Solving Chess Project

General discussion about computer chess...
Post Reply
DustinYoder
Posts: 4
Joined: Wed Jul 13, 2011 3:58 am
Real Name: Dustin Yoder

Solving Chess Project

Post by DustinYoder » Wed Mar 26, 2014 1:37 am

I wanted to announce that I have started a kick starter project to build and test a database structure to handle solving chess. I have discussed solving chess here and I think the community might be interested to read the details here https://www.kickstarter.com/projects/12 ... tep-1-of-3 I would request comments and advice on other outlets who might be interested in this project.

User923005
Posts: 616
Joined: Thu May 19, 2011 1:35 am

Re: Solving Chess Project

Post by User923005 » Wed Mar 26, 2014 9:13 pm

Re: "This is not a fast database, but one that could store positions using only 64 bits and without indexing or hashing limitations."
You cannot store a chess position using only 64 bits.

Re: "We only need to analyze possible LEGALLY reachable board positions"
Than number is about 2^155 (see http://homepages.cwi.nl/~tromp/chess/chess.html)
In decimal that is 4.5671926166590716193865151022384e+46 positions.

Now, let's suppose that you could store a chess position in a single byte.
And let's suppose you could store all the information you need about the value of the position in a single byte.
Your chess database would be 4.5671926166590716193865151022384e+34 terabytes for the positions and the same volume for the answers (9e34 bytes total).
Let's go further and imagine that you can store both the answer and the position in a single bit.
The storage for your database is 5.708990770823839524233143877798e+33 Terabytes.
At roughly $100 per terabyte for cheap jbod storage, that is $5.708990770823839524233143877798e+35 dollars to store the data.
So I would estimate that you had better charge more for participation, and figure out how to get both the board position and the answer in a fraction of a bit.
Bon chance, and happy sailing in your sea of data.

BTW, I agree that chess will be solved some day. However, I have serious doubts about your methods, as described above.

User923005
Posts: 616
Joined: Thu May 19, 2011 1:35 am

Re: Solving Chess Project

Post by User923005 » Wed Mar 26, 2014 11:32 pm

Assuming that Moore's law holds indefinitely, chess would be solved by 2250 AD:
http://home.telfort.nl/jaheruddin/f/Che ... solved.pdf

That's a fairly large if, because at some point, electrons might have to go faster than light and bit storage density would have to be smaller than the smallest elementary particle.
But, as Inigo says, "Don't bother me with trifles. There will be blood tonight!"

DustinYoder
Posts: 4
Joined: Wed Jul 13, 2011 3:58 am
Real Name: Dustin Yoder

Re: Solving Chess Project

Post by DustinYoder » Thu Mar 27, 2014 2:08 pm

I am not storing positions in the database only hard disk addresses. The organization of the data will reveal the board position. This is an idea that I think people will have to wrap their minds around and I'd like to build the first part that is required to build the actual database in step 2. So if we want to understand this database structure let's get step 1 done quickly and get on to the better part of step 2.

User923005
Posts: 616
Joined: Thu May 19, 2011 1:35 am

Re: Solving Chess Project

Post by User923005 » Thu Mar 27, 2014 8:24 pm

Here is an idea I would like you to wrap your mind around:
How much storage is needed for 4.5e46 positions?
Let's say that the above estimate is one million times too large, and the actual number of addresses needed is ONLY 4.5e40.
Obviously, a 32 bit address is not large enough, so 64 bit addresses would be required.
That volume of data is 4.5e40 * 8 = 3.6e+41 bytes.
Now, a terabyte is 1e12 bytes, so lets see how many TB that is:
3.6e+41 / 1e12 = 360000000000000000000000000000 TB
Pray tell, how much will it cost you to store that many TB of addresses. And that figure assumes that the best proven limit of chess position counts is a million times too large.

I am beginning to think you are either a troll or a nut case.

Now, let's neglect the money, disk, etc. to store it (consider also the KWH of electricity to power those disk drives -- that part alone would exceed the GNP of the USA)...
Let's assume that you can solve a chess position in a single cycle for won, lost, or drawn.
How long will it take a billion computers to solve those positions, if they can solve a position in one cycle?

User923005
Posts: 616
Joined: Thu May 19, 2011 1:35 am

Re: Solving Chess Project

Post by User923005 » Thu Mar 27, 2014 8:31 pm

(Luke 14:28-30) 28 For example, who of you wanting to build a tower does not first sit down and calculate the expense to see if he has enough to complete it? 29 Otherwise, he might lay its foundation but not be able to finish it, and all the onlookers would start to ridicule him, 30 saying: ‘This man started to build but was not able to finish.’

User923005
Posts: 616
Joined: Thu May 19, 2011 1:35 am

Re: Solving Chess Project

Post by User923005 » Thu Mar 27, 2014 8:34 pm

Please note that I am not suggesting that your exercise is a waste of time.
I think it is interesting and valuable to try to solve chess positions for won/loss/draw, and the root position would be the most interesting of all.
However, I think you need to sit down and do a little math, and figure out what a realizable goal might be and aim for that instead.
Or, at least, change your stated goal from "solving chess" to "analyzing relevant chess positions for solution" or something that is feasible.

IMO-YMMV

vittyvirus
Posts: 9
Joined: Sat Jun 21, 2014 12:52 pm
Real Name: Syed Fahad

Re: Solving Chess Project

Post by vittyvirus » Tue Jul 01, 2014 4:47 pm

Don't underestimate the money, effort and electricity you're about to need.
There are more chess games than atoms in universe..
anyway, good luck!

Post Reply