached eventually. No idea when that might happen though.

I think there is potential for a better postion accessor/strategic engine to add more "rating points" to a chess engine than say doubling the processor speed. But maybe I'm thinking too much like a human. ;)

skip to my lou
31-05-2004, 12:40 PM
Theoretically there should be no limit to the increase of clock speed. Not sure at what rate clock speed will increase.

I guess whether these high clock speed processors will make it to the average PC all depends on the future cost of alloys.

In the new technologies its just a matter of cooling vs multiplier. Heaps of people got the P4 2.4C, water cooled it and clocked it to 3.2 with no problem. I am also guessing that the 3.2 Prescott's (After cooled) can achieve 4g with no problem. Also prescott has smaller micron 0.09 from memory, so some people predict it can be clocked much higher than 4g.

How about splitting this to hardware?

Rhubarb
31-05-2004, 01:13 PM
Attention everyone.

In a bulletin board first, I am going to discontinue this discussion on the grounds that I don't know what I'm talking about.

eclectic
31-05-2004, 04:36 PM
Attention everyone.

In a bulletin board first, I am going to discontinue this discussion on the grounds that I don't know what I'm talking about.
greg,

is this the "i don't know what i'm talking about" virus ?

others here are already infected.

does it stop them from posting ?

;) :hand:

eclectic