Bit Error Rate vs Solving Chess via Computer

[Guest]

Archived from groups: rec.games.chess.computer (More info?)

I have learned, after considerable study in the field of digital data
transmission via fiber optics, that no matter how fast, how strong, how
robust, an electronic device can become, the quantum laws of physics
require that an occasional 1 in a digital logic process will show up as
a 0. I think this might be a proof that chess will never be solved by a
computer. Anyone out there wanna argue the point? Shooting down my
reasoning in flames is OK too.

Oliver

 

[Guest]

Archived from groups: rec.games.chess.computer (More info?)

If a system is cybernetically designed, the odd 0 becoming a 1 should
make no difference. Why shouldn't a computer with states between 0 and
1 solve chess, anyway?

Mark

 

[Guest]

Archived from groups: rec.games.chess.computer (More info?)

laocmo@copper.net wrote:

>I have learned, after considerable study in the field of digital data
>transmission via fiber optics, that no matter how fast, how strong, how
>robust, an electronic device can become, the quantum laws of physics
>require that an occasional 1 in a digital logic process will show up as
>a 0. I think this might be a proof that chess will never be solved by a
>computer. Anyone out there wanna argue the point? Shooting down my
>reasoning in flames is OK too.

I will take a shot at it...

Assume that a sufficiently powerful/fast computer that is no
bigger than an average galaxy can solve chess in less than a
human lifetime.

Assume that all computers have a very, very small but still
non-zero error rate on every operation.

Argument:
The logic of the argument is that the powerful/fast computer has
to do a huge number of operations, and that number is large enough so
that the very, very small error rate multiplied by the very, very
large number of operations equals a quite large chance of having an
error before the problem is solved.


Counterargument:
Use 127 redundant computers. Every few milliseconds have them
take a vote on the partial answer; majority wins. Now you have
to have 64 identical errors to get a wrong answer. If the
increased reliablility isn't good enough, use more computers.

Note that with a 127 computer majority vote, if 43 get answer A,
42 get answer B and 42 get answer C, answer A wins the vote, so
the bad computers not only have to be in error, they have to have
the *same* wrong answer.

One would, of course, also use ECC correction on all RAM and disk
storage that the individual computers have.

Using this method, one can drive the error rate as low as one wishes,
and can thus make it low enough to allow chess to be solved.

(And yes, I do know how to make a redundant votetaker circuit...)

(And yes, I do know about more efficient error correction methods.
a majority vote is easier to explain.)

 

[Guest]

Archived from groups: rec.games.chess.computer (More info?)

* <laocmo@copper.net> (19:00) schrieb:

> I have learned, after considerable study in the field of digital data
> transmission via fiber optics, that no matter how fast, how strong, how
> robust, an electronic device can become, the quantum laws of physics
> require that an occasional 1 in a digital logic process will show up as
> a 0. I think this might be a proof that chess will never be solved by a
> computer. Anyone out there wanna argue the point? Shooting down my
> reasoning in flames is OK too.

Are you sure you learned that? Might just have been an error in you
brain.

mfg, simon .... l

 

[Guest]

Archived from groups: rec.games.chess.computer (More info?)

I took a course in "Error Correcting Codes" in grad school during the
1980s (the Prof. was also an expert in game theory

They were very efficient even back then, presumably there has also been
some progress made in the meantime (the main application back then was
correcting information sent down to earth from satellites)

They used some matrix multiplication trick if I remember correctly

 

[Guest]

Archived from groups: rec.games.chess.computer (More info?)

Eh Oliver ! You never heard from "correcting code" ?
----------
<laocmo@copper.net> a écrit dans le message news:
1117818054.773858.112600@z14g2000cwz.googlegroups.com...
> I have learned, after considerable study in the field of digital data
> transmission via fiber optics, that no matter how fast, how strong, how
> robust, an electronic device can become, the quantum laws of physics
> require that an occasional 1 in a digital logic process will show up as
> a 0. I think this might be a proof that chess will never be solved by a
> computer. Anyone out there wanna argue the point? Shooting down my
> reasoning in flames is OK too.
>
> Oliver
>

 

[Guest]

Archived from groups: rec.games.chess.computer (More info?)

<laocmo@copper.net> wrote:
> I have learned, after considerable study in the field of digital data
> transmission via fiber optics, that no matter how fast, how strong, how
> robust, an electronic device can become, the quantum laws of physics
> require that an occasional 1 in a digital logic process will show up as
> a 0. I think this might be a proof that chess will never be solved by a
> computer. Anyone out there wanna argue the point? Shooting down my
> reasoning in flames is OK too.

No, it doesn't prove that. Any digital system is susceptible to error
(cosmic rays hitting things is a much more frequent occurrence than
quantum fluctuations) but error-correcting codes can take care of this, in
the sense that there are ECCs that will deliver any desired error rate.

I think it's fair to say that most people who have thought about it
believe that chess cannot (in practical terms) be solved by computers as
they currently exist. The time and memory requirements are just too
large, even before you start to consider the need for error correction.


Dave.

--
David Richerby Crystal Pickled Laser (TM): it's like
www.chiark.greenend.org.uk/~davidr/ an intense beam of light but it's
preserved in vinegar and completely
transparent!

 

[Guest]

Archived from groups: rec.games.chess.computer (More info?)

David Richerby wrote:

>I think it's fair to say that most people who have thought about it
>believe that chess cannot (in practical terms) be solved by computers as
>they currently exist. The time and memory requirements are just too
>large, even before you start to consider the need for error correction.

Emhasis on the "as they currently exist" part...

 

[Guest]

Archived from groups: rec.games.chess.computer (More info?)

Guy Macon <_see.web.page_@_www.guymacon.com_> wrote:
> David Richerby wrote:
>> I think it's fair to say that most people who have thought about it
>> believe that chess cannot (in practical terms) be solved by computers
>> as they currently exist. The time and memory requirements are just too
>> large, even before you start to consider the need for error correction.
>
> Emhasis on the "as they currently exist" part...

Those four words were included mainly with you in mind, Guy.


Dave.

--
David Richerby Electronic Sadistic Toy (TM): it's
www.chiark.greenend.org.uk/~davidr/ like a fun child's toy but it wants
to hurt you and it uses electricity!