No
Will computers ever solve chess?
Sort:
Chess is too difficult game
#6966
Extract from the above primer link #6959:
"Quantum computers will always be hybrid devices, partly quantum and partly normal. The latter is required to handle inputs and outputs, in order to interface with the lumbering apes who want to use the device. For this reason, quantum SDKs are typically embedded in a standard programming language. QISKit uses Python, so we can use a Python program to deal with both the normal parts of the program, and to construct and run jobs for the quantum part."
For the love of god… 7000 posts on something completely irrelevant. Google gave up on chess and is now doing something helpful, like unlocking the secrets of protein folding. Why not expend energy on something productive, like learning a new opening. These threads are all dumpster fires.
This is an empirical question so any post is just rampant speculation. Go spend 7000 posts arguing over how many unicorn toots it would take to fill 1000 balloons.
This is an empirical question so any post is just rampant speculation. Go spend 7000 posts arguing over how many unicorn toots it would take to fill 1000 balloons.
NikkiLikeChikki wrote:
For the love of god… 7000 posts on something completely irrelevant. Google gave up on chess and is now doing something helpful, like unlocking the secrets of protein folding. Why not expend energy on something productive, like learning a new opening. These threads are all dumpster fires.
This is an empirical question so any post is just rampant speculation. Go spend 7000 posts arguing over how many unicorn toots it would take to fill 1000 balloons.
This is an empirical question so any post is just rampant speculation. Go spend 7000 posts arguing over how many unicorn toots it would take to fill 1000 balloons.
Yeah, this thread has lost its way a long time ago. Deep Blue kickstarted a lot of research in the field of AI and so did AlphaZero. There is some research on the design of quantum chess computers. Again, any insights gained from this research in the distant future will be used to solve more complex problems. Chess serves as a "simple" enough test object/a tool for research.
"Why not expend energy on something productive, like learning a new opening."
I heard the London is really popular among chess players. Maybe its worth learning. haha 
Don’t make me block you. I blocked five London players just today!
And just remember everyone, even if chess is solved, no human will ever play into that line. There are enough openings and side variations that nobody could ever remember all the lines that are supposedly winning. Chess will be unaffected.
NikkiLikeChikki wrote:
Don’t make me block you. I blocked five London players just today!
haha don't worry. I was just joking. I don't dislike the london, but I dislike the mindset most players have when playing the london. It's shocking how many people would play this opening for their entire life. Imagine eating the same food every day.
I eat the same food most of the time. I usually have a bagel for breakfast
When chess is solved, you just know the London opening will be the main line, don’t you?
#6974
"Why not expend energy on something productive, like learning a new opening."
++ Learning a new opening is among the least productive ways to expend energy.
If solving chess is more or less productive than proving the Riemann Hypothesis is open for debate.
JuergenWerner wrote:
Will computers ever solve chess?
No. Because chess is the game of the gods/goddesses!
I guys might think that I'm trolling with the above comment but I'm serious.
#6979
God the Son: "Let us play a game of chess"
God the Father: "OK"
God the Son: "I propose a draw"
God the Father: "I agree"
btickler wrote:
Elroch wrote:
I have no reason to doubt that IBM has some 5 qubit QCs with quantum volume of 32 online. The issue is not whether these exist (or a lot more), it is whether it can be scaled up to much larger systems with adequate error correction.
It's not just about the qubits, as it would be if devices were perfect. This is why the concept of quantum volume, more indicative of computing power, was invented.
An interesting fact there is while IBM (a big player for sure) has only claimed modest quantum volumes, claims of massively higher values, indicating much more powerful QCs have been made by IonQ:
That still doesn't answer my question, though. If you can't store intermediate data, you also can't pass it...you can't make a procedural call with a parameter, nor pass back a result, you can't do iterations inside the process that require tracking any variables, you can't do object oriented programming, etc. So, solving chess would have to run in one procedure, with one output/endstate, as I understand it currently. I made a reference to the answer "42" from Hitchhiker's Guide to the Galaxy as a good analogy to that type of limited outcome, but I may have lost that post? I don't see it here.
So when they say it will run Python syntax, what do they actually mean...full Python with every bell and whistle, or just a more limited subset Python (which is already limited) using a subset of the same syntax but gutted of several fundamental capabilities?
It would not be the first time press hype would gloss over such a key distinction. The article is also less than promising for solving chess since this particular company is currently shooting for just outperforming a current PC in 2 years from now. That's quite a ways off from the millions of times faster that some people's assertions (not yours) are reliant on ...
It is a different programming paradigm, but I believe in principle entirely general. One interesting property of quantum computing is reversibility: it is essential that all quantum gates are reversible for it to work.
Elroch wrote:
It is a different programming paradigm, but I believe in principle entirely general. One interesting property of quantum computing is reversibility: it is essential that all quantum gates are reversible for it to work.
If by different paradigm you mean like Turing's Enigma breaker, that has to crank through the entire process and either succeeds in one pass or fails, then I would agree. But 10^44.5 is a large process to do in one shot 
, and it's a fundamentally iterative process evaluating backwards from mate. You cannot "evaluate" each single position standalone, that would be like solving chess another 10^44.5 times over. So you need to carry information states from one subprocess to another...ergo the problem for quantum computers.
I would liken it to a massive pattern of dominoes tipping over. One shot to do it right, no stopping, and no having certain dominoes wait for others to do their tipping. It's unidirectional in nature. The pattern falls and then you see what you got in the end. It's a tremendous limitation compared to traditional programming.
Edit: ...and sure enough, as soon as I Googled "quantum computers like dominoes", I found this...
...and this.
Maybe I should be in quantum computing. It seems like an incredibly frustrating way to program though, much like real domino pattern builders. It's well suited to massively parallel tasks, but not iterative ones.
We will see. There is already research regarding chess and quantum computers but from what I've seen so far they only made the first step out of a gazillion steps.
Whether this leads to anything substantial we don't know yet. There will be a solution in the very distant future but not with the tools we have now.
Computer is just a tool.
The Shannon Number x = 10^120
It is far to be reached.
I have no reason to doubt that IBM has some 5 qubit QCs with quantum volume of 32 online. The issue is not whether these exist (or a lot more), it is whether it can be scaled up to much larger systems with adequate error correction.
It's not just about the qubits, as it would be if devices were perfect. This is why the concept of quantum volume, more indicative of computing power, was invented.
An interesting fact there is while IBM (a big player for sure) has only claimed modest quantum volumes, claims of massively higher values, indicating much more powerful QCs have been made by IonQ:
https://arstechnica.com/science/2020/10/trapped-ion-quantum-computer-sets-new-mark-for-quantum-volume/
That still doesn't answer my question, though. If you can't store intermediate data, you also can't pass it...you can't make a procedural call with a parameter, nor pass back a result, you can't do iterations inside the process that require tracking any variables, you can't do object oriented programming, etc. So, solving chess would have to run in one procedure, with one output/endstate, as I understand it currently. I made a reference to the answer "42" from Hitchhiker's Guide to the Galaxy as a good analogy to that type of limited outcome, but I may have lost that post? I don't see it here.
So when they say it will run Python syntax, what do they actually mean...full Python with every bell and whistle, or just a more limited subset Python (which is already limited) using a subset of the same syntax but gutted of several fundamental capabilities?
It would not be the first time press hype would gloss over such a key distinction. The article is also less than promising for solving chess since this particular company is currently shooting for just outperforming a current PC in 2 years from now. That's quite a ways off from the millions of times faster that some people's assertions (not yours) are reliant on
...