Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Discussion / Fritz 11 - a big step forward
1 2 3 Previous Next  
- - By Juergen Faas (**) Date 2007-11-18 10:04
I don´t think the new Fritz is stronger than Rybka 2.3.2a despite the 7.5 - 6.5 match result:
http://f23.parsimony.net/forum50826/messages/175417.htm
But it´s obviously dangerous enough. Vas, look out! :)

Jürgen
Parent - By Bouddha (****) Date 2007-11-18 11:35
I've said it.

DF11, will be close to 232a

Ok at that time R3 shoult be out but we currently know nothing about R3

rgds
Parent - - By Uri Blass (*****) Date 2007-11-18 11:48
No
Fritz11 is weaker than Rybka even when the 32 bit version of rybka is used.

http://husvankempen.de/nunn/phpBB2/viewtopic.php?t=878

Fritz 11 vs Rybka 2.3.2a w32 1CPU    45.5-54.5 (final result)
Fritz 11 vs Rybka 1.2f x64   60.0-80.0 (final result)

Uri
Parent - - By Felix Kling (Gold) Date 2007-11-18 12:06
"Fritz 11 vs Rybka 1.2f x64   60.0-80.0 (final result)"
That's nice, those guys who bought Rybka 1 about 1 3/4 year ago still have a better engine than the latest Fritz :)
Parent - - By Luckybaer (**) Date 2007-11-18 12:55
Although it is early, those results indicate improvement from Fritz 10.

A very, very strong engine, plus the GUI and the playchess membership will make Fritz 11 a very attractive buy.
Parent - - By Felix Kling (Gold) Date 2007-11-18 13:07
Indeed, if compared with Fritz 10, the engine seems to have improved quite a bit, but if compared to Rybka I wouldn't call it "very very strong" since it's still clearly behind Rybka. But ok, it seems to be at least one of the best 5 now.
The changes in the GUI are limited, so it's indeed attractive for those who don't have Fritz 7 ,8 , 9 or 10 but if you have one of those versions and Rybka, I'm not sure why you should buy it.
Parent - - By Luckybaer (**) Date 2007-11-18 14:58
Man cannot live on Rybka alone.  :-)

I like to use multiple engines to analyze positions and such.  Rybka is definitely one of them.  I also like DS11 and H11 and F10.  They may not be stronger than Rybka, but they are still very, very strong. 

After all, Magnus Carlsen may be 71 Elo behind Vladimir Kramnik, but he is still a very, very strong GM, right?  ;-)
Parent - - By Vempele (Silver) Date 2007-11-18 15:05

> After all, Magnus Carlsen may be 71 Elo behind Vladimir Kramnik, but he is still a very, very strong GM, right?  ;-)


Most people don't have enough money to buy them both. This could lead to piracy. :-(
Parent - - By Luckybaer (**) Date 2007-11-19 12:29
The idea that "Since I can't afford both, I'll pay for one and steal the other" is sad.

If you can't afford it, don't buy it.  Or, if you know something is coming in the future, save up for it if possible.  DF11 is not being released until March.  That's 3+ months.  Try socking away $20-40 USD to help pay for the program in March.
Parent - - By Vempele (Silver) Date 2007-11-19 13:41 Edited 2007-11-19 13:52
I think Kramnik and Carlsen (where did you get the idea I was referring to Fritz?) are quite a bit more expensive... :)

>The idea that "Since I can't afford both, I'll pay for one and steal the other" is sad.


I'm tempted to start a poll to find out whether people think stealing both of them is morally better than doing that. :)
Parent - - By Luckybaer (**) Date 2007-11-20 05:05
lol.  Krammnik is probably a bit more expensive than Carlsen at the moment!  I wonder how much Anand goes for?  :-)
Parent - - By turbojuice1122 (Gold) Date 2007-11-20 10:38
It's a good comparison with the Rybka and Fritz situation.  Anand is the far less arrogant player, and yet is also a bit stronger (sure, players have strong tournaments, too, such as Kramnik's recent 2900 performance) and goes for less.
Parent - - By Carl Bicknell (*****) Date 2007-11-20 12:06
which one is arrogant? Carlsen or Kramnik?
Parent - By turbojuice1122 (Gold) Date 2007-11-21 02:19
Kramnik, of course; hopefully Carlsen doesn't go that route--Kramnik was quite humble when he was young, too.
Parent - - By Banned for Life (Gold) Date 2007-11-20 20:49
How do you know that Anand will go for less?

Regards,
Alan
Parent - - By gala.martin (**) Date 2007-11-20 21:09
I heard somewhere, a probably 1/2 years ago, that Carlsen was a cheap one. I understand this is quite an "unreferenceble" statement.
Parent - By Banned for Life (Gold) Date 2007-11-20 21:17
Carlsen would expected to be less expensive than Kramnik, when he was "World Champion" a year ago, or Anand, who is "World Champion" today. I don't see any reason to expect Anand, as WC, to be cheaper than Kramnik, as WC, but maybe I will be shown to be mistaken on this point.

Regards,
Alan
Parent - - By turbojuice1122 (Gold) Date 2007-11-21 02:20
I don't know that Anand will go for less than Kramnik, but that is my belief :-).  I see Anand as more of a crowd lover/pleaser.
Parent - - By Banned for Life (Gold) Date 2007-11-21 02:25
I believe that Anand will negotiate to get as much as the market will bear, just as Kramnik did. While he is doing this, he will smile more and play in more tournaments and will continue to be more popular than Kramnik. These things are not mutually exclusive.

Alan
Parent - - By Permanent Brain (*****) Date 2007-11-21 02:32
Talking of Kramnik and Anand, did you notice that they will play 2 rapid, advanced chess games against each other on Friday?

http://www.kramnik.com/
(see calendar)

It is mentioned on the Tal Memorial website too, but no details are given.

Rybka versus Rybka? :-D
Parent - - By turbojuice1122 (Gold) Date 2007-11-21 02:58
Yeah, we were talking about this I think about ten days or so ago concerning whether these guys would be able to hold their own against the best individual Freestylers.  I'm of the school that thinks the top Freestylers, even working only as individuals, would mostly beat them in advanced chess.
Parent - By Banned for Life (Gold) Date 2007-11-21 03:06
I am of the opinion that if Anand or Kramnik were willing to invest a hundred hours in training for this type of event, either of these gentleman could beat any of the PAL/CCC individual freestylers. Unfortunately, we'll never know...

Regards,
Alan
Parent - - By Banned for Life (Gold) Date 2007-11-21 03:53
Here is a complete tutorial on "advanced chess" from Kramnik's recent interview:

And what do you think about the so-called “advanced chess” match with Anand, where you can use the computer’s help?

This is purely a show. In this form of play, the computer’s role is 80% and yours only 20%. But this 20% is decisive. You have to know how to use the computer well, when to turn it on and when to switch it off. I hope I manage to do so.

After reading this, my money will be on Anand! :-)

Regards,
Alan
Parent - By turbojuice1122 (Gold) Date 2007-11-21 03:59
Perhaps, though he could be purposely playing ignorant to try to lure Anand into a false sense of security.  I don't think it will work, though--Anand is one who  more tends to play the situation, not the opponent.
Parent - - By Vempele (Silver) Date 2007-11-21 08:18

> You have to know how to use the computer well, when to turn it on and when to switch it off.


LOL, I sure hope he knows he should only switch it off if it crashes due to too much overclocking. :)
Parent - By Banned for Life (Gold) Date 2007-11-21 22:45
It would be very amusing if they started up the match by:

1) Ripping open their computers,
2) Installing better cooling,
3) Going into the BIOS to overclock, and
4) Installing a 64-bit OS to get a 60% speedup with Rybka

before starting up their engines.

Actually, I'm wondering which engine they will end up playing with. I'm assuming that this will be chosen by which engine distributer is willing to pony up the most sponsorship money rather than by the contestants, but I haven't heard anyone publicizing the advanced chess match, so maybe this isn't the case. I guess we'll see in a few days.

Regards,
Alan
Parent - By wem511 (**) Date 2007-11-19 14:15
I agree, nice reasoning
Parent - - By George Tsavdaris (****) Date 2007-11-19 22:35

>Indeed, if compared with Fritz 10, the engine seems to have improved quite a bit, but if compared to Rybka I wouldn't call it "very very strong" since >it's still clearly behind Rybka.


It is not clearly behind Rybka but we just have an indication that is behind in the current setting of course.....
Also except from engine-engine games there is the analysis field where Fritz 11 may be better than Rybka there. Or may not of course.....
Parent - - By Felix Kling (Gold) Date 2007-11-20 17:03
"Also except from engine-engine games there is the analysis field where Fritz 11 may be better than Rybka there"
MAybe there are some positions where Fritz's evaluation is better than Rybka's, but since Rybka is the stronger engine (and the best positional engine at the moment for sure), the quality of her analysis is simply higher than Fritz's.
I know, Chessbase is trying to tell you that playing strength and quality of analysis is sth. different, but playing strength is proportional to quality of analysis, so no wonder that the top players (Anand, Morozevich, Shirov, Grischuk, Short, ...) use Rybka! (And who else could judge better on the quality of analysis?)
Parent - - By George Tsavdaris (****) Date 2007-11-20 17:22
I agree that is logical to believe that the strength on engine-engine games in the current used time controls is proportional to the analysis level of the engines, but i haven't seen any proof about this so i can't believe it just because it is logical.....

If you have a list of a positional test suite AND a tactical test suite AND an endgame test suite, for different time periods(10 minutes per move, 1 hour per move, etc) that shows after all these 3 tests that Rybka is better than Fritz 11, then you can convince me, but until then i just know that Rybka 1CPU seems to be better from Fritz 11 1CPU in eng-eng matches only and not for analyzing purposes too.

As for the GM's you mentioned the use of Rybka does not mean Rybka is the best since perhaps all these GM's haven't compared all other existing engines for analyzing purposes.....
And also where did you see that these GM's use Rybka only and not Fritz for their analysis? And only is important since if they use Fritz too then your argument is not valid....
Parent - - By Felix Kling (Gold) Date 2007-11-20 17:39
Some of them use also other engines maybe, but they use Rybka as main engine.
Test suites only tell you sth. about the quality of the engine in extreme situation which are not as frequent as the type of positions which occure in engine-engine games. So I doubt if they can really tell you more, anyway, Rybka is pretty good in those tests, only in solving tactics some engines are faster, but Rybka winfinder is the strongest engine there, so you have got the strongest engine and the strongest tactical engine with Rybka :) . For endgames I know only this test: http://glareanverlag.wordpress.com/2007/09/07/hundert-schach-endspiele/#more-116 , Rybka wins there too.
Parent - - By Permanent Brain (*****) Date 2007-11-20 20:20 Edited 2007-11-20 20:25
WinFinder 2.2 scored best in my tactical test, but if I get it right, it's asymmetry causes it to be very good an the active/attacking sense, but not good if it's about to find the best defense, or avoid an attractive move which falls into a deep trap or something like that. In other words, is it true that WinFinder is not an "overall" tactical engine (but only in the sense of it's name, to search winning moves)?

Two examples where WinFinder 2.2 is relatively slow:

r1b1Rbk1/pp3p2/2npN2p/2qp2p1/8/1QPB3P/PP3PPB/6K1 b - - bm Bxe6; id Quick-06;


1...Bxe6 is the toughest defense.

2kr3r/ppp3pp/2pbbn2/4N3/3Pp3/2P3Pq/PP1NQP1P/R1B2RK1 w - - am Nxe4; id Quick-09;


Avoid 1.Nxe4 which is a blunder.

As for the topic Fritz 11 vs. Rybka, I think George WANTS to believe that Fritz 11 is better, and will defend this desire against any facts, as long as possible. Maybe it's like in love, where facts are almost nothing and emotions rule. :-D
Parent - By George Tsavdaris (****) Date 2007-11-20 22:03

>As for the topic Fritz 11 vs. Rybka, I think George WANTS to believe that Fritz 11 is better, and will defend this desire against any facts, as long as >possible.


You are wrong to think that.....

And can you quote any statement of mine that i say that i believe Fritz 11 is better?
Parent - - By Felix Kling (Gold) Date 2007-11-20 21:19
btw., since you can't measure "quality of analysis" (it's always a subjective thing) you will never see a proof since there can't exist a proof :)
You can only collect some facts to show that Rybka ist most likely the best program for analysis at the moment.
Parent - - By turbojuice1122 (Gold) Date 2007-11-21 02:29
The definition of "quality of analysis" is that which is most Fritz-like. :-)
Parent - By Banned for Life (Gold) Date 2007-11-21 03:08
Yes, I looked it up in the Chessbase dictionary and you are right. :-)

Alan
Parent - - By Banned for Life (Gold) Date 2007-11-20 21:01
Felix,

I'm not sure I agree with this because I think that playing style is also important. For instance, suppose you could have your games annotated by Tal, or a GM that is 100 Elo below Tal but plays in a more conventional manner? I'm not sure that Tal's analysis would be more helpful (at least to me :-)).

The second aspect is that engine analysis can (and in my opinion should) be done going backward through the game. Rybka is certainly the strongest engine going forward through a game, but because of the way things are maintained (or not maintained) in hash, I'm not sure that Rybka is strongest going backwards (i.e. it definitely has a harder time keeping multiple lines in hash than, for instance, Zappa does). Permanent hash sometimes helps, but doesn't really resolve this issue (I'm hoping Vas improves this aspect in version 3).

Feel free to rebut these comments. I certainly wouldn't analyze without Rybka, but its relative strength in analysis is not nearly as clear to me as its relative strength in engine-engine play.

Regards,
Alan
Parent - - By Felix Kling (Gold) Date 2007-11-20 21:28
"suppose you could have your games annotated by Tal, or a GM that is 100 Elo below Tal but plays in a more conventional manner? I'm not sure that Tal's analysis would be more helpful"

I don't agree on this. I'm also a more positional player but Tal's analysis would aim more fore attack, that's true, but would also show a good understanding of the position. This would mean that I could choose other lines but Tal would have explained the main strategical ideas and maybe also some quiet lines he personally wouldn't play. Reducing Tal on a "attacking only" player seems a bit too simple to me. (The same for Morphy, he was also not that bad at positional chess!)

Anyway, since all the top players use Rybka (and they have different styles for sure, also active, attacking players like Topalov and Shirov use her... which type of analysis should Fritz offer that Rybka doesn't give?) so why do you still think like this? It's somehow a very strange way of thinking.
Parent - - By Banned for Life (Gold) Date 2007-11-20 22:01
It depends what the goal of your analysis is. Assuming the goal is to allow a player to improve his performance in future games, it might not help to have Tal showing missed brilliant moves, that only Tal could find during a game. If the goal is more academic (i.e. what are the possibilities if I could play like Tal), than Tal's analysis would certainly be best. I think this is an important distinction and I don't think there is much merit to the argument that if its considered best by Topalov and Shirov (and this is unproven as far as I know), it must be best for most chess players.

Putting this a different way, if I had exactly the same clubs as Tiger Woods, I'd still be a crappy golfer.

Alan
Parent - - By Nelson Hernandez (Gold) Date 2007-11-20 22:21
Your Tiger analogy: this is a case of not knowing what you don't know vs. his not knowing what he knows. 
Parent - - By Banned for Life (Gold) Date 2007-11-20 22:31
Among other things. I do find it very amusing though when people suggest that I can dramatically improve my game by getting better clubs (I already have very good clubs). For some reason, I hear this every 2 or 3 holes! :-)

Alan
Parent - - By Nelson Hernandez (Gold) Date 2007-11-20 22:40
If that is possible, I am an even worse golfer than I am a chess player.  A 72 handicap would not be enough.  And you would need an ambulance following me from hole to hole.  Golf is a very dangerous game when I am on the course, both for me and others; careful awareness of where I am and what I am doing plus quick reflexes are essential.
Parent - By Banned for Life (Gold) Date 2007-11-20 22:50
I am also a master at hitting the ball in all directions (including once during the back-swing :-)), but I have not seriously injured anyone yet.
Parent - - By turbojuice1122 (Gold) Date 2007-11-21 02:35
Nice Donald Rumsfeld quote reference :-)
Parent - By Nelson Hernandez (Gold) Date 2007-11-21 03:07
You're alert.  Always amazed by the level of cultural literacy in this forum. 
Parent - - By Carl Bicknell (*****) Date 2007-11-19 11:39
"A very, very strong engine, plus the GUI and the playchess membership will make Fritz 11 a very attractive buy."

yes but Deep Fritz 11 would have been much more attractive. Now Deep Fritz 11 will be competing against Rybka 3...bad strategy by chessbase.
Parent - By noctiferus (***) Date 2007-11-19 13:20
even worse: they could catch customers now, and offer an upgrade fee on march
Parent - By noctiferus (***) Date 2007-11-18 12:26
What do you think about these results (leaving Rybka aside, of course)?
IMHO, it's a big jump, with respect to the other engines, isn't it?
Parent - - By George Tsavdaris (****) Date 2007-11-19 22:39

>Fritz11 is weaker than Rybka even when the 32 bit version of rybka is used.


And how do you know for sure? These are just indications that it is not stronger. But error bars and level of confidence show that we are not sure to speak with 100% confidence that Fritz 11 is not stronger.....

Also even if it is not stronger from Rybka in Fritz-Rybka games, even with 99% level of confidence and huge amount of games where range of errors does not concur, perhaps it will be stronger if we play against many other engines and get an ELO for Fritz....
Parent - - By Permanent Brain (*****) Date 2007-11-19 22:53 Edited 2007-11-19 22:58
After 820(!) 40/4m games at CEGT, the difference is 59 Elo. Their head-to-head result was Fritz-Rybka, +26 =39 -35 (45.5%)

http://www.husvankempen.de/nunn/40_4_Ratinglist/40_4_AllVersion/27.html

You can decide to believe that the truth is somewhere outside of the 95% confidence, as the error margins don't overlap, but how realistic is it? I can accept such figures easily. What is the problem? I also don't think that longer time controls will close that gap.

Btw. I noticed something strange on that list... Rybka 2.3.2a/2 CPU is rated lower than on 1 CPU, 2950 and 2973?! Error margins are small for both (600 games have been played with the 2 cpu version). Seems strange to me.
Up Topic Rybka Support & Discussion / Rybka Discussion / Fritz 11 - a big step forward
1 2 3 Previous Next  

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill