D-Wave Systems breaks the 1000 qubit quantum computing barri

[jefk]

[quote="Ozymandias"]

My point is that, chess being a draw, those things you find, when searching deeper, won't have any effect on the outcome. [/quote]

agree; but this doesn't mean that the game is finished or
it becomes boring; see e.g.
http://www-history.mcs.st-and.ac.uk/Pro ... s/Ch5.html

yes, Bob a gambler is losing with roulette if the casino has one or more zeros. No need to simulate all outcomes with a comp is there.

For chess, i believe it's a draw unless Black plays a suboptima
defense, like eg. 1e4 d5?

Anyway , referring to another thread, i believe the position after 1.e4 already is critical. Playable are 1..c5!, 1..e6, and possible 1..e5?!
After that the resulting positions often remain critical and Black has to play accurately; similar vs other strong Shite opening systems.

Just because Schaeffer, a computer guy not a fundamental mathematician/game theorist calculated (almost) all outcomes with checkers, it's not certain whether this is the only approach for games such as chess. Again, referring to Zermelo, we can think of games where it's possible to prove the outcome, without a brute force approach.

There can't be some miraculous opening move like eg. 1.f3 suddenly leading to a win in hundred after eg 300 moves or so , because ot the mathematical rigour of backsolving. Theorem by Neumann !
*If* such such a hypothetical move as 1.f3 would to a forced win,
then this would become obvious when backsolving a large posbase
like eg Arshah, and 1.f3! Would appear as best, ie as winning strategy.
But it doesn't.

In other words i don't think a rigorous brute force approach is needed
and people doing such things (or even thinking about it) imho are wasting their time. If their is a 'drawing strategy' for Black then the game is a draw per definition. And for me such a strategy means playing -very- strong defensive moves.

But the search for better openings isn't finished yet, and thus chess, although being a draw, hasn't become boring yet. It might when Chessbase would be years further, have implemented an algorithm as discussed here months ago backsolved (average of eval + constant*time*game result percentage); even then it would take
years to converge, and there still would be several playing/defending styles possible for Black, eg the sharp and difficult Sicilian or the more positional French whereby in the latter case also knowledge of eg the first 14 defending moves against the Tarrasch is useful or even needed.

When playing human otb chess who cares whether chess is a draw,
the imho same hold true for (robo)soccer(*), yet it's still a sport and
a game isn't it. And memorizing a complete Najdorf system nowadays already is quite hard even for Gm's, even imho although it's a draw


jef

(*) if (quantum?)robosoccer won't be a draw in eg 15 yrs, the first rules for the first kickoff or goal size, or so have to and will be changed, simple as that.

[jefk]

just a little addition/illustration:

[quote="jefk"]the more positional French whereby in the latter case also knowledge of eg the first 14 defending moves against the Tarrasch is useful or even needed.[/quote]

example from my book (for the chesspartner interface):
1. e4 e6 2. d4 2... d5 3. Nd2 3... c5 4. Ngf3 4... cxd4 5. exd5 5... Qxd5 6. Bc4 6... Qd7! 7. Nb3 7... Nc6 8. O-O 8... a6 9. Nbxd4
9... Nxd4 10. Nxd4 10... Qc7 11. Bb3 11... Nf6!N
( 11... Bd6?! 12. Qh5! +=)

checked with www.chessbites.com which should have all
reasonabl quality and up to date GM games and it appears
that Black move eleven already is a ' novelty'.
The move 11...Bd6?! has been played but then Stockfish
quickly is suggesting 12.Qh5 += (haven't checked it with K9).

Anyway, who cares about ' noveltie's , i don't because the current evals of eg K9 at sufficient depth imho are good enough. And the minimax theorem of Neuman was proven by him many decades ago:
https://en.wikipedia.org/wiki/Minimax

Sure, if we would only look at GM games up till now, there still might be something missing; but my point is, i have analyzed the game of chess
' from scratch' so to speak, discarding obviously losing moves, losing a piece and so on. And only with a posbase (in bookbuilder version 3.6) with some 30 million positions i already can see that Black can draw, no matter what. Forget to solve chess with the catalan, and even then Black can play Slav, or against 1.Nf3 c5! thus aver 2.c4 a symmetrical English against which Black also can equalize (with some difficulty, i admit).

Only during the opening phase i admit. So what. My claim is that if White cannot achieve a solid advantage at the end of the opening, Black will have fundamentally no problems in achieving a draw. Ergo chess = =.

jef

PS also with now the 6 man syzygy bases , it must be clear that if White would have a win, then this must be seen already in advance by the egtb's. If not, it's a draw. Now i give you guys as exercise to have a look how big such an advantage at the start of the egtb has to be at a minium in terms of 'normal' engine eval (at high depth, eg 40 ply with a strong engine and no egtbs installed). Personally i guess at least +3 or so.
Now tell me, if during the middle game the eval is eg only 0.1,
how the h.. can White increase this advantage to the requied
plus 3 ?? He cannot; mission impossible. Ergo chess = =

[jefk]

and a little rectification, sorry it's the 3d posting in a line, i would
have preferred to edit my earlier posting (including the typo's)
like eg possible on Facebook (but not here..)

[quote="jefk"]
Just because Schaeffer, a computer guy not a fundamental mathematician/game theorist [/quote]

oops, his masters was in math, but his Phd (probably) in
computer science and his job is in that area as well
(although in games, also a bit of math area, i admit).

As for my reference to Neumann, i don't mean ofcourse
positional eval with minimax, but minimax of a large posbase
with ' opening' positions. NB some GM's like Kasparov often
extended their 'opening' analysis until the endgame,
and there indeed there somtimes is a connection there
eg with sharp endgames (some Najdorf lines eg Dragon)
leading to practical endgames, where possible some
probing of the 6 men syzygy already occurs.

For positional games, this doesn't happen, but then there
are enough options for Black, and by playing the best option,
Black can ensure a draw, ie have a drawing strategy. Ergo
it's not possible for White, without a significant opening
advantage to nevertheless continue to a winning endgame.
Per definition. Because if it could, then this would become
a new opening line, and thus then later would be avoided by
Black. And so on. Ad nauseam but the point might be clear.
it =a =

jef
PS any academic interested in this research, not playing ICCF correspondence chess , and not getting the data from Arshah
(some billions of positions) after having tried that, can approach me
via a pm to ask for a copy of my 30 million posbase in the bb36 program
(backsolving aka minimax time < 30 mins). Endnode analysis/updating
with eg K9 at eg 30 ply would take more time but can be done in a few months. It often requires some finetuning, but then again my conjecture
can be seen clearly in practice; Black can always draw.
QED
an approach me via pm on this forum and get a copy

[bob]

You can always go deeper, and you can sometimes expose something that was not visible to the last iteration of the search…

My point is that, chess being a draw, those things you find, when searching deeper, won't have any effect on the outcome. Not with the current state of engines, and no unless you stumble upon a complex enough opening position, which your opponent doesn't have stored.

You can't use an assumption (chess is a draw) to prove it is a draw. You have to spend the time and do the search. All 100 gazillion nodes of it.

[bob]

My point is that, chess being a draw, those things you find, when searching deeper, won't have any effect on the outcome.

agree; but this doesn't mean that the game is finished or
it becomes boring; see e.g.
http://www-history.mcs.st-and.ac.uk/Pro ... s/Ch5.html

yes, Bob a gambler is losing with roulette if the casino has one or more zeros. No need to simulate all outcomes with a comp is there.

For chess, i believe it's a draw unless Black plays a suboptima
defense, like eg. 1e4 d5?

Anyway , referring to another thread, i believe the position after 1.e4 already is critical. Playable are 1..c5!, 1..e6, and possible 1..e5?!
After that the resulting positions often remain critical and Black has to play accurately; similar vs other strong Shite opening systems.

Just because Schaeffer, a computer guy not a fundamental mathematician/game theorist calculated (almost) all outcomes with checkers, it's not certain whether this is the only approach for games such as chess. Again, referring to Zermelo, we can think of games where it's possible to prove the outcome, without a brute force approach.

There can't be some miraculous opening move like eg. 1.f3 suddenly leading to a win in hundred after eg 300 moves or so , because ot the mathematical rigour of backsolving. Theorem by Neumann !
*If* such such a hypothetical move as 1.f3 would to a forced win,
then this would become obvious when backsolving a large posbase
like eg Arshah, and 1.f3! Would appear as best, ie as winning strategy.
But it doesn't.

In other words i don't think a rigorous brute force approach is needed
and people doing such things (or even thinking about it) imho are wasting their time. If their is a 'drawing strategy' for Black then the game is a draw per definition. And for me such a strategy means playing -very- strong defensive moves.

But the search for better openings isn't finished yet, and thus chess, although being a draw, hasn't become boring yet. It might when Chessbase would be years further, have implemented an algorithm as discussed here months ago backsolved (average of eval + constant*time*game result percentage); even then it would take
years to converge, and there still would be several playing/defending styles possible for Black, eg the sharp and difficult Sicilian or the more positional French whereby in the latter case also knowledge of eg the first 14 defending moves against the Tarrasch is useful or even needed.

When playing human otb chess who cares whether chess is a draw,
the imho same hold true for (robo)soccer(*), yet it's still a sport and
a game isn't it. And memorizing a complete Najdorf system nowadays already is quite hard even for Gm's, even imho although it's a draw


jef

(*) if (quantum?)robosoccer won't be a draw in eg 15 yrs, the first rules for the first kickoff or goal size, or so have to and will be changed, simple as that.

The odds for winning at roulette is something one can compute directly. There's no need for simulation. Chess is quite a different thing, and you can't say it is a draw until you prove it is a draw by search. Anything else is simply idle speculation, however probable it might be true. Certainly one can't say with 100% accuracy that chess is drawn because that has yet to be proven, and most likely never will.

BTW that is only going to converge on what the programs can see. That is like basing physics only on what happens on the surface of the earth. Wait. Someone did that thousands of years ago. Didn't stand up very well under more rigorous analysis either. Using a chess engine can't prove anything until the chess engine runs with a 3 point evaluation, win lose or draw. Whatever it says at that point will be the ultimate truth. Until that point in time, it is all speculation and conjecture. Perhaps chess has already been solved somewhere. But I'll guarantee you it is in some alternate universe that is far older than a puny 13.8 billion years however.

[bob]

just a little addition/illustration:

the more positional French whereby in the latter case also knowledge of eg the first 14 defending moves against the Tarrasch is useful or even needed.

example from my book (for the chesspartner interface):
1. e4 e6 2. d4 2... d5 3. Nd2 3... c5 4. Ngf3 4... cxd4 5. exd5 5... Qxd5 6. Bc4 6... Qd7! 7. Nb3 7... Nc6 8. O-O 8... a6 9. Nbxd4
9... Nxd4 10. Nxd4 10... Qc7 11. Bb3 11... Nf6!N
( 11... Bd6?! 12. Qh5! +=)

checked with www.chessbites.com which should have all
reasonabl quality and up to date GM games and it appears
that Black move eleven already is a ' novelty'.
The move 11...Bd6?! has been played but then Stockfish
quickly is suggesting 12.Qh5 += (haven't checked it with K9).

Anyway, who cares about ' noveltie's , i don't because the current evals of eg K9 at sufficient depth imho are good enough. And the minimax theorem of Neuman was proven by him many decades ago:
https://en.wikipedia.org/wiki/Minimax

Sure, if we would only look at GM games up till now, there still might be something missing; but my point is, i have analyzed the game of chess
' from scratch' so to speak, discarding obviously losing moves, losing a piece and so on. And only with a posbase (in bookbuilder version 3.6) with some 30 million positions i already can see that Black can draw, no matter what. Forget to solve chess with the catalan, and even then Black can play Slav, or against 1.Nf3 c5! thus aver 2.c4 a symmetrical English against which Black also can equalize (with some difficulty, i admit).

Only during the opening phase i admit. So what. My claim is that if White cannot achieve a solid advantage at the end of the opening, Black will have fundamentally no problems in achieving a draw. Ergo chess = =.

jef

PS also with now the 6 man syzygy bases , it must be clear that if White would have a win, then this must be seen already in advance by the egtb's. If not, it's a draw. Now i give you guys as exercise to have a look how big such an advantage at the start of the egtb has to be at a minium in terms of 'normal' engine eval (at high depth, eg 40 ply with a strong engine and no egtbs installed). Personally i guess at least +3 or so.
Now tell me, if during the middle game the eval is eg only 0.1,
how the h.. can White increase this advantage to the requied
plus 3 ?? He cannot; mission impossible. Ergo chess = =

I think you misunderstand minimax. Minimax doesn't prove ANYTHING other than it chooses the optimal path given a specific search tree AND a specific endpoint evaluation. That doesn't prove a single thing about the game of chess however. 30 years ago there were endings "proven" to be a win for white or black. Computers busted those by searching all the way to the end of the game. From simple endgames. Not from openings or middle game positions....

Minimax will give you the right answer IF you give it the right endpoint evaluations. That can't be done today.

[bob]

It would seem not everybody believes it is a draw, else nobody would waste the time playing the game.

Or they just believe their opponent, to be further from perfect play, than they are.

All I believe at the present time is that chess is a game. It might be won, lost are drawn with perfect play by white and black. However, this won't be proven in any reasonable time-frame (50 years, 500 years, 500,000 years, etc.)

[JVMerlino]

Here's a topical question. My apologies if this is a stupid question or common knowledge....

To what depth has the initial position been searched (presumably by brute force?) to PROVE that a win cannot be forced within that number of moves? In other words, has somebody said "I have proven that it is not possible to force checkmate within N ply", where N is something like 25?

Now that I think about it, I would be kind of shocked if it was even that high a number....

jm

[jefk]

[quote="bob"]Certainly one can't say with 100% accuracy that chess is
drawn because that has yet to be proven, and most likely never will.
[/quote]

no need for 100 pct accuracy with mathematical rigour. Since, 'm
a physicist, and you started talking about what happens on earh,
i challenge you to prove with 100 pct certainty that the sun
is revolving around the earth and not the other way around.
We then are getting in the area of the philosopphy of science,
now i'm not going to quote Feyerabend, but its not always
as easy as you might state. Saying straightaway what i think
of your claims that brute force calculation is needed, well i disagree,
and more politiely, i say/write i would agree to disagree.

Other example of excessive mathematical rigorous demands;
Flipping a coin will result in 50 pct chance getting heads or tails.
ok, another game than chess, but i'm going to give just an example:
If someone says nope the chance is 50.000000001 whereby
1 is approaching infinite, then most people would agree its 50 pct.
Even if someone is saying it's 49,99999 pct whereby the nr
of nines is approaching infinity.

Now if i say i'm 99,99999.... (and so on). pct certain ie have
found on the basis of my research and reasonings that
chess is a draw with the current rules, that's the same as
stating for me it is 100 pct certain. A complete rigorous 'proof' ?
Nope. Did Schaeffer did a *complete* brute force calculation ?
No he did *not*. I have his book, One jump ahead, and
he has used pruning ,and obviously has used endgame bases.
Then he claims he has found the result, checkers is a draw
and anybody believes him because he is a professor.

Well i have found chess is a draw but it's not found
at a university; nevertheless is still intend to write
paper about my findings, although i suspect that
most professors would say its not a proof already
because of the 'not invented here' syndrome.
Without having looked at my data. Why bother with
my data if you already know your answer; "chess
will never be solved' and so on (i disagree).

Then some agreements: yes the minimax depends
on the quality of the eval, while Komodo is not
perfect, the deeper the search, the better it becomes,
at least as a *start* for approaching the syzygy tables.
Again, like i said, if those tables give a draw, then
the result is a draw.

For Merlino, i not only have found chess is not mate
within x moves, i have proven it's never mate.
It's a draw. Maybe you can tell us what is a proof for you;
brute force calculation ? But we know that alfa-beta
is just as good. Alfa beta calculation ? Well i have done
that, my opening database in bb36 is pruned in a similar
way as alfa-beta. Why i use minimax to backsolve
these opening data then ? Answer because they are stored
on disc, just like eg Aquarium is using minimax in Idea.

But why bother with the data huh. I offered to demonstrate
my 30 million posbase where chess has been analyzed
from scratch, much better than eg. the CAP data.
But ofcourse got no PM from anybody. Well no problem, i
will write my article , first at academia.edu, then
maybe later at arxiv at Cornell or so. Those who
disagree may disagree. Same for ME.

So imhjo chess is a draw with 100 pct certainty and i
found that result with quite rigorous computer
analysis already some years ago. But some people
will never believe such results; you might as well
spend your time trying to find flaws in the syzygy endgame
tablebases; they *might* contain errors you know !
well have fun.

jef

PS as for the math people, is Monte Carlo analysis sound ?
If not, how sound is it then ? Can you find a winning strategy
for White with Monte Carlo analysis ? If not, what is
that saying to you ? stuff enough for some fun Phd work
(but the conclusion for chess already is clear for me).
Kuhn and many others have shown that the academic
world usually also consists of biased thinking, take
as example theoretical physics, one group is doing string
theory, the other is fighting it, and doing eg loop quantum
gravity. The truth is in the measurements. Like eg the
results in correspondence chess. Approaching 50 pct.
How much is 2 times 50 ? (question for mr Hyatt)
oh and you may add a % to the result.

[jefk]

[quote="jefk"]
i challenge you to prove with 100 pct certainty that the sun
is revolving around the earth and not the other way around.
[/quote]

Maybe you could with some transformational celestial mechanics
but in fact i meant the other way around. But not only that,
a proof please with a 100,000000000 etc pct certainty.

Copernicus couldn't (he got at 99 pct)
Galileo couldn't (he got at 99,99 pct)

Can we do it nowadays, with large computers ?
Simulating celestial mechanics. Maybe.
With brute force calculation ? And twenty decimals behind the point.
Including general relativity. Ofcourse.
Come-on..

It is not necessary.
That's how science works, you know.

Three more example for chess
A)
1. c4 0.00/0 0.0 e5 0.00/0 0.0 2. Nc3 0.00/0 0.5 Nf6 0.00/0 0.0 3. g3 0.00/0 0.6 Bb4 0.00/0 0.0 4. Nf3 0.00/0 0.0 Nc6 0.00/0 0.0 5. Bg2 0.00/0 0.0 d6 0.00/0 0.6 6. O-O 0.00/0 0.0 O-O 0.00/0 0.0 7. d3 0.00/0 0.0 a5 0.00/0 0.0 8. Bd2 0.00/0 0.0 Bf5 0.00/0 0.0 9. a3 0.00/0 0.0 Bxc3 0.00/0 0.0 10. bxc3 0.00/0 0.0 =
(thats not mate, Merlino)
B)
1. d4 Nf6 0.00/0 0.0 2. c4 0.00/0 0.0 c6 0.00/0 0.0 3. Nf3 0.00/0 0.0 d5 0.00/0 0.0 4. e3 0.00/0 0.0 Bf5 0.00/0 0.0 5. Nc3 0.00/0 0.0 e6 0.00/0 0.0 6. Nh4 0.00/0 0.0 Bg6 0.00/0 0.0 7. Nxg6 0.00/0 0.0 hxg6 0.00/0 0.0 8. a3 0.00/0 0.0 Nbd7 0.00/0 0.0 9. c5 0.00/0 0.0 Qc7 0.00/0 0.0 10. g3 0.00/0 0.0 a5 0.00/0 0.0 11. Bd2 0.00/0 0.0 Be7 0.00/0 0.0 12. b4 0.00/0 0.0 e5 0.00/0 0.0 13. Bg2 0.00/0 0.0 b6 0.00/0 0.0 14. O-O 0.00/0 0.0 O-O 0.00/0 0.0 =
C)
1. e4 c5 0.00/0 0.0 2. Nf3 0.00/0 0.0 d6 0.00/0 0.0 3. d4 cxd4 0.00/0 0.0 4. Nxd4 0.00/0 0.0 Nf6 0.00/0 0.0 5. Nc3 0.00/0 0.6 a6 0.00/0 0.0 6. Be3 e5 0.00/0 0.0 7. Nb3 0.00/0 0.0 Be7 0.00/0 0.5 8. f3 Be6 0.00/0 0.0 9. Qd2 0.00/0 0.0 O-O 0.00/0 0.0 10. O-O-O 0.00/0 0.0 Nbd7 0.00/0 0.6 11. g4 0.00/0 0.6 b5 0.00/0 0.6 12. g5 0.00/0 0.0 Nh5 0.00/0 0.6 13. Kb1 0.00/0 0.0 Nb6 0.00/0 0.0 14. Na5 0.00/0 0.0 ( 14. Nd5 0.00/0 0.0 ) 14... Rc8 0.00/0 0.0 15. a3 0.00/0 0.0 g6 0.00/0 0.0 16. h4 0.00/0 0.0 Qc7 0.00/0 0.0 17. Rh2 0.00/0 0.0 Rb8 0.00/0 0.0 18. Ka1 0.00/0 0.0 Rfc8 ( 18... Rfd8 0.00/0 0.0 )
19. Bh3 0.00/0 0.0

Whereby i leave it to IM or GM junkies of New in Chess or
Chessbase to find the socalled ' novelties' . Or claim other lines
are better for White or Black (could be, there are a lot of
drawish lines, you know

So indeed chess is a draw (*).

jef

(*) no i don't like rhetorics, but Galileo also kept repeating
that the thing is turning (around the sun) and this
isn't an academic forum anyway; it's about chess and
maybe after a while some blokes might recognize i
mean it; that = = = and stuff like that).
PS yes later Schaeffer did some more research adding
to his empirical claim that checkers is a draw.
(a claim he already made before a brute force
calculation had been done; believe it or not)
So i might do some more research, while having
some fun in correspondence chess as well, eg
with my secret 1.Nf3 lines; read my memorable
ICCf games in two decades, but in the end i will
support changing the rules of correspondence chess.
By now you might have a clue of the reasons..