Hi All.
I was wondering if people are tired of the match, and think it should end at 100 games. Houdini has been extending its lead lately and there does not seem any reason to expect anything different in another 100 games.
Maybe some change in conditions are advised before continuing. I have read that there are some time-management and\or other settings that are supposed to make Rybka play stronger. Is this even true, or worth trying? Maybe using ponder = on for both.
I was thinking that if conditions were changed, the same 50 positions could be played again to see the effect, if any, on results. Maybe you think it better to change settings but not reuse positions (but then we can not really be sure what effect they had).
Of course, maybe you are tired of the whole thing already and think it should just go away!
PawnStormZ wrote: Hi All.
I was wondering if people are tired of the match, and think it should end at 100 games. Houdini has been extending its lead lately and there does not seem any reason to expect anything different in another 100 games.
Maybe some change in conditions are advised before continuing. I have read that there are some time-management and\or other settings that are supposed to make Rybka play stronger. Is this even true, or worth trying? Maybe using ponder = on for both.
I was thinking that if conditions were changed, the same 50 positions could be played again to see the effect, if any, on results. Maybe you think it better to change settings but not reuse positions (but then we can not really be sure what effect they had).
Of course, maybe you are tired of the whole thing already and think it should just go away!
Pal, do not change anything, and continue as much as you can. 100 games are still few for error margins, they are enough only for LOS. If you change conditions, another 100 games would be a waste, as no one can decide what was a statistical fluke or a feature. To me, the most interesting part are the games themselves (half of the moves are simply uninteligible to me), the ratio win/loss at LTC on very strong hardware and the draw ratio. The problem is not that Houdini anyway wins, HOW it wins is the most important.
Continue to 200 games in identical conditions. I understand that it requires patience and passion, but you have to realize that your test would be one of the most reliable at LTC at this level, if not unique.
PawnStormZ wrote: Hi All.
I was wondering if people are tired of the match, and think it should end at 100 games. Houdini has been extending its lead lately and there does not seem any reason to expect anything different in another 100 games.
Maybe some change in conditions are advised before continuing. I have read that there are some time-management and\or other settings that are supposed to make Rybka play stronger. Is this even true, or worth trying? Maybe using ponder = on for both.
I was thinking that if conditions were changed, the same 50 positions could be played again to see the effect, if any, on results. Maybe you think it better to change settings but not reuse positions (but then we can not really be sure what effect they had).
Of course, maybe you are tired of the whole thing already and think it should just go away!
Pal, do not change anything, and continue as much as you can. 100 games are still few for error margins, they are enough only for LOS. If you change conditions, another 100 games would be a waste, as no one can decide what was a statistical fluke or a feature. To me, the most interesting part are the games themselves (half of the moves are simply uninteligible to me), the ratio win/loss at LTC on very strong hardware and the draw ratio. The problem is not that Houdini anyway wins, HOW it wins is the most important.
Continue to 200 games in identical conditions. I understand that it requires patience and passion, but you have to realize that your test would be one of the most reliable at LTC at this level, if not unique.
Kai
(+1) I echo this.
IOW, "Run the tournament as long as you can possibly stand it."
PawnStormZ wrote: Hi All.
I was wondering if people are tired of the match, and think it should end at 100 games. Houdini has been extending its lead lately and there does not seem any reason to expect anything different in another 100 games.
Maybe some change in conditions are advised before continuing. I have read that there are some time-management and\or other settings that are supposed to make Rybka play stronger. Is this even true, or worth trying? Maybe using ponder = on for both.
I was thinking that if conditions were changed, the same 50 positions could be played again to see the effect, if any, on results. Maybe you think it better to change settings but not reuse positions (but then we can not really be sure what effect they had).
Of course, maybe you are tired of the whole thing already and think it should just go away!
Pal, do not change anything, and continue as much as you can. 100 games are still few for error margins, they are enough only for LOS. If you change conditions, another 100 games would be a waste, as no one can decide what was a statistical fluke or a feature. To me, the most interesting part are the games themselves (half of the moves are simply uninteligible to me), the ratio win/loss at LTC on very strong hardware and the draw ratio. The problem is not that Houdini anyway wins, HOW it wins is the most important.
Continue to 200 games in identical conditions. I understand that it requires patience and passion, but you have to realize that your test would be one of the most reliable at LTC at this level, if not unique.
Kai
(+1) I echo this.
IOW, "Run the tournament as long as you can possibly stand it."
Yes i would love to see another 100 games under the same conditions.
Thanks for your time and testing.Much appreciated.
I can't bear to watch any more, however. Let me throw in the towel for Rybka 4-sigma superiority is enough. I'd be interested in seeing more tournaments, but from endgame positions and with a faster time control like 40/20. My 2 cents.
Sorry,I'm not interested in your match....
I have my own reasons regards,
Dr.D
_No one can hit as hard as life.But it ain’t about how hard you can hit.It’s about how hard you can get hit and keep moving forward.How much you can take and keep moving forward….
PawnStormZ wrote: Hi All.
I was wondering if people are tired of the match, and think it should end at 100 games. Houdini has been extending its lead lately and there does not seem any reason to expect anything different in another 100 games.
Maybe some change in conditions are advised before continuing. I have read that there are some time-management and\or other settings that are supposed to make Rybka play stronger. Is this even true, or worth trying? Maybe using ponder = on for both.
I was thinking that if conditions were changed, the same 50 positions could be played again to see the effect, if any, on results. Maybe you think it better to change settings but not reuse positions (but then we can not really be sure what effect they had).
Of course, maybe you are tired of the whole thing already and think it should just go away!
Pal, do not change anything, and continue as much as you can. 100 games are still few for error margins, they are enough only for LOS. If you change conditions, another 100 games would be a waste, as no one can decide what was a statistical fluke or a feature. To me, the most interesting part are the games themselves (half of the moves are simply uninteligible to me), the ratio win/loss at LTC on very strong hardware and the draw ratio. The problem is not that Houdini anyway wins, HOW it wins is the most important.
Continue to 200 games in identical conditions. I understand that it requires patience and passion, but you have to realize that your test would be one of the most reliable at LTC at this level, if not unique.
Kai
(+1) I echo this.
IOW, "Run the tournament as long as you can possibly stand it."
Playing two engines against each other is not really a reliable method to determine which one is better. At best you can say eventually: "It is almost certain that A beats B more often than B beats A", where "almost certain" can be driven up from 99% to 99.9% to 99.99999%, depending on your ptience. But even a 100% certainty that A beats B by a certain amount does not exclude that A can be weaker than B (i.e. has lower Elo).
To determine the Elo reliably you must play against a variety of opponents, distributed more or less homogeneously over an Elo range of +/- 300 around the engines.
So it would be much more meaningful to switch to other opponents after 100 games. Pit them both 100 games against an opponent 150 Elo weaker, and then each 100 games against an opponent 150 Elo stronger.
hgm wrote:Playing two engines against each other is not really a reliable method to determine which one is better. At best you can say eventually: "It is almost certain that A beats B more often than B beats A", where "almost certain" can be driven up from 99% to 99.9% to 99.99999%, depending on your ptience. But even a 100% certainty that A beats B by a certain amount does not exclude that A can be weaker than B (i.e. has lower Elo).
To determine the Elo reliably you must play against a variety of opponents, distributed more or less homogeneously over an Elo range of +/- 300 around the engines.
So it would be much more meaningful to switch to other opponents after 100 games. Pit them both 100 games against an opponent 150 Elo weaker, and then each 100 games against an opponent 150 Elo stronger.
One of my reasons for my above answer....yes....
Dr.D
_No one can hit as hard as life.But it ain’t about how hard you can hit.It’s about how hard you can get hit and keep moving forward.How much you can take and keep moving forward….
