Could it be said .....

[JAMerolle]

Programs today are underated , I noticed on playchess there are Rybkas playing with a 2300/2400 rating , its hard to beleave that any human with a 2550 rating can hold there own against these monsters. My estamate is a 300 point underating if we are looking at humans vs the engines on playchess.
I will go as far as saying anyone of these programs playing on a dual ) core (2400 rating ) can take down any human in the top 20.

Any feed back would be great


Regards

Joseph

[Harvey Williamson]

Programs today are underated , I noticed on playchess there are Rybkas playing with a 2300/2400 rating , its hard to beleave that any human with a 2550 rating can hold there own against these monsters. My estamate is a 300 point underating if we are looking at humans vs the engines on playchess.
I will go as far as saying anyone of these programs playing on a dual ) core (2400 rating ) can take down any human in the top 20.

Any feed back would be great


Regards

Joseph

These ratings are against other engines and hardware and books. I play with Hiarcs in the main room on Playchess. It has a rating well over 3000 against humans. I rarely play with Hiarcs at full strength in the main room. I set the book very wide and use high contempt.

In fact at blitz 3390 and bullet 3411. I suspect all top engines would get a very similar rating as i do not remember the last time Hiarcs did not win a game.

Yesterday Hiarcs played 4 games v GM on Playchess the score was +4

[Albert Silver]

Programs today are underated , I noticed on playchess there are Rybkas playing with a 2300/2400 rating , its hard to beleave that any human with a 2550 rating can hold there own against these monsters. My estamate is a 300 point underating if we are looking at humans vs the engines on playchess.
I will go as far as saying anyone of these programs playing on a dual ) core (2400 rating ) can take down any human in the top 20.

Any feed back would be great


Regards

Joseph

The ratings in the engine room are not related to the human ratings. In theory they are, because a human can play against the engine and their ratings will mix, but as you may notice, there isn't a long list of strong humans playing them.

[bob]

Programs today are underated , I noticed on playchess there are Rybkas playing with a 2300/2400 rating , its hard to beleave that any human with a 2550 rating can hold there own against these monsters. My estamate is a 300 point underating if we are looking at humans vs the engines on playchess.
I will go as far as saying anyone of these programs playing on a dual ) core (2400 rating ) can take down any human in the top 20.

Any feed back would be great


Regards

Joseph

No, it can be said that Elo is dependent on the pool of players used to produce the rating, and the performance of those programs against each other. You can't take any chess server rating and compare it to a FIDE or UCCF or BCF rating. If chess ratings were based on one giant rating pool, where there is lots of inter-play between the members, then this idea would work. If you take a pool of two players, one strong program and one extremely weak program, what would their ratings be? If the strong program wins roughly 15 of every 16 games, with either one loss or two draws, it would be rated at roughly N+400 where N is the rating of the weak program. But where does N come from? You could start both at 1200 and one would end up at 1000 and the other at 1400, or you could start them both at 2600 and one would end up at 2400 and the other at 2800.

You can't compare ratings between different pools of players.

[LaurenceChen]

Programs today are underated , I noticed on playchess there are Rybkas playing with a 2300/2400 rating , its hard to beleave that any human with a 2550 rating can hold there own against these monsters. My estamate is a 300 point underating if we are looking at humans vs the engines on playchess.
I will go as far as saying anyone of these programs playing on a dual ) core (2400 rating ) can take down any human in the top 20.

Any feed back would be great


Regards

Joseph

No, it can be said that Elo is dependent on the pool of players used to produce the rating, and the performance of those programs against each other. You can't take any chess server rating and compare it to a FIDE or UCCF or BCF rating. If chess ratings were based on one giant rating pool, where there is lots of inter-play between the members, then this idea would work. If you take a pool of two players, one strong program and one extremely weak program, what would their ratings be? If the strong program wins roughly 15 of every 16 games, with either one loss or two draws, it would be rated at roughly N+400 where N is the rating of the weak program. But where does N come from? You could start both at 1200 and one would end up at 1000 and the other at 1400, or you could start them both at 2600 and one would end up at 2400 and the other at 2800.

You can't compare ratings between different pools of players.

You are correct in saying that one cannot compare the ratings from 2 different pools. However, your rating analogy is wrong. If you get a rating calculator and punch some numbers you will see that playing against a week player cannot or will not inflate much a player's rating. The reason why FIDE ELO rating is inflated is the so called "club" 2700 GMs only play against themselves. This will obvious inflate the ELO rating pool. You can verify this with an ELO calculator.

[Albert Silver]

Programs today are underated , I noticed on playchess there are Rybkas playing with a 2300/2400 rating , its hard to beleave that any human with a 2550 rating can hold there own against these monsters. My estamate is a 300 point underating if we are looking at humans vs the engines on playchess.
I will go as far as saying anyone of these programs playing on a dual ) core (2400 rating ) can take down any human in the top 20.

Any feed back would be great


Regards

Joseph

No, it can be said that Elo is dependent on the pool of players used to produce the rating, and the performance of those programs against each other. You can't take any chess server rating and compare it to a FIDE or UCCF or BCF rating. If chess ratings were based on one giant rating pool, where there is lots of inter-play between the members, then this idea would work. If you take a pool of two players, one strong program and one extremely weak program, what would their ratings be? If the strong program wins roughly 15 of every 16 games, with either one loss or two draws, it would be rated at roughly N+400 where N is the rating of the weak program. But where does N come from? You could start both at 1200 and one would end up at 1000 and the other at 1400, or you could start them both at 2600 and one would end up at 2400 and the other at 2800.

You can't compare ratings between different pools of players.

You are correct in saying that one cannot compare the ratings from 2 different pools. However, your rating analogy is wrong. If you get a rating calculator and punch some numbers you will see that playing against a week player cannot or will not inflate much a player's rating. The reason why FIDE ELO rating is inflated is the so called "club" 2700 GMs only play against themselves. This will obvious inflate the ELO rating pool. You can verify this with an ELO calculator.

AFAIK you don't get extra points for playing someone rated the same or close to it, so why would a 2700 player playing another get more points than a 2400 player playing against another 2400 player?

[LaurenceChen]

Just do the calculations and you will see why...
The 2400 rated players would start to gain points and this would increase the rating of the "stronger" players. Of course, you must remember that chess is dynamic among humans, either will continue to improve their levels of play or decrease...
The same would apply to the 2400 Elo pool, eventually this pool would catch up with the 2700 Elo pool...

[Frank Quisinsky]

Hi,

end of the last year GM Jörg Hickl and myself made an interview with GM Georg Meier. Georg Meier have the opinion that Rybka 3 with 2 Cores, 32-Bit is playing with around 2.900 ELO (1 minute per move).

Georg gave the information, that different other Grandmasters have the same opinion.

GM Georg Meier is number 2 in Germany and playing so far with around 2.650 ELO.

Perhaps the information is interesting for yourself.

Best
Frank

[bob]

Programs today are underated , I noticed on playchess there are Rybkas playing with a 2300/2400 rating , its hard to beleave that any human with a 2550 rating can hold there own against these monsters. My estamate is a 300 point underating if we are looking at humans vs the engines on playchess.
I will go as far as saying anyone of these programs playing on a dual ) core (2400 rating ) can take down any human in the top 20.

Any feed back would be great


Regards

Joseph

No, it can be said that Elo is dependent on the pool of players used to produce the rating, and the performance of those programs against each other. You can't take any chess server rating and compare it to a FIDE or UCCF or BCF rating. If chess ratings were based on one giant rating pool, where there is lots of inter-play between the members, then this idea would work. If you take a pool of two players, one strong program and one extremely weak program, what would their ratings be? If the strong program wins roughly 15 of every 16 games, with either one loss or two draws, it would be rated at roughly N+400 where N is the rating of the weak program. But where does N come from? You could start both at 1200 and one would end up at 1000 and the other at 1400, or you could start them both at 2600 and one would end up at 2400 and the other at 2800.

You can't compare ratings between different pools of players.

You are correct in saying that one cannot compare the ratings from 2 different pools. However, your rating analogy is wrong. If you get a rating calculator and punch some numbers you will see that playing against a week player cannot or will not inflate much a player's rating. The reason why FIDE ELO rating is inflated is the so called "club" 2700 GMs only play against themselves. This will obvious inflate the ELO rating pool. You can verify this with an ELO calculator.

My point was not what you think. The only possible "connector" between older and more recent GMs will be young players. The young players have a chance to bridge the gap between an old and new GM to give us "some idea" of how their strengths compare. Of course the young "connector" will likely get stronger over time as well, so it is really not an easy/doable task with any significant level of accuracy.