[922] in Commercialization & Privatization of the Internet
Re: internet consumer reports on state-wide IP networks
daemon@ATHENA.MIT.EDU (Mark Moraes)
Mon Jul 8 17:48:07 1991
From: Mark Moraes <moraes@cs.toronto.edu>
To: com-priv@psi.com
Date: Mon, 8 Jul 1991 17:46:40 -0400
jsq@tic.com (John S. Quarterman) writes:
>As others have been asking lately, why haven't we seen a map of the
>whole net yet? I don't mean just the Internet, even though that's
>an increasingly large part of it. I mean the Matrix.
One possible explanation is that we have the maps (or the mapping
data), just not the technology or knowhow to represent them well to
the end user. The UUCP map, the Internet nameservers, the list of ftp
sites, comp.archives, archie, route tables, etc. represent maps of
components of the Matrix in various forms, after all.
Another explanation is that a complete map of the matrix doesn't exist
or cannot be represented. A related explanation is that's it's
changing too fast to catch a static snapshot of the data. (I assume a
map to be somewhat static by definition -- a map is a representation,
an encoding of some set of features -- my father, a cartographer by
profession, tells me that an important and hard problem in mapping is
deciding what to leave out.)
Random examples of matrix maps that indicate the number of forms maps
can take:
- A good example of a fairly static, but comprehensive map is the !%@::
book. Don't run an email facility without it.
- Brian Reid regularly puts out Usenet maps based on the UUCP map data.
- Geoff Collyer once used the results of a "sendsys tor" to draw a
Usenet map of Toronto depicting news flow.
- I've used the UofT hosts/networks files (the former is generated from
nameserver data, the latter maintained by hand) to draw maps of our
local campus networks. etc.
- I once used traceroute to draw myself a reasonable connectivity map
of places on the Internet important to me; it made the network status
reports much more meaningful.
But all this information changes and grows faster than we can put out
paper pictures (PostScript isn't all that helpful, even though we once
reverse-engineered one of Brian's maps when we urgently needed data
for a CA*net demo:-). Further, each person has their own question to
ask of the maps. The electronic forms are more pliable -- they can be
queried easily and can be cheaply updated using the Matrix itself, as
Brian Reids maps and the UUCP maps show. (also numerous network maps
available for anon. ftp) A more flexible and powerful approach is
mapping software. eg. Michael Patton's MAP program, programs that
automatically layout of graphs, network browsers that allow zoom and
filter, etc.
Mark, not a cartographer by profession, but fond of network maps.
