My machines and I are patzers (Chessbase told me so)

This forum is for general discussions and questions, including Collectors Corner and anything to do with Computer chess.

Moderators: Harvey Williamson, Steve B, Watchman

Forum rules
This textbox is used to restore diagrams posted with the fen tag before the upgrade.
User avatar
Monsieur Plastique
Senior Member
Posts: 1014
Joined: Thu Jul 03, 2008 9:53 am
Location: On top of a hill in eastern Australia

My machines and I are patzers (Chessbase told me so)

Post by Monsieur Plastique »

So in the game two days ago between Carlsen and Nakamura, the latter was described as making "...an inexplicable and costly blunder...". An "incredible" one apparently, because the Chessbase correspondent was lost for words over how a chess player could make such a devastatingly horrific move.

So prior to seeing the text move I looked at the position myself (my interest was piqued by such a sensationalist headline) and came up with 37. d6 after about 10 minutes thought. Then I see that I too would have apparently made the blunder of the century. Then I put the position into my dedicated machines...and the best they can come up with is also 37. d6. Even my Nintendo Fritz which is around 2000 ELO on the old SSDF scale couldn't come up with anything better.

So obviously by Chessbase standards I am a patzer and so are all my chess playing machines. Only my engines (Fritz 13 and HIARCS 13) are not patzers. And of course, with those engines running in the background, I couldn't be a patzer because I could instantly see the evaluation drop about 8 points after this horrific blunder.

In all honesty, if I had played 37. Qf1 it probably would have been by accident insomuch as it looks like a nice move :lol:

Anyway, I was wondering if any of the stronger dedicated machines can indeed come up with the "correct" 37. Qf1? It does seem to be a good test for a dedicated machine - it is a sort of "quiet" move that precedes a forcing line. I was actually a little surprised no machine I own played anything better than 37. d6, but then again, I suppose when you have an engine kibitzing and it solves the position in half a second, then it's really easy to knock an inaccurate move whilst sitting at the computer having a cup of coffee. :roll:

Incidentally, I don't really agree with the tone of the article, nor the verbose analysis by GM Chirila. The mistake is not inexplicable at all. It happened because Nakamura is human and he missed the winning move. He probably would have seen it if he had more time on the clock. What is so inexplicable about that? You'd think he was the first GM on the planet to ever make a blunder.

Thank goodness for Daniel King. At least he knows how to comment on games and does not treat people's errors as if they are some laughable descent into manic idiocy. Given the commentary by Chirila and Ramirez is - I assume - aimed at your average Chessbase reader, I can only assume that I am way below par for their readership and am no longer worthy. :wink:

[fen]4r1k1/1p3pPp/6q1/2pPpNP1/p1n1P3/P6Q/7R/K7 w - - 0 37[/fen]
Chess is like painting the Mona Lisa whilst walking through a minefield.
User avatar
Cyberchess
Full Member
Posts: 658
Joined: Wed Jan 08, 2014 6:10 pm

Post by Cyberchess »

Welcome me to the “patzer” club; I wanted to play 37.) d6 also.
:idea: Qf1 is the kind of move I only read about in schach books.
User avatar
Mars
Full Member
Posts: 694
Joined: Tue Nov 27, 2007 9:48 pm
Location: Bonn, Germany

Re: My machines and I are patzers (Chessbase told me so)

Post by Mars »

Monsieur Plastique wrote:So obviously by Chessbase standards I am a patzer and so are all my chess playing machines.
Yeah, that's what I thought about myself as well. Of course armchair commentators having the analysis of three engines running for them would have seen it all.

My first (well maybe second) idea was to give up the g7 pawn with Nh6+ and then support the d-pawn with Qh3-d7-e7 (after Rf8), also threatening the e5 pawn plus some pressure on f7. A plan, Houdini told me, which doesn't lose immediately but for which you need very accurate play later on. Lines I almost certainly would have missed anyway... After seeing the full analysis also "gosh, you'd never have played Qf1". So, same patzer here.

I guess this is a quite unpleasant kind of a "won" position. You need extremely precise play - one which even a 2700 player can miss under time pressure - otherwise you lose spectacularly. Now mix into this the world champ as opponent who can defend like a computer sometimes... and you are really in trouble. Commentators who in the interview later ask you "why didn't you see the obvious Qf1" nicely round up your day...

Deadly Cocktail Regards,
Martin
User avatar
Mars
Full Member
Posts: 694
Joined: Tue Nov 27, 2007 9:48 pm
Location: Bonn, Germany

Post by Mars »

http://www.zurich-cc.com/reports.html

"...And co-commentator Peter Leko added that the correct move 37. Df1 was not easy to find, even when a diagramm was pointing out that this was a winning position."

WC Challenger Regards,
Martin
User avatar
Mars
Full Member
Posts: 694
Joined: Tue Nov 27, 2007 9:48 pm
Location: Bonn, Germany

Post by Mars »

Ok, now at home, I tried the Vacouver 68020 on tournament level (2h in 40) and indeed, in the diagram position it starts to think about 37.d6 but after about 2:30 it switches to 37.Qg4 and also ends up playing this move! The reply 37.- Qb6 is answered by 38.Qf3 (with the idea Nh6+ and a hit on f7) and now i.e. 38.- Qg6 defending back f7 doesn't help:

[fen]4r1k1/1p3pPp/6q1/2pPpNP1/p1n1P3/P4Q2/7R/K7 w - - 0 3[/fen]

39.Rxh7! Qxh7 (39.- Kxh7 leads to mate on the h-file) 40.Nh6+ and either f7 falls with mate or black must give the queen. So the Lang program would not have lost that position to Carlsen... ;)

Well, by the way, in my humble opinion in the starting diagram instead of the engine solution 37.Qf1 rather the (yes, worse) 37.Nh6+ Kxg7 38.Qd7 Rf8 39.Nf5+! Kh8

[fen]5r1k/1p1Q1p1p/6q1/2pPpNP1/p1n1P3/P7/7R/K7 w - - 0 4[/fen]

40.Qe7 is more patzer compatible (although I overlooked even in this line the Zwischenzug Nf5): after 40.- Rg8 41.Nh6 there is no sensible way to defend f7 as 41.- Nd6 fails to 42.Qxd6! and black loses a piece in face of Nxf7. Since here white still needs to convert that into a win, the Vanc solution is objectively better...

A New Test Position is Born Regards,
Martin
User avatar
Steve B
Site Admin
Posts: 10140
Joined: Sun Jul 29, 2007 10:02 am
Location: New York City USofA
Contact:

Re: My machines and I are patzers (Chessbase told me so)

Post by Steve B »

Monsieur Plastique wrote:

[fen]4r1k1/1p3pPp/6q1/2pPpNP1/p1n1P3/P6Q/7R/K7 w - - 0 37[/fen]
The Mega Powerful Fidelity EAS Pre-Budapest will play d6 after 2 Min 45 Sec and a full 5ply search...scoring the position +1.8
at least i think it scores the position +1.8
it only shows Evals in Hexadecimal notation so i might have bollixed up the conversion

Yet Another Patzer Bites The Dust Regards
Steve
User avatar
Cyberchess
Full Member
Posts: 658
Joined: Wed Jan 08, 2014 6:10 pm

Post by Cyberchess »

Mars wrote:Ok, now at home, I tried the Vacouver 68020 on tournament level (2h in 40) and indeed, in the diagram position it starts to think about 37.d6 but after about 2:30 it switches to 37.Qg4 and also ends up playing this move! The reply 37.- Qb6 is answered by 38.Qf3 (with the idea Nh6+ and a hit on f7) and now i.e. 38.- Qg6 defending back f7 doesn't help:

[fen]4r1k1/1p3pPp/6q1/2pPpNP1/p1n1P3/P4Q2/7R/K7 w - - 0 3[/fen]

39.Rxh7! Qxh7 (39.- Kxh7 leads to mate on the h-file) 40.Nh6+ and either f7 falls with mate or black must give the queen. So the Lang program would not have lost that position to Carlsen... ;)
Bravo :!:
Richard Lang’s programs are especially effective in weak hardware environments. Anyone remember the 1994 milestone game whereby an Intel Pentium running Chess Genius defeated Garry Kasparov?

http://www.chessgames.com/perl/chessgame?gid=1070710

Back in those days I was running MChess in DOS. The following screenshot should bring back some memories:

http://chessprogramming.wikispaces.com/MChess
User avatar
Monsieur Plastique
Senior Member
Posts: 1014
Joined: Thu Jul 03, 2008 9:53 am
Location: On top of a hill in eastern Australia

Post by Monsieur Plastique »

Yes, well of course Lang was from the old school where the hardware was incredibly limited (even though he arguably was able to work at the "luxury" end of the market). Still, it's interesting to see how things turned out post "vintage golden age". The Morsch programs, for example (which ran on more modest hardware than the Lang contempories) kept improving over the years in the various Fritz incarnations whereas the Chess Genius line of programs tended to maintain a status quo in relative terms.

Still, there is always an attraction to those sorts of programs - a late model Mephisto Lang in a nice wooden board would be all the computer most of us would ever need.
Chess is like painting the Mona Lisa whilst walking through a minefield.
User avatar
Monsieur Plastique
Senior Member
Posts: 1014
Joined: Thu Jul 03, 2008 9:53 am
Location: On top of a hill in eastern Australia

Re: My machines and I are patzers (Chessbase told me so)

Post by Monsieur Plastique »

Steve B wrote:The Mega Powerful Fidelity EAS Pre-Budapest will play d6 after 2 Min 45 Sec and a full 5ply search
Well you were probably hoping against hope there Steve :lol: I thought you might instead break out some killer hardware. This position is clearly not as simple as it might seem. My machines remain perplexed even two half moves into the "win"!

What I am most interested in as whether the strong Novags can sort out this position. A Super Conny plays the Patzer's Favourite: d6.
Chess is like painting the Mona Lisa whilst walking through a minefield.
User avatar
Monsieur Plastique
Senior Member
Posts: 1014
Joined: Thu Jul 03, 2008 9:53 am
Location: On top of a hill in eastern Australia

Post by Monsieur Plastique »

Mars wrote: A New Test Position is Born Regards
Hi Martin,

I am actually wondering what will be more likely to live in infamy. The position itself, the "blunder" that was the actual move, or the commentary afforded by God's Gifts to Chess - GM Cristian Chirila (who I can only imagine must be under-rated by about 1200 ELO points or so and actually regularly plays around the 3700 ELO mark.

Anyway, he has a website. Next game he plays where he plays an inaccurate move, I am going to commentate it replete with double question marks and critical variations, pointing out his pathetic and unfathomable play and how it could have been vastly improved upon.
Chess is like painting the Mona Lisa whilst walking through a minefield.
User avatar
Cyberchess
Full Member
Posts: 658
Joined: Wed Jan 08, 2014 6:10 pm

Post by Cyberchess »

Monsieur Plastique wrote:Yes, well of course Lang was from the old school where the hardware was incredibly limited (even though he arguably was able to work at the "luxury" end of the market). Still, it's interesting to see how things turned out post "vintage golden age". The Morsch programs, for example (which ran on more modest hardware than the Lang contempories) kept improving over the years in the various Fritz incarnations whereas the Chess Genius line of programs tended to maintain a status quo in relative terms.

Still, there is always an attraction to those sorts of programs - a late model Mephisto Lang in a nice wooden board would be all the computer most of us would ever need.
That should certainly be strong enough for something like 99% of the chess playing population, though having access to programs of different playing styles (openings and evaluation algorithms) is always nice.

Back in “The Golden Age” of PC chess programs, myself and other members of our local chess club tried pretty much everything that was available at the time. These all proved disappointing to some extent, as their positional play was horrendous, hash tables and known endgame solutions were still years away, and dead drawn endgames were easily won by skillful human opponents.

One such early program written in assembly language was Psion by Richard Lang. Some of the early Mephisto programs were ported over and based on this strong for its time program. The PC port of Psion used boring monochrome graphics and lacked mouse input, though was stronger and more human-like than most. I actually preferred this program to the bestselling Chessmaster series with EGA graphics based on Mychess by David Kittinger.

If I had to choose only one program from the ‘80s, it would be without a doubt Sargon IV. It is no wonder that Fidelity Electronics recruited Dan and Kathe Spracklen, “The Dynamic Duo” of chess programming.

http://www.mobygames.com/game/dos/sargo ... Id,377956/

Basking in past glory!

John
RadioSmall
Member
Posts: 273
Joined: Fri Jul 02, 2010 7:12 pm

Re: My machines and I are patzers (Chessbase told me so)

Post by RadioSmall »

Monsieur Plastique wrote:So in the game two days ago between Carlsen and Nakamura, the latter was described as making "...an inexplicable and costly blunder...". An "incredible" one apparently, because the Chessbase correspondent was lost for words over how a chess player could make such a devastatingly horrific move.

So prior to seeing the text move I looked at the position myself (my interest was piqued by such a sensationalist headline) and came up with 37. d6 after about 10 minutes thought. Then I see that I too would have apparently made the blunder of the century. Then I put the position into my dedicated machines...and the best they can come up with is also 37. d6. Even my Nintendo Fritz which is around 2000 ELO on the old SSDF scale couldn't come up with anything better.

So obviously by Chessbase standards I am a patzer and so are all my chess playing machines. Only my engines (Fritz 13 and HIARCS 13) are not patzers. And of course, with those engines running in the background, I couldn't be a patzer because I could instantly see the evaluation drop about 8 points after this horrific blunder.

In all honesty, if I had played 37. Qf1 it probably would have been by accident insomuch as it looks like a nice move :lol:

Anyway, I was wondering if any of the stronger dedicated machines can indeed come up with the "correct" 37. Qf1? It does seem to be a good test for a dedicated machine - it is a sort of "quiet" move that precedes a forcing line. I was actually a little surprised no machine I own played anything better than 37. d6, but then again, I suppose when you have an engine kibitzing and it solves the position in half a second, then it's really easy to knock an inaccurate move whilst sitting at the computer having a cup of coffee. :roll:

Incidentally, I don't really agree with the tone of the article, nor the verbose analysis by GM Chirila. The mistake is not inexplicable at all. It happened because Nakamura is human and he missed the winning move. He probably would have seen it if he had more time on the clock. What is so inexplicable about that? You'd think he was the first GM on the planet to ever make a blunder.

Thank goodness for Daniel King. At least he knows how to comment on games and does not treat people's errors as if they are some laughable descent into manic idiocy. Given the commentary by Chirila and Ramirez is - I assume - aimed at your average Chessbase reader, I can only assume that I am way below par for their readership and am no longer worthy. :wink:

[fen]4r1k1/1p3pPp/6q1/2pPpNP1/p1n1P3/P6Q/7R/K7 w - - 0 37[/fen]
Hi Monsieur Plastique
Thanks for sharing your thoughts on this topic .This is superb test position for dedicated computers .Here I have results for 4 strong units that I own : In the following position :
[fen]4r1k1/1p3pPp/6q1/2pPpNP1/p1n1P3/P6Q/7R/K7[/fen]
Novag Sapphire wants to play 37.d6 at first but changes to 37.Qf1 at 2 minutes and 28 seconds (ply eight)
The RadioShack Master 2200X ,wants to play 37.d6 at first but changes to 37.Qf1 at 3 minutes and 3 seconds (ply seven)
The Fidelity Designer 2265 , wants to play 37.d6 at first but changes to 37.Qf1 at 38 minutes and 50 seconds (ply eight)
The RadioShack Champion 2250 XL wants to play 37.d6 at first but changes to 37.Qg2 after 1 minute .It however goes back to 37 .d6 finally changing again to 37.Qg2 after 8 minutes and 40 seconds (ply eight )

Radio Small
Wardy
Full Member
Posts: 940
Joined: Tue Jul 31, 2007 7:17 pm
Location: Wellingborough

Post by Wardy »

My own favourite from the 80's is Colossus 4.0 on the spectrum. For no other reason than just being in awe of this program that could search 170 moves a second. For a naive 14 year old it just seemed like a monster!

In fact my first engine match was Colossus 4.0 against Cyrus II both on a pair of spectrums..... 18 games, 2 for each of Cyrus 's levels....... And the result was a draw :-)

I now have both on emulators, I might see what they both make of this position over the weekend.....
User avatar
Monsieur Plastique
Senior Member
Posts: 1014
Joined: Thu Jul 03, 2008 9:53 am
Location: On top of a hill in eastern Australia

Re: My machines and I are patzers (Chessbase told me so)

Post by Monsieur Plastique »

RadioSmall wrote:Novag Sapphire wants to play 37.d6 at first but changes to 37.Qf1 at 2 minutes and 28 seconds (ply eight)
The RadioShack Master 2200X ,wants to play 37.d6 at first but changes to 37.Qf1 at 3 minutes and 3 seconds (ply seven)
The Fidelity Designer 2265 , wants to play 37.d6 at first but changes to 37.Qf1 at 38 minutes and 50 seconds (ply eight)
The RadioShack Champion 2250 XL wants to play 37.d6 at first but changes to 37.Qg2 after 1 minute .It however goes back to 37 .d6 finally changing again to 37.Qg2 after 8 minutes and 40 seconds (ply eight )
Very interesting results. Thanks for doing the tests. I suspected the strong Novags would get it right. What is the 2200X a clone of? I'm not good with the RadioShack machines and the 2200X isn't actually listed at schachcomputer.info.

Maybe I should retest my Nintendo as well. It has a strange characteristic in that it plays too fast from a setup position versus how quickly it moves in an actual game. So maybe it would have found the correct move as well if it had occurred in an ordinary game, given that earlier Morsch programs are at least not playing d6.
Chess is like painting the Mona Lisa whilst walking through a minefield.
User avatar
Cyberchess
Full Member
Posts: 658
Joined: Wed Jan 08, 2014 6:10 pm

Post by Cyberchess »

Wardy wrote:My own favourite from the 80's is Colossus 4.0 on the spectrum. For no other reason than just being in awe of this program that could search 170 moves a second. For a naive 14 year old it just seemed like a monster!

In fact my first engine match was Colossus 4.0 against Cyrus II both on a pair of spectrums..... 18 games, 2 for each of Cyrus 's levels....... And the result was a draw :-)

I now have both on emulators, I might see what they both make of this position over the weekend.....
I recall the Timex Sinclair being heavily advertised here in the U.S. in those days, but I had to google the “Spectrum” computer. Apparently this brand name was marketed in the U.K.

In retrospect, I think that the proprietary operating systems and programs made things rather confusing at the time. Back around ’82, I was a computer science/technology major that didn’t own a PC. Some of my friends invested in Apples and Commodores, but I held out knowing that the longer I waited, the more bang I would get for the buck so to speak. Also, cash was tight in those days, so I took full advantage of my college’s technology resources. When not busy working on our programming assignments, some of us brought in PC DOS based chess programs on 5&1/4” floppy discs, later to become 3&1/2” and less prone to damage.

I didn’t actually buy my own PC until I got a super deal on a like-new I.B.M. 386SX with peripherals from a small company that folded. It was then that I tried to make up for lost time by experimenting with as many chess and shogi programs as I could. My first PC was actually a decent position cruncher once I installed the maximum amount of R.A.M., deleted the inane Windows 3 OS shell and ran everything through a DOS batch file driven government menu system. Since I only had one complete setup, running program matches was impossible. Instead, we had lots of human vs. human matches via phone modem.

John

PS – please let us know if these older programs eventually find the correct move.
Post Reply