Leela, such a great engine (and innovation)?

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Harvey Williamson, bob

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
supersharp77
Posts: 632
Joined: Sat Jul 05, 2014 5:54 am
Location: Southwest USA

Re:CM 6000 CM 5000 Leela, such a great engine

Post by supersharp77 » Mon Jun 10, 2019 7:19 pm

mclane wrote:
Mon Jun 10, 2019 6:54 am
Millennium chess exclusive comes with the king 2.61 and is a dedicated chess computer on arm cpu 300 MHz. Elo Arround 2400+
My All time favorite Program is CM 6000 & CM 5000 did anyone every release the OPK (eng)v 2.61 from the chessmaster GUI so it could be played as a winboard engine?....Would love to have that one from CM 6000 for modern testing....used to test it manually back in the Old days against Junior 7 and Lokasoft (Chesspartner5) all done manually....quite a bit of work... :D :wink:

User avatar
Ovyron
Posts: 2044
Joined: Tue Jul 03, 2007 2:30 am

Re: Leela, such a great engine (and innovation)?

Post by Ovyron » Wed Jun 12, 2019 11:05 am

Dann Corbit wrote:
Fri Jun 07, 2019 7:52 am
Proof of best never comes from a championship. It takes many thousands of games to know which engine is strongest when they are close in strength. And by the time we know the answer, there is a new version anyway. A contest like tcec gives us a champion. And contests like ccrl and cegt tell us what's stronger on hardware mere mortals can afford. Both systems have great value.
Does anybody doubt that Magnus Carlsen is the best chess human player in the world? Did he need to play many thousands of games to prove it? Was he forced to play generic openings like the ones from TCEC or CCRL/CEGT?

Chess is chess and strength is strength, so I don't get why to prove engines' strength and human's strength radically different paths have been taken, and it's clear the most efficient one has been taken by humans, and TCEC and the rating lists are wasting incredible amounts of resources and time (while mimicking what humans do to decide who are the best humans could have been done with a fragment of those resources.)
Make someone happy today.

Dann Corbit
Posts: 9661
Joined: Wed Mar 08, 2006 7:57 pm
Location: Redmond, WA USA
Contact:

Re: Leela, such a great engine (and innovation)?

Post by Dann Corbit » Wed Jun 12, 2019 4:29 pm

The key word here is prove.
I don't know if there is someone within 20 Elo, but if there is,we can't prove who is smarter on the basis of the games played.
That does not means that we don't have a pretty good guess that is likely to be right.
If someone is 100 Elo better you can drive the LOS very, very close to 100 with a reasonable number of games.
If someone is within 5 Elo, you have a big job on your hands to prove it.
If someone is within one Elo, good luck proving which one is stronger.

Our minds sometimes do a very good job of pattern matching or interpolation.
And sometimes they don't.

And finally, there is a huge difference between "I am convinced." And , "It has been proven."
A good example of that is "Fermat's last theorem." And an ongoing one is Gauss and the zeros of the zeta function having a real!part of one half.
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.

Dann Corbit
Posts: 9661
Joined: Wed Mar 08, 2006 7:57 pm
Location: Redmond, WA USA
Contact:

Re: Leela, such a great engine (and innovation)?

Post by Dann Corbit » Wed Jun 12, 2019 4:35 pm

Examples where most people do not extrapolate correctly are the Mony Hall "Let's Make a Deal" problem and the birthday paradox.
Hilbert's Hotel is another good one.
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.

User avatar
Ovyron
Posts: 2044
Joined: Tue Jul 03, 2007 2:30 am

Re: Leela, such a great engine (and innovation)?

Post by Ovyron » Wed Jun 12, 2019 8:56 pm

Dann Corbit wrote:
Wed Jun 12, 2019 4:29 pm
And finally, there is a huge difference between "I am convinced." And , "It has been proven."
What's the difference? Really, I keep saying "Magnus Carlsen", "Magnus Carlsen" like a parrot, but it hasn't been proven to any degree that he's the strongest chess human on earth, yet "people being convinced" that he is is enough in the chess world.

Unless you can name someone else that is stronger than him (and human), what he has done should suffice, and people could do the same with chess engines to get convinced that Leela or Stockfish is the strongest chess entity. And with a very small fraction of the resources that is being used by TCEC, the CCRL and the CEGT for these purposes.

What's this obsession with "proving" something and why are only computer chess enthusiasts affected and common chess enthusiasts not? (I've never heard someone say it hasn't been "proven" that Magnus is the best human chess player.) Is it because people can play hundreds of thousands of games with chess machines but they can't do it with humans? If so, being unable to do it with humans seems like an advantage, as we can find who is best with much less time and effort (even though we can't "prove" it.)
Make someone happy today.

Dann Corbit
Posts: 9661
Joined: Wed Mar 08, 2006 7:57 pm
Location: Redmond, WA USA
Contact:

Re: Leela, such a great engine (and innovation)?

Post by Dann Corbit » Wed Jun 12, 2019 10:46 pm

Ovyron wrote:
Wed Jun 12, 2019 8:56 pm
Dann Corbit wrote:
Wed Jun 12, 2019 4:29 pm
And finally, there is a huge difference between "I am convinced." And , "It has been proven."
What's the difference? Really, I keep saying "Magnus Carlsen", "Magnus Carlsen" like a parrot, but it hasn't been proven to any degree that he's the strongest chess human on earth, yet "people being convinced" that he is is enough in the chess world.

Unless you can name someone else that is stronger than him (and human), what he has done should suffice, and people could do the same with chess engines to get convinced that Leela or Stockfish is the strongest chess entity. And with a very small fraction of the resources that is being used by TCEC, the CCRL and the CEGT for these purposes.

What's this obsession with "proving" something and why are only computer chess enthusiasts affected and common chess enthusiasts not? (I've never heard someone say it hasn't been "proven" that Magnus is the best human chess player.) Is it because people can play hundreds of thousands of games with chess machines but they can't do it with humans? If so, being unable to do it with humans seems like an advantage, as we can find who is best with much less time and effort (even though we can't "prove" it.)
Magnus Carlsen might be the strongest player in the world.
Magnus Carlsen is the strongest player in the world.
This is the difference between being convinced and being proven.

If you have 20 people in a room, 50 times out of one hundred two people will have the same birthday.
If you have 70 randomly chosen people in a room, 999 times out of one thousand two of them will have the same birthday.
In both cases, we "probably" have a match. In neither case is it decided unless we actually test the conditions for a given instance.
So what is your acceptable level of proof? Is a coin toss good enough? It might be right. It might be wrong.
If you examine a collection of trials mathematically, you can say what the certainty is that something you imagine is correct is really true.
Is it unlikely? Is it fairly likely? Is it extremely likely? Is it nearly certain? Is it certain?
Some people are easily convinced. Some people are hard to convince. Is there a different truth for these categories of people?

Now, let's suppose further that we have enough games to show that within two full standard deviations Carlsen is better than Caruana. Does that mean that if they meet head to head in a ten game match it is certain that Carlsen is going to win?

Consider this list of Fide Champions:
Anatoly Karpov (1993-1999)
Alexander Khalifman (1999-2000)
Viswanathan Anand (2000-2002)
Ruslan Ponomariov (2002-2004)
Rustam Kasimdzhanov (2004-2005)
Veselin Topalov (2005-2006)

How many of those were the best players in the world at the time they were champions?
That is an important illustration between what a champion is and who the best player is.
They are most certainly not the same thing.
Since the world champion of chess is not certainly the strongest player in the world, we should also conclude the same things about short contests involving engines that we conclude about short contests involving people.

TCEC does not name the strongest chess engine in the world. TCEC names a champion.
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.

supersharp77
Posts: 632
Joined: Sat Jul 05, 2014 5:54 am
Location: Southwest USA

Re: Leela, such a great engine (and innovation)?

Post by supersharp77 » Thu Jun 13, 2019 10:11 pm

Dann Corbit wrote:
Wed Jun 12, 2019 10:46 pm
Ovyron wrote:
Wed Jun 12, 2019 8:56 pm
Dann Corbit wrote:
Wed Jun 12, 2019 4:29 pm
And finally, there is a huge difference between "I am convinced." And , "It has been proven."
What's the difference? Really, I keep saying "Magnus Carlsen", "Magnus Carlsen" like a parrot, but it hasn't been proven to any degree that he's the strongest chess human on earth, yet "people being convinced" that he is is enough in the chess world.

Unless you can name someone else that is stronger than him (and human), what he has done should suffice, and people could do the same with chess engines to get convinced that Leela or Stockfish is the strongest chess entity. And with a very small fraction of the resources that is being used by TCEC, the CCRL and the CEGT for these purposes.

What's this obsession with "proving" something and why are only computer chess enthusiasts affected and common chess enthusiasts not? (I've never heard someone say it hasn't been "proven" that Magnus is the best human chess player.) Is it because people can play hundreds of thousands of games with chess machines but they can't do it with humans? If so, being unable to do it with humans seems like an advantage, as we can find who is best with much less time and effort (even though we can't "prove" it.)
Magnus Carlsen might be the strongest player in the world.
Magnus Carlsen is the strongest player in the world.
This is the difference between being convinced and being proven.

If you have 20 people in a room, 50 times out of one hundred two people will have the same birthday.
If you have 70 randomly chosen people in a room, 999 times out of one thousand two of them will have the same birthday.
In both cases, we "probably" have a match. In neither case is it decided unless we actually test the conditions for a given instance.
So what is your acceptable level of proof? Is a coin toss good enough? It might be right. It might be wrong.
If you examine a collection of trials mathematically, you can say what the certainty is that something you imagine is correct is really true.
Is it unlikely? Is it fairly likely? Is it extremely likely? Is it nearly certain? Is it certain?
Some people are easily convinced. Some people are hard to convince. Is there a different truth for these categories of people?

Now, let's suppose further that we have enough games to show that within two full standard deviations Carlsen is better than Caruana. Does that mean that if they meet head to head in a ten game match it is certain that Carlsen is going to win?

Consider this list of Fide Champions:
Anatoly Karpov (1993-1999)
Alexander Khalifman (1999-2000)
Viswanathan Anand (2000-2002)
Ruslan Ponomariov (2002-2004)
Rustam Kasimdzhanov (2004-2005)
Veselin Topalov (2005-2006)

How many of those were the best players in the world at the time they were champions?
That is an important illustration between what a champion is and who the best player is.
They are most certainly not the same thing.
Since the world champion of chess is not certainly the strongest player in the world, we should also conclude the same things about short contests involving engines that we conclude about short contests involving people.

TCEC does not name the strongest chess engine in the world. TCEC names a champion.
Although Not a fan Of Him or his philosophy or Playing style Magnus Carlsen without a doubt the Best Player in the world today..Caruana is approaching but not quite there yet as the WC Match showed (Match critical moments & tiebreaks)
Ditto for Karpov and Anand..their reign is unquestioned...A big fan of Khalifman..he won in match play KO's FIDE..not an easy task..Ponomariov reigned tactically supreme at his time along with Topalov for a time...Rustam K. not too sure about him..I believe he won a FIDE KO match play event...He would be the "outlier" on my list..add.. Kasparov, Fischer,V Kramnik
Tal & Petrosian and thats one List of Extremely strong chess players ! Even a Chess engine "past its prime" still has amazing moments...and can surprise even the very best...Like Stockfish 9 or SugaR 5.4a...or Houdini 3 CK :) :wink:

User avatar
Ovyron
Posts: 2044
Joined: Tue Jul 03, 2007 2:30 am

Re: Leela, such a great engine (and innovation)?

Post by Ovyron » Mon Jun 17, 2019 11:23 pm

Dann Corbit wrote:
Wed Jun 12, 2019 10:46 pm
How many of those were the best players in the world at the time they were champions?
That is an important illustration between what a champion is and who the best player is.
Are you saying we never knew who was the best player in the world in any time period because it was never proven scientifically?

I think it was clear who was the best in several time periods (even if, as you said, it didn't coincide with who was champion), and we didn't need hundreds of thousands of games to know.

You can only say "Y was chess champion, but X was best player", if you know X was best player, and you know that without all the requirements to know what chess engine is best.
Make someone happy today.

Dann Corbit
Posts: 9661
Joined: Wed Mar 08, 2006 7:57 pm
Location: Redmond, WA USA
Contact:

Re: Leela, such a great engine (and innovation)?

Post by Dann Corbit » Mon Jun 17, 2019 11:47 pm

Ovyron wrote:
Mon Jun 17, 2019 11:23 pm
Dann Corbit wrote:
Wed Jun 12, 2019 10:46 pm
How many of those were the best players in the world at the time they were champions?
That is an important illustration between what a champion is and who the best player is.
Are you saying we never knew who was the best player in the world in any time period because it was never proven scientifically?
No.
I am saying that a contest like Fide's world championship often does not have the strongest player come out on top.
In a similar way, contests like TCEC (or any other championship with a limited game count) do not always have the strongest entrant win either.
Which engine is strongest is much more reliably described in long running calibration tournaments like CCRL and CEGT.
I think it was clear who was the best in several time periods (even if, as you said, it didn't coincide with who was champion), and we didn't need hundreds of thousands of games to know.
I have never said one hundred thousand games are necessary. If there is decent separation in the real strength a couple thousand games is plenty.

You can only say "Y was chess champion, but X was best player", if you know X was best player, and you know that without all the requirements to know what chess engine is best.
The number of games played in carefully controlled circumstances is like having more and more tiny, accurate marks on the ruler.
With a short championship contest of any sort, it is like using ropes with knots tied about every 18 inches.
If you have 2000 games, then you have a nice Stanley tape measure with tick marks every 64th of an inch.

We think that our human guesses are accurate. And in general, they are pretty good.
But sometimes, it is like what Inigo said to Vincini, "That word you keep using. I don't think it means what you think it means."
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.

User avatar
Ovyron
Posts: 2044
Joined: Tue Jul 03, 2007 2:30 am

Re: Leela, such a great engine (and innovation)?

Post by Ovyron » Tue Jun 18, 2019 12:02 am

Dann Corbit wrote:
Mon Jun 17, 2019 11:47 pm
I think it was clear who was the best in several time periods (even if, as you said, it didn't coincide with who was champion), and we didn't need hundreds of thousands of games to know.
I have never said one hundred thousand games are necessary. If there is decent separation in the real strength a couple thousand games is plenty.
So, are you saying that the only times we were able to know who was the best human player were those where the separation between them and second best was big enough that the games that they played in that time period were enough to know they were best?
Make someone happy today.

Post Reply