Date: Sat, 17 Dec 94 11:10:52 -0500 (EST)
From: Jerry Bryan <BRYAN@wvnvm.wvnet.edu >
~~~ Subject: How Big is Big?

Some of the notes in the last day or two about whether or not ten levels
deep is too large to search reminded me of a note I have been meaning
to send for a long time. Just how big is 4.3*10^19, and can we ever
hope to search it all?

First of all, 4.3*10^19 is really about 10^18. That is, we could safely
confine ourselves to searching M-conjugate classes, and there are about
0.9*10^18 classes, which we might as well call about 10^18. But how big
is that?

Suppose were trying to buy enough disk space. I claim that you
could store each position in a byte with clever indexing. Actually,
you could store each position in 5 bits, or 5/8 of a byte, but leave
it as a byte per position. Let's say that you can purchase
10^3 USD). (We are buying good quality used disks for mainframes
for about 1,000 USD per gigabyte; new prices are closer to 4,000 or
5,000 USD per gigabyte. Both SCSI and IDE disks for the desktop,
PC or UNIX, are just now down to around 500 USD per gigabyte, and I
have seen firesale type prices closer to 300 USD per gigabyte).

At 10^3 USD per 10^12 bytes, the cost would be 10^9 USD per 10^18
bytes. Well, 10^9 USD is a lot of money, but it is a lot less
than the cost of going to the moon, or the cost of an aircraft carrier.
In fact, Bill Gates could afford it if he so chose.

There are other ways to think about the problem. The size of
chess is about 10^75 states, and Go is about 10^120 states. The
standard 3x3x3 Rubik's cube is vastly smaller than either of these.
In fact, Go (and maybe chess, I can't remember for sure) is usually
described as being bigger than the universe.

A handy number in these types of comparisons and in determining "how big
Avogadro's number is the number of molecules (or atoms, for substances
which occur atomically) in the gram molecular weight of a substance.
For example, molecular hydrogen has a molecular weight of 2, so
2 grams of hydrogen contain 6*10^23 molecules. Iron is atomic with
an atomic weight of 56, so 56 grams of iron contain about 6*10^23
atoms. If you had 56 grams of iron, and if you could store magnetically
each cube position in no more than 6*10^5 iron atoms, then you could
store the whole Rubik's cube.

By comparison to the size of the universe, the mass of the sun is
about 10^30 grams, consisting mostly of atomic hydrogen, so there
are about (10^30)*(10^23)=10*53 hydrogen atoms in the sun. I can't
remember for sure, but I think there are about 10^11 stars in the
Milky Way. If the sun is typical star, that would leave about
10^64 hydrogen atoms in the Milky Way. I don't know how many galaxies
there are, but we are clearly getting close to the size of Chess
at 10^75 being about the same as the size of the universe, and of Go
at 10^120 being much larger than the size of the universe. Rubik's
cube is small potatoes.

A couple of more items: the human genome is being mapped. I cannot
remember the exact size of the problem, but I do remember when I
read about it that it was a larger problem than Rubik's cube. Finally,
the Chronicle of Higher Education had an article in the last few weeks
send hundreds or thousands of magnetic tapes to each other via standard
mail (snail mail to E-mail folks -- but mailing magnetic tapes can
yield tremendous data transfer rates if you actually calculate bytes
per second). According to the article,
the physicists are already sending gigabytes over the Internet. They are
planning soon to start sending petabytes (10^15) over the Internet.
10^15 is getting interesting close to the size of Rubik's cube
(never mind that I thought that the proper term for 10^15 bytes was
terabytes.)

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
Robert G. Bryan (Jerry Bryan)                        (304) 293-5192
Associate Director, WVNET                            (304) 293-5540 fax