Complexity Physics and Tipping Points in Chess

[jedishrfu]

https://arstechnica.com/science/202...-finds-crucial-tipping-points-in-chess-games/

The game of chess has long been central to computer science and AI-related research, most notably in IBM's Deep Blue in the 1990s and, more recently, AlphaZero. But the game is about more than algorithms, according to Marc Barthelemy, a physicist at the Paris-Saclay University in France, with layers of depth arising from the psychological complexity conferred by player strategies.

Now, Barthelmey has taken things one step further by publishing a new paper in the journal Physical Review E that treats chess as a complex system, producing a handy metric that can help predict the proverbial "tipping points" in chess matches.
...

 

[jedishrfu]

It's interesting that no one is interested except me.

 

[javisot]

It's interesting that no one is interested except me.

Personally, I found this very interesting. As a chess player, I've always wondered if there exists a finite, simple, and perhaps dynamic set of tips that, when executed, lead to only winning games.

That is, if we divide the set of all possible chess games into winning games (for me) or losing/draws games (for me), what are the differences between these sets? And, above all, could these differences be summarized in a finite, simple set of tips so I don't have to remember predetermined matches?

(Not taking into account possible human errors resulting from lack of concentration)

It was very interesting, thanks for sharing it.

 

[jedishrfu]

I guess you proved and disproved my point. You are the first to comment after 7 months of inactivity.

—-

Back to chess: A few months ago, I saw a Youtube video of a retired grandmaster playing a winning game by only moving his pawns His first 17 moves was to fill the board with his pawn forces and throwing a monkey-wrench into his opponents game strategy.

How do you combat such a seemingly slow moving strategy where every move you make opens you to a pawn attack?

 

[javisot]

How do you combat such a seemingly slow moving strategy where every move you make opens you to a pawn attack?

In that game, there comes a point where pawn development blocks any attack by black's, the rest of white's pieces can quickly enter the game. The distribution at a certain point is clearly favorable for white's. The way to combat this would have been from the beginning, knowing that his intention was to develop only pawns. Developing only pawns is a strategy with few options; it depends on your opponent not understanding what you want to do. Since it's an unusual strategy, it may happen that an opponent doesn't realize that you're going to develop only pawns, as seems to be the case in this game.

In any case, it's a complicated exercise. Suppose his strategy from the beginning was to develop only pawns, and his opponent prevents him from doing so, then he would have resorted to another strategy, developing other pieces.

 

[Halc]

You are the first to comment after 7 months of inactivity.

Maybe because the OP offered only a line and a snip from that link, and no actual commentary from yourself.

The pawn thing was interesting, but had little direct relevance to the OP.

I've always wondered if there exists a finite, simple, and perhaps dynamic set of tips that, when executed, lead to only winning games.

It has been generally shown (not proven) that no such strategy can exist. Chess is a game of mistakes. If both sides make none, it's going to be a draw.
Still, there are these critical so called tipping points, the places in the game where your thinking time is best spent. The moves immediately following one are likely similar to the opening: Fast reactive moves.

The article in the OP apparently proposes a metrc to help identify these points. I didn't read it.

 

[javisot]

It has been generally shown (not proven) that no such strategy can exist. Chess is a game of mistakes. If both sides make none, it's going to be a draw.

I've heard this many times too, but I think it's an outdated and unproven view of chess. We're clear that chess is a finite game; it was discussed here https://www.physicsforums.com/threads/possible-number-of-chess-games.1067754/ . In order to debate this, we don't really need to know the exact number of possible games, just that it's a finite amount.

The idea that if we have two players who have memorized all the games and can use that information in a reasonable amount of time without making mistakes then they should always draw doesn't seem correct.

The turn seems to be an important factor. Imagine a distribution of pieces on the board such that, if white moved, they would checkmate, but if black moved, they would checkmate. Clearly, there is a "turn advantage," and the question remains: Is there an initial advantage for the one who starts or for the one who answers? (This question has not yet been answered with complete precision)

It seems reasonable that such an advantage exists. If we look at the percentages for players and bots, the percentage of draw games is always minimal compared to the percentages of lost/won games. These are the percentages of a game in which you essentially win or lose.

 

[jedishrfu]

I need to say your comment about my first post is rather offensive. I posted the article for others to read and comment. I am a site mentor and this is how we post articles in addition to our other duties.

My comment to @javisot was not targeted at him or meant to be any kind of insult. It was a humorous comment of the state of PF where many posts seldom get any responses.

 

[sbrothy]

I guess you proved and disproved my point. You are the first to comment after 7 months of inactivity.

—-

Back to chess: A few months ago, I saw a Youtube video of a retired grandmaster playing a winning game by only moving his pawns His first 17 moves was to fill the board with his pawn forces and throwing a monkey-wrench into his opponents game strategy.

How do you combat such a seemingly slow moving strategy where every move you make opens you to a pawn attack?

Is this the famous game by Philidor who liked to say that pawns are the soul of chess? He played a game where he crushed his opponent while letting him take all his officers. This was in the 18th century or thereabouts though, if I remember correctly.

EDIT: I think this is the one Andrew Smith vs. François-André Danican Philidor: The pawn storm

 

[jedishrfu]

The grandmaster I posted from 1985 was nicknamed "The Pawnmower Man"

 

[sbrothy]

The grandmaster I posted from 1985 was nicknamed "The Pawnmower Man"

I see. But Philidor is a watch worth too though.

 

[sbrothy]

I see. But Philidor is a watch worth too though.

I play chess and I've studied a lot of openings. Naturally I know the name Reti but I'll admit I haven't really looked too much into his style.

EDIT: Nimzowitsch on the other hand. His Indian attack/defense in particular.
EDIT2: What I mean by attack is that I always liked playing white against the indian defense.

 

[jedishrfu]

I used to know a lot of chess stories when I was a teen, playing with my friend, who played really well. It was like a mashup of Capt Kirk (me) vs Spock(him). He would be very methodical in analysis and writing down all moves for later study. I was more of an out-of-the-box thinker with no notion of strategy, only the ability to make unorthodox moves. I knew when I succeeded because he would stop flipping his pen, and that was a moment to savor.

After about 50 games, we played one more, and this time he didn't record the moves. My unorthodox move was to sacrifice my queen, and unexpectedly, after he captured her, I saw a way to checkmate. He was stunned and didn't know what to say. Finally, he said, "Let's play another game." and my response was nope, nope, nope, I think I need to get home. That was the best, and while we were still friends, we didn't play again; he went off to college.

 

[sbrothy]

I used to know a lot of chess stories when I was a teen, playing with my friend, who played really well. It was like a mashup of Capt Kirk (me) vs Spock(him). He would be very methodical in analysis and writing down all moves for later study. I was more of an out-of-the-box thinker with no notion of strategy, only the ability to make unorthodox moves. I knew when I succeeded because he would stop flipping his pen, and that was a moment to savor.

After about 50 games, we played one more, and this time he didn't record the moves. My unorthodox move was to sacrifice my queen, and unexpectedly, after he captured her, I saw a way to checkmate. He was stunned and didn't know what to say. Finally, he said, "Let's play another game." and my response was nope, nope, nope, I think I need to get home. That was the best, and while we were still friends, we didn't play again; he went off to college.

Hah. I know the feeling. Lately I've mostly been having fun with it. I'll sacrifize my bishop by smashing it into the enemy's king-side, preventing him from castling and using any official openings. Against computers it fails but against humans it often freaks them out enough to win me enough tempo to chase their king all over the board. Unfortunately it keeps my rating under 1000 but it's fun.

I played a real life game against this guy I know who's on the autistic spectrum. He has an online ELO rating >2200. After two hours it became painfully obvious that the first one to make a mistake would be toast. Even he became bored so we called it a draw. We never played again.

 

[sbrothy]

I see. But Philidor is a watch worth too though.

Hah. The wordplay just now dawned upon me: The "Pawnmower" man! Embarrassing.

 

[jedishrfu]

I imagine the chess players of today feel like the Okinawan Go masters who didn’t progress as fast as their Japanese counterparts due to the distance between their two respective islands.

For chess players of today, the chess engines can seem daunting unless you persevere to become greater than the machine.

Lee Sedol, the world-class Go champion got beat by AlphaGo from Google. The first three went to AlphaGo but the fourth went to Sedol when made an unexpected move that was part defensive and part offense never considered by AlphaGo before. Everyone gasped thinking Sedol had made a tactical error until they saw AlphaGo falter and lose many moves later. The fifth game went to AlphaGo.

Go has a higher branching complexity of $10^1270$ vs chess at $10^120$

Sadly, Lee Sedol retired in 2019 saying “AI cannot be beat.” I think thats the wrong view to take but I can understand the sentiment.

 

[sbrothy]

I imagine the chess players of today feel like the Okinawan Go masters who didn’t progress as fast as their Japanese counterparts due to the distance between their two respective islands.

For chess players if today, the chess engines can seem daunting unless you persevere to become greater than a machine.

Lee Sedol, the world-class Go champion when he got beat by AlphaGo from Google. The first three went to AlphaGo but the fourth went to Sedol when made an unexpected move that was part defensive and part offense never considered by AlphaGo before. Everyone gasped thinking Sedol had made a tactical error until they saw AlphaGo falter and lose many moves later.

Go has a higher branching complexity of $10^1270$ vs chess at $10^120$

Sadly, Lee Sedol retired in 2019 saying “AI cannot be beat.”

Sad. I was under the impression that Go was still the last frontier. Given the complexity of the game it's hard to think otherwise.

 

[jedishrfu]

Im sure today grandmasters will find a way but then the machine will learn from it and continue to dominate the games.

For AI developers winning these games was a challenge and once completed, they aren’t interested in pursuing it further.

Look at the IBM Chess match where DeepBlue defeated Garry Kasparov, a great marketing win for IBM. Afterwards, the winning team was disbanded. IBM like so many other tech companies has no soul.

Kasparov accused IBM of special assistance to DeeBlue and demanded a rematch (IBM declined and dismantled the machine).

 

[sbrothy]

Im sure today grandmasters will find a way but then the machine will learn from it and continue to dominate the games.

For AI developers winning these games was a challenge and once completed, they aren’t interested in pursuing it further.

Look at the IBM Chess match where DeepBlue defeated Garry Kasparov, a great marketing win for IBM. Afterwards, the winning team was disbanded. IBM like so many other tech companies has no soul.

Kasparov accused IBM of special assistance to DeeBlue and demanded a rematch (IBM declined and dismantled the machine).

Why am I not surprised?

 

[Hornbein]

Sad. I was under the impression that Go was still the last frontier. Given the complexity of the game it's hard to think otherwise.

When AlphaGo beat Lee Sedol I knew the revolution was here. AlphaZero that soon followed was much more impressive.

 

[jedishrfu]

I imagine in the future, we will be taught by machines using active learning to accelerate our progress in education.

Competitions would be among humans with the machines judging the participants.

We could even be physically trained by machines using active learning to master a variety of martial arts, gymnastics, music and arts.

The list is endless…

 

[javisot]

Kasparov said something about this a few days ago at an economic/political forum https://www.elmundo.es/internacional/2025/10/04/68e15e4321efa0b41c8b4575.html , comments on AI:

"There are things that are beyond our ability to control. Progress cannot be stopped. People will continue to invent. Machines have made us stronger and faster, helped us live longer, and should make us smarter. We must accept that they are already part of our world. A realistic approach is in order. We should not see them as a harbinger of utopia or dystopia. They are there, they are technology, and we must find a way to relate to them. ChatGPT and language models do not threaten our livelihood. In education, we must reform the system. It is not about accumulating knowledge, that can be found. It is about understanding patterns and being creative".