Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!utgpu!water!watmath!clyde!rutgers!husc6!seismo!epiwrl!epimass!jbuck
From: jbuck@epimass.UUCP
Newsgroups: news.misc
Subject: Re: The USENET Backbone (Last changed: 30 April 1987)
Message-ID: <1127@epimass.UUCP>
Date: Tue, 5-May-87 12:33:22 EDT
Article-I.D.: epimass.1127
Posted: Tue May  5 12:33:22 1987
Date-Received: Thu, 7-May-87 02:10:33 EDT
References: <15219@gatech.gatech.edu> <3753@oddjob.UChicago.EDU>
Reply-To: jbuck@epimass.UUCP (Joe Buck)
Organization: Entropic Processing, Inc., Cupertino, CA
Lines: 30

In article <3753@oddjob.UChicago.EDU> matt@oddjob.uchicago.edu (Sherlock, the keen-eyed detective) writes:
>In article <15219@gatech.gatech.edu> usenet@gatech.edu writes:
>) [Most recent change: 30 April 1987 by spaf]
>) 
>) A Usenet "backbone" site is one which exchanges every (non-local) news
>) article it receives with at least two other backbone sites; or which is
>
>After a couple unanswered mail messages to Gene I still don't
>understand the application of this definition.  If all the links
>in a chain like
>		ihnp4 -- x -- y -- hao
>are low-delay and full feeds, are x & y then backbone sites?
>Is the answer different in a case like ihnp4 -- z -- hao?

I wrote a set of programs that perform analysis on Path: headers
(available from mod.sources, er, comp.sources.unix).  Among other
things, there's an awk script called "closure" that, roughly,
implements Gene's definition of a backbone site.  In Matt's example
above, it would add x, y, and z to the backbone if the given strings
appeared more than a certain number of times in the paths.

The result, which has been consistent every time I've run it, is that
the "real backbone" -- the sites most responsible for circulating the
news -- is about twice as big as "Gene's backbone", and much more of
it is on the Internet than is shown in Gene's map.


-- 
- Joe Buck    {hplabs,ihnp4,sun,ames}!oliveb!epimass!jbuck
	      seismo!epiwrl!epimass!jbuck