more on gnutella and a good interview

[Daniel Chudnov <daniel.chudnov@yale.edu>]

First off, if you haven't seen the Lars Ulrich interview on slashdot it's
worth ten minutes of your time:

  http://slashdot.org/article.pl?sid=00/05/26/1251220&mode=thread


Second, I can report a bit back on life with gnutella.  Since I started
the thing up yesterday at around this time and loaded about 12 gutenberg
texts in my upload dir, there have been roughly 150 downloads from that
directory.  A good range of the titles I put up but mostly the hacker's
dictionary (aka jargon file), human genome project 21, x, or y, and a lot
of cia world factbooks too.

Since several folks said so I know that some folks on this list looked
specifically for these, but I am certain that many of these downloads
weren't from oss4libers.  I'm most certain of this because some of them
were already being downloaded before I posted.  So we can probably guess
that lots of folks on gnutellanets are looking for "hacker" or "cia".  No
big surprise really.

More #s (gnutella keeps some interesting stats):  52,000,000+ gnutella net
messages; 800,000+ searches of my upload dir; 72 connections to other
servents (jargon for the client+server dual roles a single gnutella plays)
right now; 200,000 errors, 16,000,000 dropped messages.

It seems very resilient.  When you're keeping a minimum of, say, eight
other machines and you drop five at once, it only takes a few more seconds
to get back up to eight.  The gnutella discussions seem full of folks
concerned with search spoofing but I haven't really seen very many
brownouts.  Or if I have it didn't look like it.  Most of my searches come
back with lots of results within ten minutes or so.  

It is very interesting to watch your search results change as other
servents connect/disconnect.  It is also very strange to watch the search
monitor.  Man folks are looking for everything.  #1 music, #2 porn, but
lots and lots of other things too.  Movies.  Software.  Books.

Hmmmm.

Interesting to see a few conventions kinda pop up.  One is that some folks
are putting up many short html files which open to ads for their porn
sites but with very commonly searched filenames like 'britney' or 'buffy'.
Another is that a handful of individuals are putting up hosts of
fairly current tech books (eg teach yourself blah blah blah), a
handful of which have filenames preceeded by '(library)'.  Then of course
there are also very short html porn ad files named 'library' too.


One thing is for certain.  This thing needs librarians.  But it isn't at
all obvious how (though we may interpret Lars' comments in some obvious
ways) to help.  But there is work for us to do...  take a look if you
haven't yet.


  -dc