Who is stronger at chess? Computers or Humans?

[reflectionofpower]

Computers, hands down!

[Father]

Does anybody know where Eduard Nemeth is ? Thanks in advance

[mwyoung]

Human Grandmasters are still much better than computers. It is also logical because humans were the creators of computers and a creation can never be better (or smarter) than the creator.

Computers are just tools, nothing more.

And this post proves it...

[Sedat Canbaz]

Bayesian Elo calculation tool is also made by human, but as we see since 2000 years is started to calculate in favor for machines

In those years (10-20 years ago) the processors were Pentium II,Celeron...and the strongest engines were Fritz 5, Hiarcs 6, Junior 5

And since 2006,
There is no any official Man vs Machine, I wonder a lot why ???

Don't be too afraid...we will use again Fritz 10 on Pentium II

And nowadays, I can't imagine...what will be the performance of Stockfish,Houdini,Komodo..on latest fast decent 2x Xeon E5-2697 v2 24 cores .

I guess at least there will be 600 Elo+ over best human player!!

Code: Select all

Rank Name                            Elo    +    -  games score oppo. draws
     1.Machine 2000-2007             2903   19   19   390   74%  2774   34%
     2.Kasparov, Garry               2850   18   18   628   61%  2776   56%
     3.Anand, Viswanathan            2836   10   10  1996   62%  2754   53%
     4.Kramnik, Vladimir             2829   11   11  1602   60%  2762   56%
     5.Ivanchuk, Vassily             2802   10   10  2254   59%  2744   53%
     6.Aronian, Levon                2797   15   15   887   57%  2748   54%
     7.Topalov, Veselin              2795   12   12  1377   55%  2758   47%
     8.Karpov, Anatoly               2792   10   10  2186   60%  2727   53%
     9.Morozevich, Alexander         2788   15   15   987   56%  2741   37%
    10.Grischuk, Alexander           2785   15   15   909   57%  2736   51%
    11.Carlsen, Magnus               2784   16   16   797   56%  2743   49%
    12.Svidler, Peter                2783   12   12  1337   57%  2738   55%
    13.Gelfand, Boris                2779   11   11  1828   55%  2744   56%
    14.Shirov, Alexei                2779   17   17   729   48%  2795   50%
    15.Leko, Peter                   2775   12   12  1286   52%  2760   63%
    16.Ponomariov, Ruslan            2771   16   16   743   54%  2740   53%
    17.Adams, Michael                2769   11   11  1561   56%  2730   53%
    18.Mamedyarov, Shakhriyar        2768   19   19   558   56%  2726   51%
    19.Radjabov, Teimour             2766   16   16   735   53%  2747   59%
    20.Fischer, Robert James         2763   23   23   408   65%  2656   41%
    *  Machine 1977-1999             2588   11   11  1802   51%  2590   23%

[Sedat Canbaz]

One thing more,
We should not forget to mention Rybka Cluster too
And Rybka Cluster 296 cores + superior book probably will pass 1000 Elo+ over any strongest human player!!!

[Uri Blass]

Bayesian Elo calculation tool is also made by human, but as we see since 2000 years is started to calculate in favor for machines

In those years (10-20 years ago) the processors were Pentium II,Celeron...and the strongest engines were Fritz 5, Hiarcs 6, Junior 5

And since 2006,
There is no any official Man vs Machine, I wonder a lot why ???

Don't be too afraid...we will use again Fritz 10 on Pentium II

And nowadays, I can't imagine...what will be the performance of Stockfish,Houdini,Komodo..on latest fast decent 2x Xeon E5-2697 v2 24 cores .

I guess at least there will be 600 Elo+ over best human player!!

I guess not and I guess the advantage of computers may be only 300 elo relative to the best human player.

comp-comp rating lists get bigger difference relative to human-human games for the following reasons:
1)They are played without opening book and humans can choose an opening book that is good to make a draw.
2)The weaker programs have not the knowledge that they are weaker so they do not try to draw when they have a small advantage.

Strong humans are going to play for a draw when they get a small advantage from the opening against a stronger opponent and not take risks.

3)books clearly gives advantage to humans because humans can play some known book lines when the computer need to choose between going to the main theory that the human knows well and can draw or playing slightly inferior move so the human start the game from some small advantage.

[Sedat Canbaz]

Bayesian Elo calculation tool is also made by human, but as we see since 2000 years is started to calculate in favor for machines

In those years (10-20 years ago) the processors were Pentium II,Celeron...and the strongest engines were Fritz 5, Hiarcs 6, Junior 5

And since 2006,
There is no any official Man vs Machine, I wonder a lot why ???

Don't be too afraid...we will use again Fritz 10 on Pentium II

And nowadays, I can't imagine...what will be the performance of Stockfish,Houdini,Komodo..on latest fast decent 2x Xeon E5-2697 v2 24 cores .

I guess at least there will be 600 Elo+ over best human player!!

I guess not and I guess the advantage of computers may be only 300 elo relative to the best human player.

comp-comp rating lists get bigger difference relative to human-human games for the following reasons:
1)They are played without opening book and humans can choose an opening book that is good to make a draw.
2)The weaker programs have not the knowledge that they are weaker so they do not try to draw when they have a small advantage.

Strong humans are going to play for a draw when they get a small advantage from the opening against a stronger opponent and not take risks.

3)books clearly gives advantage to humans because humans can play some known book lines when the computer need to choose between going to the main theory that the human knows well and can draw or playing slightly inferior move so the human start the game from some small advantage.

Dear Uri,

I don't know how much you are experienced in book making,or how many books did you test so far... ?!

But from my experience I can say (probably I tested more than 1000 books),
During my book testings I noticed that there are such private books which are beast, unbeatable, killer books

Grandmasters have 0 (zero) chances to win, maybe in 100 games we can see a draw...of course I am talking about Houdini, Komodo, Stockfish, Rybka Cluster with super strong private books on latest fast machines

Note also that the period between 1970-2000 years:
There were not many super strong engine books, but nowadays you can't imagine how much the opening theory is changed since that period of time
I mean there is a HUGE opening book progress since 1990 years till nowadays (2014)

Btw, the above rating list is based on games which are played only Computer vs GMs
I mean all those 396 engine games are played against GMs

So...if you still don't agree:
1) Then Bayesian Elo program is buggy (this tool does not give us right results)
2)Ed Schröder or Me published wrong results...

Or maybe it's time to change your view about the strength of the latest top chess engines ?!

In my calculations,
I noticed that in around 2005 years, the engines were approx. 2900 Elo stronger than humans (in 2000-2002 years, top engines were almost equal to strongest GMs)

In 2000-2005 years, the engines versions were Fritz 6/7, Chess Tiger 14,Rebel...
In 2000-2005 years, the processors were Celeron,Pentium IV,AMD Athlon

In other words,
1) Maybe you remember that we made some calculations with you a few days ago, so we noticed that between Celeron 1.7GHz and QX9650 there is approx. 400-500 Elo difference

2)Between Fritz 6/Chess Tiger 14,Junior 6 <----> Stockfish 150214, Houdini 4,Komodo,Gull 2.8b,Rybka 4.1 = there is more than 700 Elo difference

To be honest (between Top GMs vs Top Chess Engines),
I expect higher than 600 Elo, the right number probably would be 800 Elo
But I preferred to say 600 Elo just in case of Top X is using not so strong book

If you still don't agree with my statements,
Then I suggest you to participate with your book in my next
SCCT Book Championship I or if you wish you can participate without engine :
http://www.sedatcanbaz.com/chess/?page_id=601




And then there is no doubt that you will see the power/the strength/the progress of the latest opening theory!

Hopes this helps...

Best,
Sedat

[Uri Blass]

Bayesian Elo calculation tool is also made by human, but as we see since 2000 years is started to calculate in favor for machines

In those years (10-20 years ago) the processors were Pentium II,Celeron...and the strongest engines were Fritz 5, Hiarcs 6, Junior 5

And since 2006,
There is no any official Man vs Machine, I wonder a lot why ???

Don't be too afraid...we will use again Fritz 10 on Pentium II

And nowadays, I can't imagine...what will be the performance of Stockfish,Houdini,Komodo..on latest fast decent 2x Xeon E5-2697 v2 24 cores .

I guess at least there will be 600 Elo+ over best human player!!

I guess not and I guess the advantage of computers may be only 300 elo relative to the best human player.

comp-comp rating lists get bigger difference relative to human-human games for the following reasons:
1)They are played without opening book and humans can choose an opening book that is good to make a draw.
2)The weaker programs have not the knowledge that they are weaker so they do not try to draw when they have a small advantage.

Strong humans are going to play for a draw when they get a small advantage from the opening against a stronger opponent and not take risks.

3)books clearly gives advantage to humans because humans can play some known book lines when the computer need to choose between going to the main theory that the human knows well and can draw or playing slightly inferior move so the human start the game from some small advantage.

Dear Uri,

I don't know how much you are experienced in book making,or how many books did you test so far... ?!

But from my experience I can say (probably I tested more than 1000 books),
During my book testings I noticed that there are such private books which are beast, unbeatable, killer books

Grandmasters have 0 (zero) chances to win, maybe in 100 games we can see a draw...of course I am talking about Houdini, Komodo, Stockfish, Rybka Cluster with super strong private books on latest fast machines

Note also that the period between 1970-2000 years:
There were not many super strong engine books, but nowadays you can't imagine how much the opening theory is changed since that period of time
I mean there is a HUGE opening book progress since 1990 years till nowadays (2014)

Btw, the above rating list is based on games which are played only Computer vs GMs
I mean all those 396 engine games are played against GMs

So...if you still don't agree:
1) Then Bayesian Elo program is buggy (this tool does not give us right results)
2)Ed Schröder or Me published wrong results...

Or maybe it's time to change your view about the strength of the latest top chess engines ?!

In my calculations,
I noticed that in around 2005 years, the engines were approx. 2900 Elo stronger than humans (in 2000-2002 years, top engines were almost equal to strongest GMs)

In 2000-2005 years, the engines versions were Fritz 6/7, Chess Tiger 14,Rebel...
In 2000-2005 years, the processors were Celeron,Pentium IV,AMD Athlon

In other words,
1) Maybe you remember that we made some calculations with you a few days ago, so we noticed that between Celeron 1.7GHz and QX9650 there is approx. 400-500 Elo difference

2)Between Fritz 6/Chess Tiger 14,Junior 6 <----> Stockfish 150214, Houdini 4,Komodo,Gull 2.8b,Rybka 4.1 = there is more than 700 Elo difference

To be honest (between Top GMs vs Top Chess Engines),
I expect higher than 600 Elo, the right number probably would be 800 Elo
But I preferred to say 600 Elo just in case of Top X is using not so strong book

If you still don't agree with my statements,
Then I suggest you to participate with your book in my next
SCCT Book Championship I or if you wish you can participate without engine :
http://www.sedatcanbaz.com/chess/?page_id=601




And then there is no doubt that you will see the power/the strength/the progress of the latest opening theory!

Hopes this helps...

Best,
Sedat

Dear Sedat,
My point is that I expect humans to play for a draw and strong humans know lines that give them good draw chances with white(they may not be good for your book because 50% with white is bad result but humans do not have problem to allow chess engine to force a draw with black if it gets them closer to get the desired draw).

comp-comp tests are not relevant because we talk about rating against humans and not about rating against chess programs.

I expect Magnus Carlsen to be able to draw half of the games with white in case that he plays for a draw against the best programs with the best book that you have.

I do not say that I can do it but I am not Magnus Carlsen so it proves nothing.

[Sedat Canbaz]

I have facts, games, rating list calculated by Bayesian Elo...if you still disagree:
It is not my problem...

In my opinion (in the future),
No any Top Grandmaster (including Magnus Carlsen too) will accept Man vs Machine match

Because Top GMs know very well the power of the latest top chess engines!

Maybe you remember Man Machine match in 2006
Where Fritz 10 won against Kramnik 4:2

If you say only a few games....then you need to check again my publish rating list
And if still you will need more data then I can inform you with many several tournaments (mainly played around 2000-2007 years) where Machines are beyond GMs!

Note that your engine (Movei 1 core) is near strenght to many GMs

Note also that many chess engines 1700-1800 Elo have possibility to make draws against Top GMs too, but I guess in every 50 - 100 games
But of course, engines of 1700-1800 Elo will loose mostly of the games

As you know,
Instead of comments and guessing I prefer data...

Actually it is not so hard...just a little bit mathematica is needed
See please the performance of Fritz 6, Chess Tiger 14,Hiarcs 7.32, Movei
All these mentioned engines are near strenght to GMs
And then compare those mentioned engines with Houdini,Stockfish...
And then check again please the Elo diffeernce between 2000 year hardwares and latest decent processors...

And in the end, the formula (calculation) shows min. 800 Elo difference between Top GMs and Top Engines

In other words,
We have already enough data,facts,games,tournaments,rating lists...
What we need more ???
And it is not needed to be proven any more ... we should agree that !


Best,
Sedat

[asanjuan]

I would say Carlsen, but I can't find the option!