7 ms·

This reminds me of a question posed by an online user of a message board I frequent: Will the likes of Deep Blue and Alpha Zero kill competitive chess? My init

by ddxxdd 6y ago

This reminds me of a question posed by an online user of a message board I frequent: Will the likes of Deep Blue and Alpha Zero kill competitive chess? My initial thoughts were, "No, of course not! Cars and trains didn't kill track-and-field competitions!". But now that I think about it, I really do feel a slight aversion towards getting back into chess. Although practiced concentration is a side-benefit of playing chess regularly, the fact of the matter is that nowadays, it is impossible to play in a manner that hasn't already been thoroughly examined by some AI. There's no new ground to discover; no interesting techniques or thought-processes that an amateur or a world champion can invent; there is no more sense of human innovation in the game. If I picked up a chessboard for the first time in a decade, there would always be a chance that I can come to defeat Gary Kasparov someday; that is not the case with AlphaZero or DeepBlue. Say what you want about the benefits of forced concentration or the social benefits, but that desire to play chess has been extinguished for me, and I'm sure for many others, once AlphaZero and DeepBlue hit the scene.

pea 6y ago
I see where you're coming from, but I think the practical chances of a given chess player who hasn't played in a decade becoming GM-level are realistically similar to beating Alpha Zero. Ben Finegold tweeted about this once: https://twitter.com/ben_finegold/status/1316936206844821505?lang=en-gb https://twitter.com/ben_finegold/status/1316936206844821505?...
- Rerarom 6y ago
  Yeah but this would to me be an argument against chess in general, whether pre- or post-AI. Math is better, I feel like I can really contribute stuff that no one else would otherwise.
- nemo44x 6y ago
  Chess has to be the only game that regular people actually think they could become elite at. If you’re just picking it up in your late teens, there is basically no chance of becoming GM or even IM. Nearly every top-rated kid will never become GM for that matter. It seems like people really downplay difficulty of getting to that level. Even comments from the parent about hypothetically being able to beat Kasparov imply a certain misunderstanding of the vast gap between the 1% best players and the .01% best players. I suppose it’s fair to have fantasies about the possibility of beating the best of the best but the reality is for almost every person, a player like Kasparov in his prime is no different than a computer.
danparsonson 6y ago
A more suitable analogy to me would be - did Formula 1 kill off all other motor racing series? I don't think it did. Do you feel like, because you couldn't beat a supercomputer there's no point in playing all the humans on the way towards competing at the highest levels? Also, was there much innovation left in the game even before the AIs took over? As a casual chess player I (probably naively) thought that the game had been explored pretty exhaustively already. No sarcasm intended, as I said I'm strictly a casual player so I enjoy the game for the intellectual challenge and the social element, but all the upper echelon stuff is a mystery to me.
matsemann 6y ago
I don't get this line of thinking at all. It seems like such a random goalpoast. Why does it matter if the computer is better? If you're an 1800 rated player, you're still so far behind the best humans that it's irrelevant. Or why care if a computer has examined some similar positions before? As an amateur the positions won't be novel in qny way anyway. No more human innovation is blatantly false. If anything, computers have led to lots of new ways to play. And humans play humans, what a computer would do in a position is irrelevant.
- mdorazio 6y ago
  For me, AlphaZero was the overwhelming proof that a game with a fixed board and a limited set of rules is fundamentally an exercise in pattern recognition and response, not of actual innovation or creativity. This applies every bit as much to chess and go as it does to checkers. I realized that I really didn't want to try to be a better pattern recognizer (aka computer) than other humans, so I stopped playing and put that time toward endeavors that humans actually excel at: creating entirely new things.
  - 35fbe7d3d5b9 6y ago
    > creating entirely new things Like a writer?[1] Or a photographer?[2] A graphic designer?[3] Careful, the computers are beginning to horn in on everything :) [1] https://www.theguardian.com/commentisfree/2020/sep/08/robot-wrote-this-article-gpt-3 https://www.theguardian.com/commentisfree/2020/sep/08/robot-... [2] https://www.thispersondoesnotexist.com/ https://www.thispersondoesnotexist.com/ [3] https://brandmark.io/ https://brandmark.io/
  - perfmode 6y ago
    could you ever find solid ground?
dmurray 6y ago
> but that desire to play chess has been extinguished for me, and I'm sure for many others, once AlphaZero and DeepBlue hit the scene. It was extinguished in 1996, when DeepBlue played? Or 2018, when AlphaZero was announced? In the intervening time, an entire generation of chess players - and many generations of chess computers - has come and gone. I understand the feeling that there's nothing new to discover - but if you conflate the state of competitive chess in 1996 with that in 2018, you probably don't know how much you don't know. And if you could have hypothetically beaten Kasparov, you'd certainly have had a chance against Deep Blue.
tasuki 6y ago
I (used to) play go competitively (European 4dan). AlphaGo took me by surprise, but apart from the occasional cheater, its consequences are mostly great: - I have a superhuman teacher available anytime, ready to analyze all my games and answer all my questions (though the "why?" question sometimes takes a fair bit for the teacher to explain). - A lot of widely accepted dogmas in go were shaken, that was just a whole lot of fun. - Learning joseki (corner sequences) finally makes sense (they used to develop and change every year, which on my level wasn't worth keeping on top of). - It's a sneak peek into what's coming with AI. I watched the Lee Sedol match live, some of the AlphaGo moves were amazing. The human commentary was enlighteningly wrong: "oh he lost the first game, surely he'll win the next four"
IAmNotAFix 6y ago
On the other hand, the fact that it stays unsolved and isn't predicted to be solved in the near future (and it's not sure whether we'll one day have enough computing power to solve it by brute force) indicates that it probably has a long life ahead. Games only are really ended when they're solved.
ApolloFortyNine 6y ago
>Will the likes of Deep Blue and Alpha Zero kill competitive chess? In a way it's led to more preparation requirements for serious tournaments. You can load up a board in any position and Stockfish will tell you the best possible move from that point. Once you review your games in Stockfish/similar, it's honestly hard to think about how you would learn what you did wrong in the past, on your own at least. Unless you had a GM teacher, even your teacher may not be 100% all the time, but chess engines have gotten to the point where they essentially are 100% correct.

by ddxxdd 6y ago

pea 6y ago

I see where you're coming from, but I think the practical chances of a given chess player who hasn't played in a decade becoming GM-level are realistically similar to beating Alpha Zero. Ben Finegold tweeted about this once: https://twitter.com/ben_finegold/status/1316936206844821505?lang=en-gb https://twitter.com/ben_finegold/status/1316936206844821505?...

Rerarom 6y ago
Yeah but this would to me be an argument against chess in general, whether pre- or post-AI. Math is better, I feel like I can really contribute stuff that no one else would otherwise.
nemo44x 6y ago
Chess has to be the only game that regular people actually think they could become elite at. If you’re just picking it up in your late teens, there is basically no chance of becoming GM or even IM. Nearly every top-rated kid will never become GM for that matter. It seems like people really downplay difficulty of getting to that level. Even comments from the parent about hypothetically being able to beat Kasparov imply a certain misunderstanding of the vast gap between the 1% best players and the .01% best players. I suppose it’s fair to have fantasies about the possibility of beating the best of the best but the reality is for almost every person, a player like Kasparov in his prime is no different than a computer.

danparsonson 6y ago

A more suitable analogy to me would be - did Formula 1 kill off all other motor racing series? I don't think it did. Do you feel like, because you couldn't beat a supercomputer there's no point in playing all the humans on the way towards competing at the highest levels? Also, was there much innovation left in the game even before the AIs took over? As a casual chess player I (probably naively) thought that the game had been explored pretty exhaustively already. No sarcasm intended, as I said I'm strictly a casual player so I enjoy the game for the intellectual challenge and the social element, but all the upper echelon stuff is a mystery to me.

matsemann 6y ago

I don't get this line of thinking at all. It seems like such a random goalpoast. Why does it matter if the computer is better? If you're an 1800 rated player, you're still so far behind the best humans that it's irrelevant. Or why care if a computer has examined some similar positions before? As an amateur the positions won't be novel in qny way anyway. No more human innovation is blatantly false. If anything, computers have led to lots of new ways to play. And humans play humans, what a computer would do in a position is irrelevant.

mdorazio 6y ago
For me, AlphaZero was the overwhelming proof that a game with a fixed board and a limited set of rules is fundamentally an exercise in pattern recognition and response, not of actual innovation or creativity. This applies every bit as much to chess and go as it does to checkers. I realized that I really didn't want to try to be a better pattern recognizer (aka computer) than other humans, so I stopped playing and put that time toward endeavors that humans actually excel at: creating entirely new things.
- 35fbe7d3d5b9 6y ago
  > creating entirely new things Like a writer?[1] Or a photographer?[2] A graphic designer?[3] Careful, the computers are beginning to horn in on everything :) [1] https://www.theguardian.com/commentisfree/2020/sep/08/robot-wrote-this-article-gpt-3 https://www.theguardian.com/commentisfree/2020/sep/08/robot-... [2] https://www.thispersondoesnotexist.com/ https://www.thispersondoesnotexist.com/ [3] https://brandmark.io/ https://brandmark.io/
- perfmode 6y ago
  could you ever find solid ground?

dmurray 6y ago

> but that desire to play chess has been extinguished for me, and I'm sure for many others, once AlphaZero and DeepBlue hit the scene. It was extinguished in 1996, when DeepBlue played? Or 2018, when AlphaZero was announced? In the intervening time, an entire generation of chess players - and many generations of chess computers - has come and gone. I understand the feeling that there's nothing new to discover - but if you conflate the state of competitive chess in 1996 with that in 2018, you probably don't know how much you don't know. And if you could have hypothetically beaten Kasparov, you'd certainly have had a chance against Deep Blue.

tasuki 6y ago

I (used to) play go competitively (European 4dan). AlphaGo took me by surprise, but apart from the occasional cheater, its consequences are mostly great: - I have a superhuman teacher available anytime, ready to analyze all my games and answer all my questions (though the "why?" question sometimes takes a fair bit for the teacher to explain). - A lot of widely accepted dogmas in go were shaken, that was just a whole lot of fun. - Learning joseki (corner sequences) finally makes sense (they used to develop and change every year, which on my level wasn't worth keeping on top of). - It's a sneak peek into what's coming with AI. I watched the Lee Sedol match live, some of the AlphaGo moves were amazing. The human commentary was enlighteningly wrong: "oh he lost the first game, surely he'll win the next four"

IAmNotAFix 6y ago

On the other hand, the fact that it stays unsolved and isn't predicted to be solved in the near future (and it's not sure whether we'll one day have enough computing power to solve it by brute force) indicates that it probably has a long life ahead. Games only are really ended when they're solved.

ApolloFortyNine 6y ago

>Will the likes of Deep Blue and Alpha Zero kill competitive chess? In a way it's led to more preparation requirements for serious tournaments. You can load up a board in any position and Stockfish will tell you the best possible move from that point. Once you review your games in Stockfish/similar, it's honestly hard to think about how you would learn what you did wrong in the past, on your own at least. Unless you had a GM teacher, even your teacher may not be 100% all the time, but chess engines have gotten to the point where they essentially are 100% correct.