Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Discussion / empirical estimate of God's elo rating
1 2 3 Previous Next  
- - By 8lrr8 (***) Date 2007-12-30 10:39
the difference btwn an undefeatable opponent and God is simply that God converts a handful of games that end up as draws, into wins.  so suppose we have a program that never loses.  suppose after 100 games against an opponent (or a batch of opponents) is: +60=40-0.  against the same opponent(s), God may score something like +70=30-0.

rybka 2.3.2a (64-bit), on a quadcore has an elo rating of 3106 (http://www.husvankempen.de/nunn/40_120_ratinglist/Quad/1.html) on "classical time controls."  i use parentheses because we dont know if this is 120/40 w/ no time extension, or if it's truly 40 moves/2hrs, 20 moves/1hr, + 30 min.  anyway, this translates to an expected value of 0.731/game, against an avg. opponent strength of 2933.  wiping out the 20 loses rybka suffered and converting them into draws, such a program would have an EV of 0.76/game (i.e. +182=168-0).  against a 2933 elo opponent (or in this case, opponents) yields an elo rating of ~3130 (http://www.littlethorpechess.co.uk/elo_calculator.htm).

let's take this another step further.  suppose we have an even stronger program, a program that is able to convert 25% of the drawn games (does anyone really think this is *too* low??) and turn them into wins.  thus, it scores +224=126-0, for an EV of 0.82/game (against 2933 opposition).  such a performance is equivalent to 3200.

moral of the story?  there is no way in hell God can achieve +4000 elo (at least not at classical time controls).  no matter what vasik thinks.  :-)  the reason is because the draw barrier is too large for 4000 elo to exist, plus the nature of how games are scored and subsequently translated into the elo rating system.  i will even go another step further and say even if God was white for every game, he still wouldnt achieve +4000 elo against, say, rybka 4.0 on an vaporchill 16-core (assuming efficient multiprocessor scaling).
Parent - By 8lrr8 (***) Date 2007-12-30 10:46
the above assumes a kick-ass opening book (e.g. jereon, dagh, nelson) is used, of course.
Parent - By BB (****) [gb] Date 2007-12-30 10:51
the difference btwn an undefeatable opponent and God is simply that God converts a handful of games that end up as draws, into wins.

The "handful" would be quite large in my estimation. Being all-knowing (of how the opponent chooses moves, in particular) allows the game to be directed down a path where the program will err. The only exception would be if the program never erred in turning a draw into a loss.
Parent - - By Vasik Rajlich (Silver) [hu] Date 2007-12-30 12:20
I doubt that Rybka 2.3.2a on a quad could draw more than a couple of games out of a hundred against God. He has among many other things a very good understanding of contempt.

Vas
Parent - - By Venator (Bronze) [de] Date 2007-12-30 12:26
I think He will predict all of our moves until the end of the game, so it easy to come up with a killer answer :-).
Parent - - By Vasik Rajlich (Silver) [hu] Date 2007-12-30 12:31
Ok, but that's not fair. Maybe Rybka could get a draw or two against a just God. :)

Vas
Parent - - By garyf919 (**) [us] Date 2007-12-30 23:16
Hi all, i ususally just read the posts here, but I have to comment here.
Rybka just recently lost a match to Zappa Mexico and now to Hiarcs!
and you want to consider Rybka could have a few draws against God?
What are you thinking? This is most ludricious.

What Vas needs to consider from a business point of view is how not
to wait till he creates a program that can beat God and how to not get
run over by Zappa and Hiarcs before he releases Rybka 3.0. No?

I would appreciate any comments on this. Thanks.
Parent - - By turbojuice1122 (Gold) [us] Date 2007-12-30 23:18
Hiarcs?  What in the world are you talking about?   Try a match on your machine between Rybka 2.3.2a with any decent opening book against Hiarcs 11.2 with its tournament opening book or anything you can find that's better, and see what happens...
Parent - - By garyf919 (**) [us] Date 2007-12-30 23:22
I dont have Zappa or Hiarcs. I am only going from these tournament results,
which would not be those programs anyway, but I am sure they will be coming,
maybe even befor Rybka 3.0.
Parent - By turbojuice1122 (Gold) [us] Date 2007-12-30 23:34
Well, then you're making a judgment based on a tie in a 7-game Swiss tournament.  You would be much better off checking the rating lists since you have nothing else to go on.  In the case of Zappa, however, you can see individual games, and notice that when they're played on Zappa's turf, Zappa wins, but when they're played on Rybka's turf, Rybka wins.
Parent - By ebutaljib (****) Date 2007-12-30 23:20 Edited 2007-12-30 23:26

> Ok, but that's not fair.


Why is this not fair? Aren't you trying to predict your opponents move when playing chess? I know I'm trying to do it, Anand does it,  Rybka does it, everybody does it.

If we assume that the game of chess is drawn if both sides play perfectly (which is probably true) then there is no way to win against god. To be able to draw one "just" has to play the moves which don't change the outcome of the game (draw). There is nobody (or anything) that would be capable of doing that at the moment. If someone or something would be able to do it, then it would never lose a game. No human and no chess engine is invincible for others. Everybody lost at least once.

I think someone should write a program whose goal is not to win the game, but not to lose against anybody. I think it's impossible. The engine would lose a game sooner or later eventhough it would play the most defensive, non-risk taking and passive chess.
Parent - - By bluemax (**) [ie] Date 2008-01-01 02:04
Against a just and mercyful God maybe even a win or two.
Parent - By Uly (Gold) [mx] Date 2008-01-01 02:41
No, if God can beat you, why would s/he let you win? That would be unfair to her/him.
Parent - - By Uri Blass (*****) [il] Date 2007-12-30 12:42
I think that there is something to do against it.
choose a random move so god cannot predict your move.

Unfortunately I am afraid that there is no way to choose random move that is going to be random from god's point of view.

As a feminist(some radical feminists do not define me as a feminist but I am not going to discuss it here) I also see no reason to treat god as "he" and not as "she"
I suggest using the word "it" for god.

Uri
Parent - By Venator (Bronze) [de] Date 2007-12-30 12:46
At the University where I studied, they had a nice saying on the door of the toilet:

'When God made men, She was only testing'  :-)
Parent - - By turbojuice1122 (Gold) [us] Date 2007-12-30 15:43
There is a way of choosing a random move: have the randomness be generated by quantum mechanical phenomena, such as the location of a particle in an infinite square potential.  If there are N possible drawing moves, put the electron in the Nth energy level, and which move made will be dictated by the closest proximity of which particular antinode the particle is found.  Of course, this would need to be performed in reality, something that is quite difficult, not just "on paper".  In reality, the best that God can do is instantly (and perfectly) do the same calculations that we would do on paper to calculate the probability that the particle will be found in such and such range--but since it is an infinite square potential, the ranges between the two nodes of each particular antinode have exactly the same probability.  Of course, there are no infinite square potentials in the real world, so in reality, there would be some antinode ranges that would have slightly greater probability than others--but even so, God doesn't know where is the particle until it's measured.
Parent - By Arrière Pensée (Gold) Date 2007-12-30 17:29 Edited 2007-12-30 17:32
Humanity (in our individuality) is the fragmentation of God's consciousness on earth. God would have to get all of us on the same Grid-mind and schedule a game against Rybka. Only God could keep all of humanity on the same Grid-mind long enough to play a tournament. Of course some of the minds would burn out before the game-causing a mind-brown-out. But, Rybka is used to that kind of temporary inconvenience.

Grid-Mind vs Rybka

sidserious-God! I'm on your side.

P.S. God would have to put a gag order on Turbo! :-)
Parent - - By Uly (Gold) [mx] Date 2007-12-30 19:27

> There is a way of choosing a random move: have the randomness be generated by quantum mechanical phenomena


But randomness doesn't exist for God, and it's just an illusion for humans, as they don't understand some events and think that they happen at random.

God knows if your coin is going to be head or tails, before time, as s/he has a perfect calculation of the events of the universe.
Parent - - By turbojuice1122 (Gold) [us] Date 2007-12-30 19:32
You're thinking classically (i.e. not quantum mechanically), where it is indeed true that God knows whether the coin will fall on heads or tails, every time.  However, even God must obey the laws of quantum mechanics, or else quantum mechanics would not exist (i.e. all wavefunctions would be collapsed for all eternity, something that is in violent disagreement with our existence!).  However, unlike humans or any other instrument, if you take a given physical setup, God will correctly be able to calculate the probabilities of such and such happening with infinite precision, and that must be how His decisions are made in such a case.
Parent - - By Hetman (*****) Date 2007-12-30 20:04
I am not sure if the laws of quantum mechanics are obeying ?

I see that you are doing the theology study here and you are designig your own god.
It is very slippy area.

Rgds
Hetman
Parent - - By turbojuice1122 (Gold) [us] Date 2007-12-30 20:06
I am simply saying that God must obey His own laws.  Yes, I agree that it is a difficult area to discuss theologically, but this is something in which I have very high confidence, as does pretty much any other physicist who believes in God.
Parent - - By Hetman (*****) Date 2007-12-30 20:23
I doubt if it is good assumption.
Laws are for people and are not obeying everyone (even some people are, were over the law).
Parent - - By turbojuice1122 (Gold) [us] Date 2007-12-30 20:30
Laws are for people and not obeying everyone

We're talking about laws of the Universe here.  Humans don't make these laws--they only discover them and test them to make sure that they're really laws.  After lots of testing, theories are formed that explain why the laws work this way.  The points about which I have been talking thus far in quantum mechanics are laws that are part of the current theory of how quantum mechanics works.  There are laws of gravity--but massive bodies still attracted before human existence (if not, then we obviously would never have been here).

There are some people out there who believe that God uses the possibilities in quantum mechanics to influence events in the Universe.  Unfortunately, this would seem to go against the idea that there is any free will.  Einstein said, "God does not play dice with the Universe."  Bohr responded, "Stop telling God what to do!"  In order for God to know not only the probabilities, but also the precise outcomes of the probabilities (and thus break His laws of quantum mechanics), not only would God be playing dice with the Universe, but would be cheating, and once the die lands, flip it precisely to the number He wants to come up.  This would, of course, collapse every wavefunction that could have ever existed even before their existence, and by definition, they wouldn't exist, which would be a contradiction.
Parent - - By Hetman (*****) Date 2007-12-30 20:38
People are over the laws created by themselves so .... there is possibility to be over the Universe laws.  ;-)

Einstein has said everything is relative :-).
Parent - By turbojuice1122 (Gold) [us] Date 2007-12-30 20:53
Actually, as explained in Nigel Calder's book "Einstein's Universe", Einstein did not make that remark at all, and that he really wanted to call his main 1905 result "theory of invariance".  I mulled over this statement for the better part of a decade until I studied the mathematics of special and general relativity in more detail during my undergrad years, where I immediately realized the importance of "invariants" throughout the framework.

There is the "possibility" for God to be over the Universe laws, but I believe that He chooses not to be, as in that case, nothing would really make sense in the Universe.
Parent - - By Banned for Life (Gold) Date 2007-12-30 20:31
People who try to violate the laws of physics usually don't fare very well. :-)

Regards,
Alan
Parent - By turbojuice1122 (Gold) [us] Date 2007-12-30 20:32
I once quipped to someone who asked what would happen to someone who violated the laws of physics that the penalty would be an eternity in an infinite square well.
Parent - By Hetman (*****) Date 2007-12-30 20:35
People ... :-)
Parent - - By Uly (Gold) [mx] Date 2007-12-30 20:47

> I am simply saying that God must obey His own laws.


You believe in a very limited God. In my belief, God could just copy the universe, and paste it somewhere else, break some laws to gain information, that will lead to the destruction of the universe's copy, and then come back to our universe to beat any non-perfect player at chess.
Parent - By turbojuice1122 (Gold) [us] Date 2007-12-30 20:59
Actually, there is no problem with God doing exactly what you propose--in fact, many physicists believe that the equivalent is being done all the time (see Everett's "Many Worlds Interpretation").  Unfortunately, it doesn't give the results that you might wish: if God used this knowledge from the destroyed Universe, He wouldn't come to correct conclusions any more often than using His perfect knowledge of all wavefunctions in the Universe.
Parent - - By noctiferus (***) [it] Date 2007-12-30 20:45
Hi Turbo: maybe my quantum mechanics is a bit dated, but collapsing to a pure state should happen when you perform a physical experiment: are we so sure that God would need a physical experiment to know the state of universe? This needs some theology, not physics, IMO.
Parent - - By turbojuice1122 (Gold) [us] Date 2007-12-30 20:57
The equivalent of this statement was actually a major point of debate among physicists before the 1960's, with Einstein taking your view.  However, Bell inequalities and experiments thereafter showed that there are no "hidden variables" of any sort.  The reason this is important and equivalent to this discussion is that hidden variables would occur when there is, "on the outside", an "actual" state to the system that one simply cannot measure.  That's why it is far more difficult than one might think, even theologically, for God to do anything more than simply have perfect knowledge of all wavefunctions in the Universe.
Parent - By noctiferus (***) [it] Date 2007-12-30 21:02
wow,my qm is more obsolete than I thought. Interesting, thanks.
If you would be so kind to give me some tech reference by PM, thanks again :)
Parent - By BB (****) [gb] Date 2007-12-31 09:28
but even so, God doesn't know where is the particle until it's measured.

God doesn't "know the future", because with God, there is no future, only the eternal now. [I won't digress on the metaphysics of the Copenhagen interpretation].

Or, more poetically (Knox I think, at least in the second part, perhaps as a response to the Balliol Hegelian of the first):

There was a young man who said, "God
Must think it exceedingly odd
That the juniper tree
Just ceases to be
When there's no one about in the quad."

Dear Sir, it is not at all odd,
I am always about in the quad.
Thus the juniper tree
Never ceases to be
Since observed by, Yours faithfully God.
Parent - - By BB (****) [gb] Date 2007-12-31 11:44

> If there are N possible drawing moves, put the electron in the Nth energy level, and which move made will be dictated by the closest proximity of which particular antinode the particle is found.


This is a perfectly reasonable (if experimentally clumsy) method for how to quantify the desired ELO estimation without conflating it theologically. You could altematively simulate randomness via Lava Lamps, for instance. Then we could discuss chaos rather than non-contextuality...

> However, even God must obey the laws of quantum mechanics


At best, you could claim that the actions of God as perceived in visible creation must appear to obey the laws therein.

> [...] influence events in the Universe.  Unfortunately, this would seem to go against the idea that there is any free will.


This would seem to be a strange conclusion. If one action is not free, does it follow that all actions are not free?

> That's why it is far more difficult than one might think, even theologically, for God to do anything more than simply have perfect knowledge of all wavefunctions in the Universe.


I'm not sure what theology would claim this, but, most likely, the ever-industrious ninth century Syrians (particularly those attuned to negation heory) probably already proposed it. :)
Parent - - By turbojuice1122 (Gold) [us] Date 2007-12-31 14:06
could altematively simulate randomness via Lava Lamps, for instance

Unfortunately, lava lamps wouldn't work because they obey the laws of classical chaos, not quantum chaos.  God has perfect knowledge over all future outcomes involved in classical chaos, and someone making moves based on such actions from among a group of apparently good moves would lose if God chose to exercise His potential knowledge of the classical future.  The quantum picture is, as it sounds like you already know, a very different story: the fact that God's actions are limited by the very framework of this has very noticeable, observable, and testable consequences in your Universe.  This is closely linked to your statement below:

At best, you could claim that the actions of God as perceived in visible creation must appear to obey the laws

I agree, but I also believe that not only in the realm of physics, but in the realm of God, this is the same thing as saying that God must obey the laws of quantum mechanics.

That's an interesting note on the Syrians--I don't know much about their ancient theology.
Parent - - By BB (****) [gb] Date 2007-12-31 14:38
That's an interesting note on the Syrians--I don't know much about their ancient theology.

Actually, maybe it was the fifth century... :-P [It is true that the theory of negation was a Syriac theory of about this time (later influencing, say, Maimonides), but it is also joked that almost every hair-brained philo-theo-sophical scheme imaginable was proposed by an obscure Syrian monk at some time - so the statement might be almost an inside joke in some circles].

this is the same thing as saying that God must obey the laws of quantum mechanics

So are we back to the identification with God and Nature (as per Spinoza, but with free will intact), or have we exploded into (say) Tipler's Cosmological Singularity? :) My personal contention would be that the subsuming standard ideas of time and causality into deistic notions is, at the outset, unwise.
Parent - - By turbojuice1122 (Gold) [us] Date 2007-12-31 15:14
On the second point, we're back with the identification with God and Nature.  I think that the above arguments about God's chess-playing "ability", as laid down by the rules and possibilities in the game of chess, still hold if this identification is only approximately true.  I'm glad to see that you know the Tipler paper :-).
Parent - By BB (****) [gb] Date 2007-12-31 15:46
I'm glad to see that you know the Tipler paper :-)

Actually, I ran across his book (on the Physics of Immortality - he now has a recent one on the Physics of Christianity also) in a random campus-town bookshop maybe 8-9 years ago. And then a few years later, when musing about models of computation, I ran across the redoubtable Warren D. Smith's "review" of it...
Parent - - By 8lrr8 (***) Date 2007-12-30 12:38
if, say, rybka 5.0+ is able to win ~80%+ of its games against v. 3.0, i may be forced to admit i am wrong, but i dont see this happening.  anyway, in a few yrs people can test out my conjecture.
Parent - - By Vasik Rajlich (Silver) [hu] Date 2007-12-30 12:41
Yes, we'll be able to test it. You should for the record give your estimate of God's rating.

I haven't seen any Elo asymptote since I started to measure this in my work on Rybka.

Vas
Parent - - By Venator (Bronze) [de] Date 2007-12-30 12:48
I fear we must make book draws, coming out of the opening with an equal position that is in
the tablebases. But I am afraid God would go 1.e4 a5 and prevent that :-).

My estimation is that God is like Chinook in checkers and will never lose and win most games.
I don't know which rating that corresponds with...
Parent - - By Vasik Rajlich (Silver) [hu] Date 2007-12-30 14:06
Book preparation for God could be quite difficult :)

Vas
Parent - - By ebutaljib (****) Date 2007-12-30 23:32
God knows every possible outcome from every possible setup, so the only usefull preparation against god would be 32 men tablebases :)
Parent - - By Uri Blass (*****) [il] Date 2007-12-31 10:21
32 men?

Why not 32 women.

I am also not sure if god knows every possible outcome from every possible setup.
Maybe you overestimate the knowledge of god.
Parent - - By ebutaljib (****) Date 2007-12-31 11:50
If he doesn't know it, then he is not god. There is nothing god doesn't know. By definition he is almighty, so he knows everything. He knows exactly what you are going to do and what the outcome will be before you even learned to play chess :)

By the way I find all this idea about the allmighty supernatural entity that is "watching over us" as an utter nonsence. Only humans can be stupid enough to make up (and believe in) something like that.

But if we assume that such an entity does exist (allmighty, creator of everything, etc.) then there is no way that he wouldn't know everything.
Parent - - By Uri Blass (*****) [il] Date 2007-12-31 13:51
"if we assume that such an entity does exist (allmighty, creator of everything, etc.) then there is no way that he wouldn't know everything."

You assume more than one assumption.
People believe that there was somebody who created the world in order to reply the question how the world exist.

They call this thing god.
They cannot get the conclusion that god knows everything based on the assumption that they believe and they even cannot get the conclusion that god is not dead now.

Uri  
Parent - By ebutaljib (****) Date 2007-12-31 16:03
So he created this universe and just left? Or even die? Interesting, but this is not the idea most people have about god.
Parent - - By Vasik Rajlich (Silver) [hu] Date 2008-01-03 15:39

> Maybe you overestimate the knowledge of god.


Doesn't being the creator of the universe earn you some benefit of doubt :)

Vas
Parent - - By Banned for Life (Gold) Date 2008-01-03 19:07
Not from Uri! :-) :-)

Alan
Up Topic Rybka Support & Discussion / Rybka Discussion / empirical estimate of God's elo rating
1 2 3 Previous Next  

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill