Skip to Content.
Sympa Menu

internetworkers - Re: [internetworkers] 20:02, 11-11, 2002

internetworkers AT lists.ibiblio.org

Subject: Internetworkers: http://www.ibiblio.org/internetworkers/

List archive

Chronological Thread  
  • From: "Michael D. Thomas" <mdthomas AT mindspring.com>
  • To: <internetworkers AT lists.ibiblio.org>
  • Subject: Re: [internetworkers] 20:02, 11-11, 2002
  • Date: Wed, 13 Nov 2002 00:54:34 -0500

> > I like it, assuming you can do a db dump of such a monster. I expect that
no
> > one has bothered to implement that function fully. :-)
>
> Why do a db dump? Just take an image of the filesystem. :)

Assuming that the terraserver is on one file system. It looks like it is
partitioned across several, though the wording is a bit ambiguous on that
point and I'm better versed in OracleSpeak than SQLServerSpeak:

http://terraserver.homeadvisor.msn.com/About/AboutDbstats.htm

(From this, it also looks like a lot of the data is hosted here in Raleigh.)

> I was -- the db file's "real value" is being the db. Granted, it isn't a
> human-readable string, but what human is going to sit down and read in
> entirety the best approximation of pi?

That is the point that is ticklish... and which makes the problem more
complicated than determining the largest ball of twine. Like twine, given any
two strings, one is longer or both are the same size. Thus, it's obvious that
there is some (small) set of strings of equivalent length that are longer than
all other strings. If this were Hollywood, there would only be one string in
the set: The King Of The Strings!

In some ways, it should be easier to determine the longest character string.
Unlike twine, a character string is always the same width and thickness.
Guiness doesn't have to send someone out to approximate the length by taking
the circumference of a big ball of character string.

The real work is in constraining the problem. You either eliminate the
strings that aren't interesting, or take a lesson from the sport of boxing and
divide the strings in to different classes so that the fight is fairer. First
to go: strings of infinite length. It's trivial to create formulas that derive
strings of infinite length. Instead, I think the focus should be strings of
finite length.

With infinite strings dismissed, the top class would probably be theoretically
derived finite strings. My contender -- a string representation of all the
sub-atomic particles in the universe, with each description describing the
relationship with every other sub-atomic particle in the universe, including
distance. In binary, of course.

(Oh yeah, I think you always render strings in binary, since, given any set of
things -- in this case, the alphabet of the string -- a member of such set can
be represented uniquely in binary notation. Besides, it just isn't fair for
non-binary strings to compete with binary strings. Reminds me of being in
elementary school and having some looser chosen for the team instead of me,
just because he was binary. Sucked. I can be binary, too, ya know.)

Depending on your philosophical bent, you could imagine a Turing Machine
representation of the universe. A Turing Machine is a long string that
represents a computational machine, so if you believe the universe is a big
computational machine, then the Turing Machine representing the universe would
be pretty long. But if you also believe that the universe is infinite, then I
suspect that your Turing Machine representation would be infinite in
length. If so, your universe wouldn't be allowed in our contest. Sorry. But
please don't be tempted in to changing your belief system just so you can win
a silly contest :-) We simply can't be liable for the damage this may cause.

But concocting theoretical finite strings is largely a parlor game -- it's
not like these strings physically exist, right? It's not like they are part
of our real world, right? This is where I propose the class of "naturally
occurring strings," one of whose members would be OED. But the next problem is
what does it mean for a string to "physically" exist? It could be the largest
physically existing document. But since we're computer folks, why not include
digitally stored strings in the study?

However, it's complex to constrain the class of digitally stored strings. For
instance, is it important to actually be able to "see" the string? If you have
some yottabyte database but no function exists to output the db one
character at a time, does it count? If you get in the business of writing
functions to create long finite strings, then how "natural" is
that?

Once you start writing functions, the Internet may be the longest character
string in it's class, assuming a simple robot that knows how to traverse the
Internet. But as we all know, that particular string would change radically in
length by the time such a robot completed. Considering that the Internet tends
to grow in size over time, your string would certainly not be the longest
string at the time that your robot finished (if it finished). If you wish to
declare that your robot runs instantaneously and submit the size of your
string as the estimate of bits on the Internet, then you've landed back in the
theoretical space and must complete with the likes of the heavyweights listed
above.

If you constrain the class to static strings -- or perhaps, "strings that
could remain static for the period of time it would take to render the
string" -- then I guess you look for the largest disk drive, file system or
database, depending on how natural you want to be. Which brings me back to the
original point -- do we further constrain this class to "strings that could
remain static for the period of time it would take to render the string AND
are already capable of being rendered to an output device"?

Now I must stop, lest I'm accused of attempting to win the longest string
contest with this discourse.;-)




















Archive powered by MHonArc 2.6.24.

Top of Page