![[us]](/mwf/v12/flags/us.png "United States")
Hello Vas and All,
I'm new to the forum, but just watched the Rybka - Ehlvest pawn odds match live. Firstly, Congratulations to
Vas and Larry for the remarkable performance and score for Rybka!
My question would be for Vas. Do you now have an idea of when you might release Rybka 2.3.2? Will the
match have any effect as to the release date? Will Rybka 2.3.2 include all know fixes and improvements
to date, inclduding anything learned from this match such as improved endgame performance, etc.?
Gary
I'm new to the forum, but just watched the Rybka - Ehlvest pawn odds match live. Firstly, Congratulations to
Vas and Larry for the remarkable performance and score for Rybka!
My question would be for Vas. Do you now have an idea of when you might release Rybka 2.3.2? Will the
match have any effect as to the release date? Will Rybka 2.3.2 include all know fixes and improvements
to date, inclduding anything learned from this match such as improved endgame performance, etc.?
Gary
![[is]](/mwf/v12/flags/is.png "Iceland")
"A full update to Rybka 2.3, fixing a number of issues, will be made sometime in March 2007." (Vas)
For more information look here: http://rybkaforum.net/cgi-bin/rybkaforum/topic_show.pl?pid=3872#pid3872
For more information look here: http://rybkaforum.net/cgi-bin/rybkaforum/topic_show.pl?pid=3872#pid3872
Yes that is an interesting question.Will the Rybka-Ehlvest match have any impact on the release of 2.3.2? I just have to say that 2.3.1 is a remarkable accomplishment as the strenght of this program (I think) is the strongest yet.Will the lessons learned be implimented in the new version2.3.2?More than likely your findings will not be conclusive when 2.3.2 is released concerning the past match and hence no implications will be added.But just checking as some lessons concerning Rybka's positional behaivor are no doubt to be learned from this GREAT match.
FWCC
FWCC
I would vote for Rybka 2.3.2 being "delayed" to fix all know bugs and to incorporate the potential improvements
gleaned from the Elhvest match. Ofcourse complete testing of these improvements and fixes as well.
Maybe a rematch with Elhvest before release?! :)
Gary
gleaned from the Elhvest match. Ofcourse complete testing of these improvements and fixes as well.
Maybe a rematch with Elhvest before release?! :)
Gary
![[gb]](/mwf/v12/flags/gb.png "United Kingdom")
I disagree, the problem with fixing bugs is that you create new ones! :-(
![[ua]](/mwf/v12/flags/ua.png "Ukraine")
sounds like neverending story. is there any programm free from bugs?
Yes, I once wrote a program called "Hello world" :D
![[hu]](/mwf/v12/flags/hu.png "Hungary")
You forgot the punctuation, it's "Hello, world." :)
Vas
Vas
You're a Grammar teacher now?
Why, should I keep my day job?
Vas
Vas
Upsa, ok, there were "grammar bugs" in it :)
Were ?! I still haven't even seen the code :)
Vas
Vas
Arghl... you answer too fast, I always need some time to correct my answers since my English isn't that good.. :)
I think it was like this (I deleted it):
<?php
echo 'Hello world!';
?>
:D
I think it was like this (I deleted it):
<?php
echo 'Hello world!';
?>
:D
What language was the OS on your computer? :)
Vas
Vas
(Just checking for improper localization :))
Vas
Vas
Yes, but it's not a one for one deal. Eventually you will get a bug free product.
Actually it is estimated that no large product is bug free or even close to that...
The size below which this is possible can be the topic of a beautiful and never ending argument.
I guess a "Hello, world" program may be small enough (actually that's an argument for taking the "," off :)
Besides, do not forget the "Rule of cybernetic entomology":
"There is always one more bug".
The size below which this is possible can be the topic of a beautiful and never ending argument.
I guess a "Hello, world" program may be small enough (actually that's an argument for taking the "," off :)
Besides, do not forget the "Rule of cybernetic entomology":
"There is always one more bug".
Yes, quite true.
What I like to do sometimes is just read Rybka's inner-loop code. Not with any purpose in mind, just to have a look and admire it :) Always (and I really mean always), I end up changing something.
Vas
What I like to do sometimes is just read Rybka's inner-loop code. Not with any purpose in mind, just to have a look and admire it :) Always (and I really mean always), I end up changing something.
Vas
Which brings up the question: What percentage of the time do you do full regression testing after each change? :-)
In the past, I would usually collect some data or look at some positions specific to the change. This can't in good conscience be called 'regression testing' :)
This is changing now, though. Computer chess is actually a field which is well-suited to good testing.
Vas
This is changing now, though. Computer chess is actually a field which is well-suited to good testing.
Vas
This version will:
1) Fix known bugs
2) Be a bit better tested
3) Be released in the not-too-distant future
The rest is not 100% clear yet. We also haven't decided yet if the anti-human version will be released.
Vas
1) Fix known bugs
2) Be a bit better tested
3) Be released in the not-too-distant future
The rest is not 100% clear yet. We also haven't decided yet if the anti-human version will be released.
Vas
![[ph]](/mwf/v12/flags/ph.png "Philippines")
[quote]We also haven't decided yet if the anti-human version will be released.[/quote]
I suggest a separate engine version, or as a parameter that's set to "off" by default. I'm sure only a few would want one
I suggest a separate engine version, or as a parameter that's set to "off" by default. I'm sure only a few would want one
![[fi]](/mwf/v12/flags/fi.png "Finland")
> I suggest a separate engine version, or as a parameter that's set to "off" by default. I'm sure only a few would want one
You missed the point. He's not sure if he should keep it private (to get an advantage in future matches against humans). Several people have already expressed interest.
![[dk]](/mwf/v12/flags/dk.png "Denmark")
In my opinion people just want too much for their 34 and 59 euro. It is not enough that they got all Rybka versions so far that have been released and they want more.
If that is the case they should open their vallets and pay a little more for this extra service provided by Vas.
I myself have come up with this suggestion to Vas in creating anti human mode in Rybka, and I wouldn't mind in paying a little more to get it. Considering all this time Vas has been spending improving his engine, getting rid of bugs, and everything I hardly think he ever sleeps anymore. Surely he needs some effort for all his hard work.
If that is the case they should open their vallets and pay a little more for this extra service provided by Vas.
I myself have come up with this suggestion to Vas in creating anti human mode in Rybka, and I wouldn't mind in paying a little more to get it. Considering all this time Vas has been spending improving his engine, getting rid of bugs, and everything I hardly think he ever sleeps anymore. Surely he needs some effort for all his hard work.
Oh I almost forgot, for taking all the crap from unhappy customers and everything else people write in this forum.
![[ee]](/mwf/v12/flags/ee.png "Estonia")
So true! I think the Rybka is the best value for money in the chess engine market. I can't understand those who are complaining that 2.3 version was disappointmet for them because it has no big ELO jump compared with previous versions. I think Vasik is doing really great job and deserves to be properly rewarded for that.
I second that.
And Vas I would definitely support you if you put the anti-human rybka out for an extra charge. You deserve it.
But if you plan to use it for more human matches you can still release it.
This is because kaufman said that any strategy against rybka will only work once.
Also, even Kramnik had fritz before the match and still couldn't do much against it.
And Kramnik is a lot stronger than Elhverst (even vs. computers) and anti-human rybka is a lot stronger than Fritz.
Thus anti-human rybka >>>>>>>>>>>>>>>>>>>>>>>>>>>> Elhverst.
So by logic anti-human Rybka will still be the favorite in any near future human match if it is released before.
And Vas I would definitely support you if you put the anti-human rybka out for an extra charge. You deserve it.
But if you plan to use it for more human matches you can still release it.
This is because kaufman said that any strategy against rybka will only work once.
Also, even Kramnik had fritz before the match and still couldn't do much against it.
And Kramnik is a lot stronger than Elhverst (even vs. computers) and anti-human rybka is a lot stronger than Fritz.
Thus anti-human rybka >>>>>>>>>>>>>>>>>>>>>>>>>>>> Elhverst.
So by logic anti-human Rybka will still be the favorite in any near future human match if it is released before.
Alright! alright, what the hell is an "anit-human" chess engine, considering the fact that we get our butts kicked by "all" chess engines-unless your IM-So! what is the distinction. Does the "anti-human" chess engine when it beats the crap out of go on to destory your hard drive and some how fry you in bargin?
Sidserious
I'll take my answer logged-out. Thanks
Sidserious
I'll take my answer logged-out. Thanks
Ha Ha Ha
Good point!
But it will be mainly for people aiming to use an anti-computer strategy (Like lock the pawns up).
It will kill them (literally) if they try.
Good point!
But it will be mainly for people aiming to use an anti-computer strategy (Like lock the pawns up).
It will kill them (literally) if they try.
By Michael Waesch ![[de]](/mwf/v12/flags/de.png "Germany")
Date 2007-03-12 23:46
Maybe we should name it "Anti-Pablo" for more clarity ...
Mike
Mike
Mike,
Are you talking "Paulo" if so I don't think there would be any recognition that there was an end-its like that little bunny with the batteries.
sidserious
Actually, recognition is the wrong word- it would be a lack of "Cognition."
Are you talking "Paulo" if so I don't think there would be any recognition that there was an end-its like that little bunny with the batteries.
sidserious
Actually, recognition is the wrong word- it would be a lack of "Cognition."
By Michael Waesch
Date 2007-03-13 02:02
I meant Pablo, nick "Father" on Playchess who is known for generating stonewall/fortress positions in play against engines, shuffling his pieces behind left and right very fast in order to exploit a time-GUI bug, forcing the engine to accept a draw by rule or forfeit on time. Some say this isn´t even chess - I myself just find it boring. If you saw one og his games, you have seen `em all.
Mike
Mike
Actually, I'm very happy to improve the code. That's the main thing right now. It hardly matters if it's called 2.3.2 or 3.0 or whatever.
Re. anti-human Rybka, Larry and I need to discuss it. Of course, I know that some people would like us to release it - we will see.
Vas
Re. anti-human Rybka, Larry and I need to discuss it. Of course, I know that some people would like us to release it - we will see.
Vas
Is there an anti-Computer version that is going to be released?
I think all Rybka owners will agree with me that all releases of Rybka past and present are already strong against any other chess software.
This is good. Small incrementally sound steps are probably best at this point. Save all the real good stuff
for 3.0 when you have a chance to get compensated for a new major relase. Am looking forward to 2.3.2.
and 3.0 down the road.
Gary
for 3.0 when you have a chance to get compensated for a new major relase. Am looking forward to 2.3.2.
and 3.0 down the road.
Gary
Yes! An Anti-Human version would be great. Also an even stronger Winfinder would be great. But definitely an anti-human Rybka would be very useful for us grateful humans.
Helo all,so Vas I HATE to bring up dates,but you did say March for 2.3.2 release?Do you have an approximate date or perhaps you would not like to say because the WHOLE WORLD will hold you to that maybe you should just surprise us.I was just wondering,thanks.
Yes, it is almost time for Jessica to weigh in on Vas's schedule for 2.3.2. :-)
Shaddup, Alan.
Ahh, my hero! Congrats on your performance in the freestyle! #1! I will be 100 times more impressed if you tell me you were playing alone on your P4. :-)
Anyway, best of luck in the finals!
Regards,
Alan
Anyway, best of luck in the finals!
Regards,
Alan
Thanks... Check out round #7 vs. the formidable Dufrek team. Really threaded the needle in that one; huge complications.
I think it's better to make this one a surprise. :)
Vas
Vas
Anti-Human and Anti-Computer versions of Rybka would be rather beside the point. After all, there are relatively few occasions for which these versions would even have applicability.
The strongest opponent should always be a Centaur,even at the Super GM level, and Vas and team,imho,should continue to develop Rybka in only one direction...for the purpose of
making her the strongest possible chess playing engine in the world.
No diffusion of effort,no sidetracks.Another way of putting it would be that one day Rybka would only be beatable by another Rybka in more skillful hands and/or more powerful hardware.
As mentioned earlier in this thread,Kaufman is on record saying that anyone divining a method to defeat Rybka would only succeed one time. I assume that Larry means until the next
Rybka iteration which would have plugged the hole which had been exploited.
I know that nothing great is ever achieved in a straight vertical line but let's not add to Vas's difficulties by throwing up relatively secondary wish lists. Except for face-to-face play,
how can we ever know with certainty that the human we think we're playing isn't computer-assisted,or that the computer engine we imagine is playing us on automatic pilot isn't,
in fact,human-assisted?
We can't, which means that Anti-Centaur Rybka development is the paradigm which should guide Vas and team's efforts.
The strongest opponent should always be a Centaur,even at the Super GM level, and Vas and team,imho,should continue to develop Rybka in only one direction...for the purpose of
making her the strongest possible chess playing engine in the world.
No diffusion of effort,no sidetracks.Another way of putting it would be that one day Rybka would only be beatable by another Rybka in more skillful hands and/or more powerful hardware.
As mentioned earlier in this thread,Kaufman is on record saying that anyone divining a method to defeat Rybka would only succeed one time. I assume that Larry means until the next
Rybka iteration which would have plugged the hole which had been exploited.
I know that nothing great is ever achieved in a straight vertical line but let's not add to Vas's difficulties by throwing up relatively secondary wish lists. Except for face-to-face play,
how can we ever know with certainty that the human we think we're playing isn't computer-assisted,or that the computer engine we imagine is playing us on automatic pilot isn't,
in fact,human-assisted?
We can't, which means that Anti-Centaur Rybka development is the paradigm which should guide Vas and team's efforts.
Actually, the term "anti-human" is not really very precise. Maybe it should have been called a "handicap" version. What we mainly wanted to avoid was a situation where Rybka helps Ehlvest trade all the pieces into a winning endgame.
Vas
Vas
Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill