1993-03-26 - Re: A New Usenet…

Header Data

From: mccoy@ccwf.cc.utexas.edu (Jim McCoy)
To: clark@metal.psu.edu (Clark Reynard)
Message Hash: 68dd37b581eeb7b97bd481c5c35d6d375162e5e2fcc8a21da6ef86cf5353a5b8
Message ID: <9303261704.AA06403@tramp.cc.utexas.edu>
Reply To: <9303260717.AA27522@metal.psu.edu>
UTC Datetime: 1993-03-26 17:05:51 UTC
Raw Date: Fri, 26 Mar 93 09:05:51 PST

Raw message

From: mccoy@ccwf.cc.utexas.edu (Jim McCoy)
Date: Fri, 26 Mar 93 09:05:51 PST
To: clark@metal.psu.edu (Clark Reynard)
Subject: Re: A New Usenet...
In-Reply-To: <9303260717.AA27522@metal.psu.edu>
Message-ID: <9303261704.AA06403@tramp.cc.utexas.edu>
MIME-Version: 1.0
Content-Type: text

> From: Clark Reynard <clark@metal.psu.edu>
> irt the idea of a new usenet, i doubt that a 'new' USEnet is possible;
> it is so firmly entrenched, by 'tradition' et cetera, that it is very
> likely that what will exist is merely an improvement and expansion of
> the existing usenet;

This was what I was thinking about.  Something that would sit on top of the
existing usenet and provide different levels of service and information.  I
realize that it would be next to impossible to replace the usenet as a
whole, what I am thinking about is extended services and newsgroups that
exist in parallel with the current usenet paradigm.  For example, a "new
usenet" server could sit on prot 119, and when it gets a connection it can
work like a normal usenet server if no special information is given to it
or it can take articles for the authenticated service if additional
commands or flags are given.  To the regular user it would just appear as a
new hierarchy (newusenet.sci.crypt, etc).  If the person connecting wanted
to post articles to the new groups it would require them to use commands or
a client that extends the commands available in the current RFC.

> [s/n ratio has increased, but we just need smarter readers...]  thus,
> rather than have kill files, having scanning programs 
> capable of filtering out particular TYPES of data, rather than the poster
> him- or her-self, will be of far more use than excluding a usenet poster
> who may very well post a greatly useful file amidst a welter of
> useless files.  thus, some sort of syntax/subject/type of data file
> scanner (which requires asi--my acronym for artificial semi-intelligence)
> is necessary to make the usenet actually new, and not merely IMPROVED--
> usenet improves constantly. 

As someone who has spent the past several years working in an AI lab, let
me tell you that this is a very, very difficult task.  The easiest method
is to do something like user-supplied tags, which is what is happening now
with the overview package and increased usage of the References line.
Don't expect the "tell me what is in this group that would interest me"
newsreaders to appear any time this decade and don't expect them to be

It is still much easier for a person to do this kind of filtering and the
current structure of usenet is designed for this.  The newsgroup naming
scheme allows readers to select groups based upon thier interests
(topic/name of the group) and moderation in groups allows filtering.  The
problem that I see now is that there are some unmoderated groups that are
too popular for thier own good.  Take comp.org.eff.talk for example.  I
would love to stay current on this group, but there is so much noise
(mostly in people repeating the same thing in 5 different subject lines...)
that I will often skip it if I don't have the time.  What would be nice
would be for a group to exist that selects articles and threads from that
group and puts them in some group like comp.org.eff.talk.best-of...

Eh, either way I will start coding....