Did you know that a human body in dna stores more data than the entire internet which is well over a zettabyte
Will computers ever solve chess?
Sort:
May 7, 2016
0
#303
0110001101101000 wrote:
DragonPhoenixSlayer wrote:
0110001101101000 wrote:
Some people think computers will get faster forever. To them chess is just a "big number" and computers "can do big numbers eventually."
But if "eventually" isn't 100 years, but instead 1 million years, it might as well be impossible. In any case, computers can't continue to double in speed forever.
At some point you would to make them bigger instead of better
Ok.
How many times do you suppose we could double the size of a computer until it's as big as a building? 10 times? 15 times? Eventually that also must end.
Computers used to be as big as buildings. They weren't faster. The limiting factor is the speed of light. The electricity actually has to flow across the transistors. Ideally, they are very small, making that problem minimal. Then the limiting factor becomes the size of atoms and quantum uncertainty about the position of electrons. The current smallest size of transistor in commercial production is about 14nm, 14 billionths of a meter. Most engineers think the smallest possible is between 3nm and 5nm, due to reasons having to do with the structure of atoms, and therefore not solveable, ever. Some think 1nm may be possible. Regardless, that's only a few more doublings of what we have now, so the current path is ending, at least for what we currently call computers (Von Neumann architecture, etc.). While other approaches may be possible, then again they may not.
May 7, 2016
0
#304
DragonPhoenixSlayer wrote:
Did you know that a human body in dna stores more data than the entire internet which is well over a zettabyte
I didn't know that. Looking it up one site estimates as much as 150 Zettabytes. (This seems like too much, but I'll just go with this estimate). I had to look up that a zettabyte is 10^21 bytes.
The number of possible positions in chess is, lets say, 10^40
With each position, you will also need to store the evaluation (win, loss, or draw).
You will of course need much more, but lets say you can store 1 position with 1 byte, and also 1 evaluation with one byte.
So plugging in these numbers real quick, it seems you'll only need 10^17 humans. These numbers are very generous, so you'd actually need many more. But still, 10^17 humans would be a planetary sized object lol. (edit, ok, maybe not that big.)
And then how would you even retrieve this data?
May 7, 2016
0
#305
RCMorea wrote:
0110001101101000 wrote:
DragonPhoenixSlayer wrote:
0110001101101000 wrote:
Some people think computers will get faster forever. To them chess is just a "big number" and computers "can do big numbers eventually."
But if "eventually" isn't 100 years, but instead 1 million years, it might as well be impossible. In any case, computers can't continue to double in speed forever.
At some point you would to make them bigger instead of better
Ok.
How many times do you suppose we could double the size of a computer until it's as big as a building? 10 times? 15 times? Eventually that also must end.
Computers used to be as big as buildings. They weren't faster.
Yeah, but modern computers the size of a building would be faster 
May 7, 2016
0
#306
Supercomputers are quite big these days, and require huge amounts of power.
You can always buy yourself one, with plenty of money it's all about charisma, all the way down.
If you have enough money you don't need charisma to get things done
But you do need money to get things undone.
Or charisma
RoepStoep wrote:
Or charisma
with money, in this case, you do need a frack, a wheel and plenty of money, needless to say an insider as well, but you can always buy me a drink, get politics into play, you do love your contry, right
I have sold my country, basically, keep your junk out. The difference between stupidity and genius is that genius has its limits.
RCMorea wrote:
Therefore, even if computers get a billion, billion times faster (a billion billions), they can not solve chess.
And they will still take half a minute to boot up. 
May 7, 2016
0
#315
Let's argue that chess is a draw.
Let's further argue that we can create a computer that plays perfectly 100% of the time.
We would just have the engine play against itself and that would prove the draw.
To test the draw, you wouldn't need to store every possible move and their evaluation. You could just have the ability to input every game for evaluation. The computer will demonstrate where every mistake was made.
There will never be a decisive game without an error. To prove that statement, you don't have to play every conceivable game, you just need an engine strong enough to play perfect chess. We're fairly close to such a machine.
s23bog wrote:
I knew someone who worked closely with designers of some of the first computers. They told him that they designed it with 1 kilobyte of memory. They thought about making it with 2 kB, but said, "Who would ever need that much?!"
Some of the first computers had quite small memories. The person in question was perhaps not aware of the huge amount of opening knowledge, let alone endings and calculations in general, in almost any given position. He must have been a novice in chess, if that.
May 7, 2016
0
#317
Computers used to be as big as buildings. They weren't faster.
Yeah, but modern computers the size of a building would be faster
True. In fact, it's not a matter of "would be." They are. But they still are limited by the speed of light and the fact that you need to make transistors at least several atoms across. Bottom line, we are running up against these limits now, and they are limits based on the structure of the universe, not limits that we can solve with cleverness.
The point people aren't focusing on, the really interesting one in my opinion, is quantum computers (if they are indeed possible). A regular computer depends on data stored as bits, either 0 or 1. Two bits can have a total of 2 squared states. Three is 2 to the third. A gigabit, is a billion bits, that's 2 to the billion possible states...it's an insanely large amount of possibilities (but still much too small to hold chess). But, a quantum computer would try to store information in electrons. Then, it would take advantage of quantum uncertainty, and the fact that electron position and spin isn't just unknown, but actually undetermined until observed, and therefore neither at once or both at once. It's conjectured that this would allow all possibilities to be examined at the same time, instantly, assuming you could set it up in the first place. Not necessarilly possible, but pretty cool if it turns out to be possible.
May 7, 2016
0
#318
SmyslovFan wrote:
Let's argue that chess is a draw.
Let's further argue that we can create a computer that plays perfectly 100% of the time.
We would just have the engine play against itself and that would prove the draw.
To test the draw, you wouldn't need to store every possible move and their evaluation. You could just have the ability to input every game for evaluation. The computer will demonstrate where every mistake was made.
There will never be a decisive game without an error. To prove that statement, you don't have to play every conceivable game, you just need an engine strong enough to play perfect chess. We're fairly close to such a machine.
As long as you can prove "this machine plays perfect chess" it's fine.
How do you propose doing this without having played every position though?
May 7, 2016
0
#319
0110001101101000 wrote:
SmyslovFan wrote:
Let's argue that chess is a draw.
Let's further argue that we can create a computer that plays perfectly 100% of the time.
We would just have the engine play against itself and that would prove the draw.
To test the draw, you wouldn't need to store every possible move and their evaluation. You could just have the ability to input every game for evaluation. The computer will demonstrate where every mistake was made.
There will never be a decisive game without an error. To prove that statement, you don't have to play every conceivable game, you just need an engine strong enough to play perfect chess. We're fairly close to such a machine.
As long as you can prove "this machine plays perfect chess" it's fine.
How do you propose doing this without having played every position though?
Yeah, true. If we just assume that we have a machine that plays perfect chess, that is like saying "Let's assume we have the solution. And therefore, we have the solution." The statement "we are fairly close to having such a machine" is at least dubious. Probably false.
May 8, 2016
0
#320
s23bog wrote:
True. In fact, it's not a matter of "would be." They are. But they still are limited by the speed of light and the fact that you need to make transistors at least several atoms across. Bottom line, we are running up against these limits now, and they are limits based on the structure of the universe, not limits that we can solve with cleverness.
The point people aren't focusing on, the really interesting one in my opinion, is quantum computers (if they are indeed possible). A regular computer depends on data stored as bits, either 0 or 1. Two bits can have a total of 2 squared states. Three is 2 to the third. A gigabit, is a billion bits, that's 2 to the billion possible states...it's an insanely large amount of possibilities (but still much too small to hold chess). But, a quantum computer would try to store information in electrons. Then, it would take advantage of quantum uncertainty, and the fact that electron position and spin isn't just unknown, but actually undetermined until observed, and therefore neither at once or both at once. It's conjectured that this would allow all possibilities to be examined at the same time, instantly, assuming you could set it up in the first place. Not necessarilly possible, but pretty cool if it turns out to be possible.
You claim we are running against hard limits imposed by the Universe. I would like to offer a correction to that, if I may.
We are running against hard limits imposed by our understanding of the universe using the method we are currently using.
Rather than looking at a computer as a single device, it may be useful to look at the cloud of computers on the internet as a whole. There are many users, and many processors, but there is great disconnect between the individual nodes. Data sharing is extremely limited, and there are a bunch of unruly people at the controls.
No, respectfully, the cloud isn't going to help in any material way, even if the whole world cloud were devoted to nothing but chess. Linear increases in power, such as adding a billion processors or a trillion processors, can never defeat problems which are exponentially too large. That is, 10^120 positions (Claude Shannon's estimate of the size of chess), is not even scratched by application of a trillion times more power even over a period of a trillion years.
Meanwhile, yes, the hard limits of the universe as we understand them. But if we happen to be wrong about everything, it becomes a philosophy discussion in which there is no point discussing anything, since we have no framework for doing so. Provided we must make our computers out of atoms, provided we can't really know the exact locations of electrons, provided the speed of light is constant--that sort of thing--then chess is not going to be solved by computers.
Some people think computers will get faster forever. To them chess is just a "big number" and computers "can do big numbers eventually."
But if "eventually" isn't 100 years, but instead 1 million years, it might as well be impossible. In any case, computers can't continue to double in speed forever.
At some point you would to make them bigger instead of better
Ok.
How many times do you suppose we could double the size of a computer until it's as big as a building? 10 times? 15 times? Eventually that also must end.