Re: The List again :-)

smb@ulysses.att.com
Wed, 10 Oct 90 22:44:16 EDT

I should have replied before this, but life has been very hectic even by my
usual standards....

--Steve Bellovin

-----

Here is the latest copy of the list. What I tried to do was to take
Mark's suggestion about dates and apply it to what I already had.
I also took the comments of other people and incorporated them into
the list. If you have time and the interest please look this over
and see if there's anything you want to comment on. Maybe the way
to deal with the list would be to break it down into smaller pieces.
I'm open to suggestions as to where to make the breaks.

This is still mostly Brad's list. The lines that begin with (BT)
are from his original work. Additional topics are provided as
noted. Comments on the list are credited to the originators.
[My own questions about specific things are in brackets -bj]

Lastly what the list needs most are dates. The year of an event is
fine but if you can approximate month and year please put them in
and return the list. I'll do a recompilation in a week or so.

List Begins Here:

(BT) The human-nets digest on the arpanet

>From stargate!mark@cis.ohio-state.edu Wed Sep 26 06:34 PDT 1990
> human-nets and sf-lovers were both equally important.

Both were equally important to Usenet, but I believe that human-nets
came first on the ARPANET.

(BT) The uucp program

The release of the uucp program with v7 UNIX provided the initial impetus.
So did the Bourne shell. The very first version of netnews was a 3-page
shell script. It supported multiple newsgroups, cross-posting, and
subscription lists implemented as environment variables. As best I can
tell, this script has not survived. (A few years ago, I did search for
it, to no avail.) Another motivation was some sort of local news system.
On V6, Duke and UNC had a local news system that came from Somewhere. But
articles were limited to 512 bytes, and we didn't carry it forward to v7.
A prime requirement was that there be a very efficient way to test for
the presence of news (hence the checknews program).

The original idea for netnews came from Tom Truscott and Jim Ellis.

(BT) Notesfiles

>From stargate!mark@cis.ohio-state.edu Wed Sep 26 06:34 PDT 1990
> Notesfiles were around before Netnews at U of I - they belong after "u
ucp prog"

Notesfiles existed on Plato before this, but I don't think there was a UNIX
version until after netnews. Even if there was an earlier one, the cultural
thread was netnews-driven -- notesfiles was adapted to speak the netnews
protocols.

(BT) The A-news system (Bellovin, Truscott)

Actually, I wrote a C language translation of my shell script; on an 11/45
that didn't even have 256K of memory, the shell was just too slow. This
version (pre-A) is also lost. It was never released past Duke and UNC.
Truscott and Steve Daniel wrote the version of A news that was released.
Netnews was announced at Boulder Usenix (Jan '80), though the code wasn't
quite ready.

Some day, I'll write up what I remember of our design criteria. It's worth
noting now that given the speed (or the lack thereof) of the machines we
had, we utterly relied on the ease of writing shell scripts to experiment
with protocol variants; compilation would have taken much too long.

(BT) The first newsgroups
"fa" groups
"net." groups

>From trt@rti.rti.org Tue Sep 25 06:57 PDT 1990
> The very first news groups were "NET." and local groups
> such as "dept". Later Horton et al. oversaw the lower-casing of NET.
> Only when ucbvax joined the net did "fa" appear.
> Indeed I was unaware of the Arpanet mailing lists
> such as human-nets until ucbvax enlightened us.

Correct. The original concept was that most of the traffic would be of
the form now known as unix-wizards (or whatever it's called this week).
Growth was slow until Mark started feeding the mailing lists in because
there was nothing to offer prospective customers. Given a ready source
of material, people were attracted.

It's interesting to note that (in my not very humble opinion) the quality
of dialog (and the moderator's efforts) on human-nets are still essentially
unsurpassed. Only comp.risks is a serious rival (and probably a superior);
I exempt otherrealms from discussion because it's a different sort of beast.

>From stargate!mark@cis.ohio-state.edu Wed Sep 26 06:34 PDT 1990
> i.e. the renaming of NET to be net

(BT) The mistake of net.general
[What was the mistake of net.general? -bj]

net.general was a group intended to be read by everyone. This worked while
the net was still small.

(BT) Creating groups by typo or whim - net.joke, net.bizarre,
net.flame, net.gdead net.wobegon and others

The first joke newsgroups were net.suicide (the original article called
for a meeting of some university suicide club on the roof of a building)
and net.dead-babies, which contained a series of AP wire stories on
a school bus accident. I seem to recall a specific incident that prompted
a particular individual to create them, but I no longer remember what it
was.

(BT) Net goes international (Canada) (Spencer, Templeton)

I've been wondering of late if that will ultimately be the greatest
contribution of the net to history -- that it provides a cheap, reasonably
accessible, mechanism for multi-way international communication.

(BT) The B news software (Horton, Glickman)

The important point here is that B news -- as most of the enhancements
to netnews software -- was load-driven. The original A news had a number
of design choices that made it unsuitable for a large net. (We estimated
a maximum size of 100 sites, and 1-2 articles a day, net-wide....). Many
of the deficiencies could have been patched around (and indeed, many were
in A+ news); the key problem was that the last-read time was a global
concept that applied to all newsgroups; you couldn't read things out
of order.

The goal there (and in many other spots) was to have software free of
databases, and especially central databases. Instead, we chose to let
the file system do the work. This also worked to eliminate most of the
explicit synchronization. For example, article id's were limited to 14
characters or less, and named the file in /usr/spool/news that held the
text. No history file was needed for duplicate detection; stat() would
tell you that. Etc.

(BT) Pathalias software (Honeyman)

>From rsalz@bbn.com Mon Sep 24 16:20 PDT 1990
> The first version of pathalias was written by Steve Bellovin while
> working on his doctorate at the University of North Caroline at Chapel
> Hill. He posted it to Usenet in around 1982, and included patches to
> picked it up and installed it on allegra, a Bell Labs machine at Murra
y
> Hill, NJ. At that time, allegra was the center of the UUCP mail world
(a
> role since assumed by sites named ihnp4, seismo, and now uunet), and a
t the
> time when Usenet was starting to explode all over the place. Before h
e
> left for Princeton, Peter had moved from tweaking the options, to doin
g
> serious redesign of most of the algorithms. Peter is still maintainin
g
> the program, and it has followed him out to the University of Michigan
.

Rich's chronology is about right, though I no longer remember the exact
dates. (Incidentally, there's apparently a line missing from the above
paragraph. Peter brought the code up on allegra.) I also tried to gather
the data, but the effort was too much for me. The data arrived in very
``dirty'' form, and especially after I join the Labs I had no time to
reconcile it all. And yes, the code is essentially all Peter's now;
virtually nothing remains of my original.

>From stargate!mark@cis.ohio-state.edu Wed Sep 26 06:34 PDT 1990
> (I think pathalias came later, and was written by Bellovin.)

(BT) Decvax and Bill Shannon's contribution

This was moderately early; they picked up the slack when assorted AT&T
sites no longer could/would.

>From stargate!mark@cis.ohio-state.edu Wed Sep 26 06:34 PDT 1990
> The first used car ad probably belongs in here somewhere.

The true meaning here is rather amusing. When we were arguing over the
design of the protocols, I argued strongly for many local -- single-machine --
groups, with the sole group NET to be broadcast. Jim Ellis pointed out
the need for regional group for things like used car ads; thus, NET.* was
born. (Originally, network-wide groups were identified syntactically;
hence the use of all caps to distinguish them from anything local.) Jim
prevailed (fortunately); naturally, we were all very amused when the first
used car ad showed up on NET.general. The original designers all sent
congratulatory messages to the perplexed poster.

We discussed, but never resolved, the difference between a distribution
and an interest group. No solution seemed right. 11 years later, the
current implementation suffers from all the faults we talked about way
back when.

(BT) Chain letters over the net.

This was later, I think.

(BT) AT&T's contribution (allegra, ihnp4 etc.)

Research and mhtsa were much earlier, and I think vax135.

(BT) Widespread net growth and international growth

(BT) Forged article critical of Unix (1981)

This was not so much a pure forged article as a forgery designed to conceal
the unauthorized posting of a critical paper by Don Norman.

(BT) K News (proposal)

>From rsalz@bbn.com Mon Sep 24 16:20 PDT 1990
> K News is a non-topic; it generated, what, a month or two of discussio
n?

(BT) Domain style naming (Postel et al)

I don't think this is particularly related to Usenet. The impact of
Usenet on the ARPANET was more as a (strong) catalyst to force re-examination
(and benign neglect) on the strict policies against interconnection. Uucp
mail into the ARPANET became a major force long before it was legit. And
it was obviously known to, and ignored by, many of the Powers that Were.

(BT) The net.jokes.q creation and deletion

(BT) Formation of the "Backbone"
>From spaf@cs.purdue.edu Mon Sep 24 18:58 PDT 1990
> backbone creation, circa 1984

>From spaf@cs.purdue.edu Mon Sep 24 18:58 PDT 1990
> monthly postings of FYI stuff about 1984
[What's this one all about? -bj]

(BT) net.women.only experiment

(BT) Arpa mailing lists merge with usenet groups.

Comparatively early.

(BT) Subgroups to divert traffic:

(BT) net.startrek, net.abortion, source code discussion

(BT) Hoaxes etc. (kremvax)
[Is there a copy of the kremvax article floating around somewhere?
I once saw a copy but ... -bj]

(BT) The re-emergence of mailing lists

(BT) net.columbia and naming debates

>From stargate!mark@cis.ohio-state.edu Wed Sep 26 06:34 PDT 1990
> The UUCP Project (Horton et al)
[Mark places this here chronologically if I read his message
correctly. Is this different from the earlier mention of the
uucp program? -bj]

I assume this is the UUCP Mapping Project, which was designed to supply the
data for pathalias.

(BT) The emergence of voting for creating newsgroups

(BT) RN & Kill files (Wall)

Note again that this was a load-driven development.

(BT) News batching and compression

Ditto.

(BT) Checkgroups messages (Spafford)

>From spaf@cs.purdue.edu Mon Sep 24 18:58 PDT 1990
> checkgroups -- late 1986

The whole issue of control messages was and is troubling. The original
versions of netnews had no control functions because we knew we couldn't
authenticate the messages. (Yes, we knew about public key cryptography
and digital signatures. If nothing else -- and I had read the RSA paper
by then -- V7 included enroll/xsend/xget.)

Some of the code to handle some messages was dangerously buggy in the
early days. I interviewed at the Labs very soon after B news came
out. When I entered the office of a Very Prominent UNIX guru, the first
words out of his mouth were ``Netnews B is a tool of the devil''.
(As an example of an early bug, the code would honor requests to rmgroup
../../../../../.. -- and recall that v7 did not honor setuid if root
did the exec.)

(BT) The problems with the old releases of B news

(BT) Line Eater Bug

(BT) Moderated newsgroups:
a) mod.announce
b) mod.newprod and commercial information on the net

Here we see a return to syntactic decisions.

>From spaf@cs.purdue.edu Mon Sep 24 18:58 PDT 1990
> moderated groups -- late 1986 (after summer Usenix in Atlanta)

(BT) netnews software implemented on VM (IBM Mainframes)

(BT) Stargate (Weinstein)

(BT) NSA Baiting in messages

Not really significant of anything but paranoia, if you ask me.

(BT) Big Net personal fights
a) some result in lawsuit threats (Mark Ethan Smith)
b) some result in people being bounced temporarily from the net
(Maroney)
c) some result in people being removed permanently.

(BT) The great renaming (Horton)

>From rsalz@bbn.com Mon Sep 24 16:20 PDT 1990
> The renaming was instigated and done by Rick Adams because his
> news/sys file was getting unwieldy.

>From spaf@cs.purdue.edu Mon Sep 24 18:58 PDT 1990
> grand renaming -- started July 86, ended March 87

>From stargate!mark@cis.ohio-state.edu Wed Sep 26 06:34 PDT 1990
> Actually, Rick Adams was the prime force behind this.

(BT) The merging of moderated groups into the hierarchy.

(BT) The decline in propagation for groups outside of "comp" and "news"

This was part of the hidden (but widely admitted) agenda of the renaming.
In particular, there was a struggle to keep many groups out of ``talk'',
which everyone knew would be the first to go.

(BT) Creation of rec.humor.funny (first ultra-moderated non-mailing-list
)
(Templeton)

(BT) Battle over the expiry dates on rec.mag.otherrealms. (Von Rospach)

(BT) The attempt to form comp.protocols.tcp-ip.eniac. (Webber)

(BT) Monthly postings, newuser group, commonly asked questions.

>From spaf@cs.purdue.edu Mon Sep 24 18:58 PDT 1990
> monthly postings of newsgroups -- pre 1981

(BT) The misnaming of soc.culture.china -> 30 day voting time.

(BT) Attempt to remove talk.bizarre from the net for the "VOLUME,VOLUME,
VOLUME" game.

(BT) "jj@portal" (Rob Noha) begs for money on the net.

(BT) General problems with Portal, and the issue of pay-for-access net s
ites.

(BT) Attempts to form groups for drugs and sex.

This belongs earlier.

(BT) Creation of the "alt" hierarchy.

>From rsalz@bbn.com Mon Sep 24 16:20 PDT 1990
> John Gilmore and Brian Reid, with help from Carl at pyramid, created
> alt. I think the reasons were that Brian didn't like mod.gourmand-->
> rec.recipes, and that John didn't like having no unmoderated source
> group. I would contact them at first, however.

Certainly true of Brian.

(BT) UUCP project and .US domain. (Gilmore, Horton)

>From rsalz@bbn.com Mon Sep 24 16:20 PDT 1990
> The UUCP project was started by Horton, with Weinstein, and Mel.

(BT) Arbitron news statistics (Reid)

(BT) Telebit releases the Trailblazer.

(BT) UUNET arrives (Adams) (Seismo fades)

(BT) Internet & TCP/IP begins to pervade the net. The backbone begins t
o fade.
>From spaf@cs.purdue.edu Mon Sep 24 18:58 PDT 1990
> backbone goes away, 1987

(BT) NNTP causes news to propagate too fast.

>From rsalz@bbn.com Mon Sep 24 16:20 PDT 1990
> NNTP and the large-scale gatewaying into newsgroups is Erik
> Fair's doing.

(BT) PC-Pursuit

(BT) Reliable links to Europe

(BT) Crazy creation of groups in the alt hierarchy. (Weiner)

(BT) alt.gourmand kepr rather than being renamed. (Reid)

(BT) AT&T decides not to forward mail.

(BT) "biz" and "inet" hierarchies. "Gnu" hierarcy.

(BT) comp.sys.next violates voting rules

(BT) AT&T complains about source code on Killer, shuts it off temporaril
y(?)

(BT) comp.binaries.ibm.pc goes to 8 megs/month -- gets moderated in firs
t,
and only, moderator election. (Dhesi)

(BT) The forming of comp.society.women (proposed as comp.women) (Roberts
)

(BT) Oct/88 Internet Worm (perhipheral to usenet)

(BT) Nov/88 The battle over rec.humor.funny (Templeton, Richmond)

(BT) C News (Spencer, Collyer)

(BT) NN (actually started earlier)

(BT) B News 3.0 (Raymond)

(BT) alt.fusion, sci.physics.fusion

(BT) More rec.humor.funny -- Jokebook, Expansion to GEnie, Denninger's v
ote

(BT) Aquaria alt, rec, sci (sci.skeptic?)

(BT) Battles about naming, Australian voting etc.

(BT) shareware on usenet

(BT) flow mapping

(BT) bitnet merging into USENET, more newsreaders etc.

(BT) more moderator copyright (telecom, sci.med.aids)

(BT) June 89/ClariNet

(BT) In Moderation Network

(BT) More forgery

(BT) alt.sex, alt.sex.bondage and "Cindy's Torment" -- cutoff at U of T,
etc.

(BT) AlterNet, NYSERNET/PSI, NSFNet, FARRnet
>From rsalz@bbn.com Mon Sep 24 16:20 PDT 1990
> AlterNet, NYSERNET/PSI -- non-issues.

(BT) Confiscation of USENET sites involved with phrack.

(BT) EUNet and its policies (pay for feed, limit feed, not all groups)

(BT) rec.arts.erotica

(BT) Alt explosion more alt bannings, rise of alt.sex

(BT) trial newsgroups

(BT) USENET over 10 megs/month

>From stargate!mark@cis.ohio-state.edu Wed Sep 26 06:34 PDT 1990
> Surely you mean over 10 megs/day. It's about 12.5 now.

(BT) BIFF and further use of forgery

(BT) Alt.sex becomes #1 read USENET group

(BT) comp.unix.sco -- debate over corporations and the net (yet again)

(BT) June 90/"Internet Porno Ring"/Houston Chronicle

This page last updated on: Jul 1 09:16