Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!ucsd!ogccse!blake!milton!max!wcn
From: wcn@max.u.washington.edu (W C Newell Jr)
Newsgroups: news.admin
Subject: Re: Is USENET stagnating? (long)
Message-ID: <9499@max.u.washington.edu>
Date: 31 Oct 89 10:19:16 GMT
References: <40056@looking.on.ca> <1989Oct30.040015.3272@alembic.acs.com>
Organization: University of Washington, Seattle WA
Lines: 289

WARNING:  Some opinions expressed herein touch on highly sensitive issues and
may be offensive to readers.  Private e-mail replies are welcome, but malicious
flames will be forwarded to site managers.  Since I am a University of
Washington employee, standard disclaimers apply.
 
> From: brad@looking.on.ca (Brad Templeton)
>
> [background information deleted]
>
> The only significant thing I can point to of late is NNTP.  [...]
> As a transmission mechanism, it has not had a direct affect.  Instead, its
> effect comes in tandem with NSF's funding of the internet, and the vastly
> reduced transmission times that come with it.
 
NO, NO, NO! :)  It is painfully obvious that very few people on the UUCP side
understand how the Internet works.  NSF funds the high-speed backbone, a set of
T1 links which connect the regional academic subnetworks and the NSF-sponsored
supercomputer centers.  However, the subnets are funded by the participating
universities and affiliates.  The UW is part of NWNET, and happens to be the
connection point between the NSFnet backbone and the NWNET ring.
 
It is certainly true that NNTP allows accelerated receipt/transmission of the
feeds for sites on the NSFnet backbone or one of the regional subnets.  The
protocol doesn't really matter, however.  The speed advantage will last only as
long as there is excess bandwidth, which may not exist beyond the next few
years.  In addition, big sites such as ours tend to place their large, general
purpose systems a hop or two behind the gateway, to facilitate local sub-
netting, and there may be delays on these hops which make them behave more like
well-connected UUCP nodes in terms of the user's perceived turnaround time.
 
NNTP >has< had a major effect as a transmission mechanism because of the ever-
increasing connectivity to the Internet.  We are now providing Usenet access to
the entire campus population, a total of 50,000 students, faculty and staff.
Other major universities are in the process of doing the same thing.  We are
talking about a potential audience of 10M people just on the academic domains
within the US.  This has never been practical to implement using UUCP links
alone.
 
The consequence of this is that academia has now recognized Usenet as a
production news and conferencing service, and is looking at its viability for
various applications related to the instructional and research missions of
higher education.  The future of Usenet will be dramatically influenced by the
success or failure of these activities, as judged by the academic community.
I will elaborate on this point, below.
 
> So what is new?  I like to think I'm doing new things, with stuff like
> an ultra-moderated newsgroup, electronic publishing, fancy feed/filtering
> tools and classification tools.  But I can't deny that many of these
> things are not truly new.  Hell, you could read UPI on "The Source" when
> it opened up in 1978.  Better article classification has existed on
> other nets for some time, too.
 
Technical issues in the field of information science are of great interest to
me, but I will pass on the subject for the time being, so as not to obscure
the central thrust of this reply.
 
> The only thing that really distinguishes USENET these days is the
> strongly moderated group.  Most other nets don't have that, or don't
> use it much.  (No surprise, on things like CIS the people pay to post,
> and hardly want to write things to have a moderator toss it away.)
 
Again, NO.  The Usenet model does provide for editing/moderation, but the vast
majority of general-interest newsgroups don't use it!  Commercial services such
as CIS actually have a greater level of moderation built in, because many of
their customers are willing to pay to have someone monitor the input and keep
out the flakes.  (This doesn't work in certain areas, for reasons beyond the
scope of this discussion.)  There are many hundreds of moderated discussions on
the Internet, of course, that exist as open-subscription mailing lists.  Usenet
has not begun to address the task of incorporating these services into a single
entity.
 
> Much of the "news" on USENET these days isn't about revolution in
> computer networking.  It's about whether aquariums are hobby or science.
> Or about what new tweak should be made to newsgroup creation rules.
 
You obviously don't read the same newsgroups I do.  A lot of useful information
is being exchanged in comp.*, some groups in sci.*, and even rec.games.misc.
(I also think rec.humor.funny deserves passing mention, since I use it in my
networking classes.)  I sense, by reading between the lines, that you are
dissatisfied with the loss of decorum in news.* and are worried about the
implications that has to future Usenet administration.  I agree.  The NNTP side
of Usenet is not going to embroil itself in such trivial debates, however...
 
> I do know one reason, and it comes from my own experience as (I think) one
> of the few people to try to do completely new things in this network
> environment.   You get pointlessly flamed whenever you try to innovate.
 
It's advantageous to follow this line of thinking a few steps further.  The
flaming is not the problem in and of itself.  People argue in every conceivable
forum, and human nature dictates that at some point it will start getting
personal.  Other forums maintain better order by enforcing a clearly-defined
cutoff point.  As examples, consider the letters column in your favorite
journal or your local newspaper.  Do they print every crank letter they
receive?  Of course not, and the reasons are obvious.
 
The old-guard UUCP community, which currently holds administrative control over
Usenet, has made a conscious decision to allow more-or-less unconstrained
public access, with very few rules and essentially no enforcement of those that
do exist.  This has allowed Usenet to evolve into an amazingly diverse forum
being read by an even more amazingly diverse audience.  The extremes can be
seen quite clearly when looking at recent newsgroup creation proposals and the
discussions thereof.  There is a price to pay for this openness, however.  Some
of your best new ideas may be lost because the innovators take a look at the
system as a whole and dismiss it as inherently umanagable.
 
I would personally like to propose the creation of a whole set of new groups of
particular relevance to the academic applications, things like
 
    comp.math                           edu.*
    comp.math.mathematica               edu.physics.problems
    comp.math.benchmarks                edu.chemistry.problems
    comp.stat                           edu.math.problems
    comp.stat.*                         edu.logic.problems
                                        (etc...)
    sci.information
    info.conferences
    info.proceedings
    info.abstracts
 
But I can't see how to get there from here.  The newsgroup creation process
discourages experimentation.  There is some risk that radical elements (read:
idiots) will sabotage our efforts before they are given time to take root.  Can
such groups survive unmoderated, or will they sink to the lowest common
denominator and then be abandoned by those who seek a true academic forum?  Who
would be willing to moderate such groups knowing that it may mean accepting
personal abuse from complete strangers who have no stake in the academic
mission?  What leverage is there to deal with troublemakers?
 
I teach classes at this center on how to use e-mail and other network services,
including Usenet, and I see wildly enthusiastic reactions from faculty and grad
students toward its possibilities.  In many cases, however, this is quickly
tempered by the user's first hands-on experience.  It starts with a list of
unsubscribed groups in alphabetical order, with alt.sex and alt.sex.bondage on
the very first screen.  A typical continuation is to post a simple question in
comp.whatever, which generates perhaps four e-mail replies, one with useful
information and three flames for wasting net bandwidth.  Since our users tend
to be of moderately high intelligence, they quickly assume that the net is
dominated by a small subset of users with no academic or corporate affiliation,
little or no accountability, and in certain cases exposing emotional problems.
They are thus unable to take Usenet seriously.
 
It doesn't help that we have seen this model evolve on our own campus-wide
bulletin board system with parallel results.  As soon as the undergraduate
student population was given free access, they immediately set up discussion
topics entitled "Satan", "Great*Sex", "Anal*Sex", and other non-academic
(perhaps 'immature' is a better word) diversions.  There are an assortment of
networking 'addicts' who frequent the center, some of whom will tie up ports
for 12-14 hours/day doing nothing but read the local bulletin boards and
Usenet.  I ask myself, "where is the educational value in that?"  We've even
come close, once or twice, to legal actions alleging libel and/or sexual
harassment, and one wonders when and where the first big Usenet scandal along
these lines will strike.
 
The bottom line is that Usenet is now visible to a much larger audience, one
which has expectations of the network as well as new ideas to help bring them
about.  Usenet must come to terms with the fact that it is no longer an
experiment, or a club for state-of-the-art technologists, but rather a public
service catering to disparate interest groups with conflicting needs.  This is
a prerequisite to realizing major progress along the lines we are discussing.
 
> I want to find out why we aren't
> watching *major* changes, and why we aren't watching a lot of them.
> And what we can do about it.
 
OK, since you asked...
 
1)  It should be possible to add/modify/rename groups by a consensus of expert
users rather than by a popular vote.  The time has come for the creation of
some official Usenet by-laws, and in particular for the establishment of a
governing body with elected representatives from the various constituent
groups.  This body should have ultimate authority on administrative issues such
as the naming of groups, the status of moderators and the certification of
software updates.
 
2)  The major security holes now in place must be dealt with ASAP.  Our users
do not appreciate having to repeatedly refuse subscriptions to bogus groups
such as alt.sex.bestiality.  Forgeries may eventually become a real problem;
this may require some enhancements to the protocols.
 
3)  Site managers have to be impressed with the importance of ensuring that
their local users adhere to Usenet guidelines and reasonable net etiquette.  We
could probably split off a heated debate on how exactly to go about this.  As a
corollary, major sites such as ours may need some legal protection in actions
which stem from libelous or obscene material we may receive as part of the
feeds.  This would suggest the development of a "rights and responsibilities"
document of some form, which would be signed by all participating sites as a
condition of membership.
 
4)  Sites with problem mailers (e.g. Portal) should be tarred and feathered.
 
4a) Seriously, there needs to be a concerted effort to get more UUCP sites to
register with valid domain addresses and to upgrade their mailers.  This will
greatly improve the reliability of Usenet as perceived by the new user.
 
5)  Every established newsgroup should have a moderator and an archive site, or
even more than one of each for very active groups.  The moderator does not have
to clear each posting in advance (although this may be desirable for certain
applications), but should have the ability to cancel, route to a different
group and, most importantly, mark for preservation.  There should be a means
for sites to identify valuable postings and keep them on-line indefinitely, so
the often-asked questions and answers don't have to be reposted over and over
again, especially since most major sites can't commit disk space for more than
two weeks of feeds at a time.  The lack of knowledge of what has gone before is
probably the most intimidating aspect of Usenet to new users.
 
6)  Usenet needs a mechanism for maintaining a database of mail distribution
lists at each backbone site, and for passing updates to this database along
with the feeds.  This is why many of the most popular academic discussion
groups are still administered using the LISTSERV utility on BITNET.  It should
be possible to use the newsreader to request a subscription to an open mailing
list (and to examine a list of lists), even though the message distribution is
via mail rather than a newsgroup feed.  Many special-purpose applications are
best served by separating the message traffic from the feeds, for both
political and performance reasons.  Also, a lot of people are reluctant to use
mailing lists now because they feel their application needs the exposure that
the Usenet feeds provide.  If mailing lists fit into the namespace somewhere
and could be queried by the newsreader, it would substantially reorganize the
information and expedite the matching of new users to their most immediate
interests.  As a side benefit, it would greatly reduce the bandwidth wasted on
flames about the wasted bandwidth.  This almost certainly requires protocol
enhancements.
 
I do not wish to slight the valuable suggestions put forward by others
regarding local newsreader enhancements and the improvement of searching and
indexing capabilities.  I do feel that the above-described changes are of
fundamental importance if Usenet is to remain a single entity.
 
The consequence of not implementing these changes is that the academic side of
the network will, in all probability, explore the idea of maintaining a
separate distribution hierarchy with its own administrators, and as the
software evolves to meet the high volume of academic users, the public sites
will be left behind.
 
> From: amanda@intercon.com (Amanda Walker)
>
> Another thing, which I do think is revolutionary, is that the edges are
> blurring rather quickly these days.  From the newsgroups & occasional
> gatewayed mailing list, we now have AP & UPI feeds, Fidonet conferences,
> gateways (at least experimental ones) to commercial services like CompuServe,
> Genie, AppleLink, and so on.  Usenet is quickly become the place for "one
> stop shoppping" when it comes to computer-based telecommunication.
 
This may sound like a quibble, but most of the above is attributable to service
enhancements on the Internet, and does not represent any change to Usenet
itself.  Public and commercial users tend to lose sight of the fact that the
Internet costs money, and that it is primarily underwritten by universities and
the federal government for the specific purpose of furthering the academic
mission.  The Internet is not, nor will it become, an unregulated gateway for
the private sector to access commercial services.  This has considerable
bearing on what can and cannot be done on Usenet.
 
> From: wayne@dsndata.uucp (Wayne Schlitt)
>
> personally, i think this change is great.  this means that most people
> arent using usenet because it's a neat new revolutionary networking
> concept, they are using it to get "real things" done.
 
With many more applications still to come, I hope.
 
> From: brad@looking.on.ca (Brad Templeton)
>
> In some ways, the commercial nets are way behind USENET and don't intend
> to catch up.  In other ways USENET is left in the dust.   When was the
> last time you saw something on USENET that excited you as a new development
> in computer networking?
 
Never mind .aquaria, it's sci.forest vs. alt.trees!
 
> From: csu@alembic.acs.com (Dave Mack)
>
> ...  If we are to make major improvements to Usenet, I believe it will
> be at the cost of access to quite a few of the users. I suggest that the
> way to do this with minimum disruption to the rest of the net is to
> do it in another hierarchy - effectively a different net, with different
> rules, that doesn't try to cater to the lowest common denominator, that
> will set standards rather than obeying them.
 
We have just completed a cycle of protocol and transport enhancements.  It is
now time for experimentation on the applications side, to see what will and
will not work within the scope of the present model.  If the changes I
enumerated above don't occur on Usenet, then I agree that we will see the next
cycle of software updates occuring in the context of a separate hierarchy.
 
(bill)
 
Bill Newell
Systems Analyst, Applications Consulting Group
University of Washington
WCN@MAX.ACS.WASHINGTON.EDU