Living and working in modern societies is made up of an increasing amount of communication. In view of the increase in communication - deeds must be communicatively prepared or assessed or even expressed by way of communicative activities only - our society can be described as a "talkative one" (Knoblauch 1996, 19) The different types of network communication - Mailbox "noticeboards", discussion forums, video conferences, groupware, email, IRC channels, the virtual worlds of MUDs and MOOs or network radio, to name but a few - have contributed in no small part to the multiplication of communicative forms and traditions1. Network communication may only have become accessible to a large public since the Internet's increase in popularity, but many of their forms have existed for quite some time. ²

The period during the late Seventies and early Eighties was a time of particular development for electronic communication. Computer networks were increasingly seen as a communication medium, rather than mere code crunchers.³ During this period, as well as the first mailbox networks and the first commercial online service, Usenet developed and with it, News - or, more or less synonomously, Netnews, a collection of subject-oriented newsgroups. Originally an offshoot of Unix, news becomes the Internet's most popular communication service over the years. During mid-1993, Usenet's participants are estimated at approximately three million.⁴ For many people, reading news has become part of their daily routine: "I read netnews right after my mail most mornings." (Raymond 1994, 298)

During the mid-1990s, Usenet represented a service with a tradition on the one hand and on the other hand, it was subject to an influx of new users to whom Internet culture was more or less unknown: the customers of commercial online services and Internet Service Providers. Seen from this background, this paper aims at answering the following questions: Is Usenet an "information-ecological niche" (Schmid & Kubicek 1994, 188) or is it also fit to be used as a service for the masses? In other words: Will Usenet's medial order prove to be scalable or will its "collaborative anarchy", as the network is known, collapse under the strain of its increasingly heterogenous participants? Do comparable crisis situations exist in Usenet's history and how were they overcome? Is Usenet growing in a merely quantitative sense or can we speak of an upward transformation and what changes will this bring about? Will Usenet stay "the same" or will it become something different?

The paper is divided into six parts. First, the conceptual background and method used will be described (The medium as artefact). Then follows a chronological sketch of Usenet's development (Periods of medial (dis)order). After that, we will turn to the resources of establishing order on the network ("How to do things with words"). The next two parts are devoted to Namespace management and Dealing with net abuse. Finally, the network's characterisational framework and limits for action ("Default policy") are dealt with.

In Germany, the discussion on the computer's peculiarities - and related to this - on forms and consequences of informatisation took the opposite route, from the computer as machine or tool to the computer as medium. ⁵ When applied to network communication, an integrated view of computers which talks of "a programmable medium" (Coy 1994) or "an instrumental medium" (Nake 1993) has proven increasingly fruitful. In the interplay of transfer and processing, the computer is seen as a "mediator who not only connects and links, as does every medium, but can also bring about changes to whatever is connected to it" (Nake 1993, 182). Seen from this background, network communication can be characterized as "telematic interaction", a concept coined by Esposito (1995). On the one hand, the concept of "telematic interaction" includes the computer as a technical object which acts. On the other hand, this form of interaction gives users the possibility of symbolic agency in which keystrokes can be used beyond merely interacting with other participants to interact with structures and technical processes. As well as the traditional text which relates to the communication of ideas, network communication also includes an "operative text" with which users can interact with the programme-controlled operation process of messages. In "telematic interaction", instrumental and medial aspects of computer use are intertwined; the medium's habitual latency has been revoked.

The idea of the existence of different textual levels in network communication fits in with the idea of conceptualising network interaction as a new form of writing. This concept is based on the realization that network commnication eludes the anthropomorphous model of situated communicative acts. What goes on during the medial use of a computer cannot be adequately described within a categorial framework of interpersonal interaction (cf. Kräer 1997). The fact that network interaction can manifest itself as or is at least based upon an intertextual communication process is empirically comprehensible in face of hypertext systems such as the WWW. Hypertexts as a specific type of discourse infrastructure are not the only form of "electronic writing" (Wehner 1997). On the Internet, a plethora of new forms of writing and reading can be encountered. Individual network services differ in both the number of textual layers as well as characteristic and valid operations. Mailing lists, Usenet, Internet Relay Chat channels and the virtual worlds of Multi-User Domains (MUDs) all make up their own spaces of interaction.⁶

The focus of this paper is the medial materiality of communicative acts within Usenet. Usenet is viewed as an artefact both from a historical as well as a systematic perspective. The examination is directed towards processes and forms of a medium's creation and change during its use. The element of communicative acts whose subject or object was the medium itself is examined. Our perspective towards communicative acts within Usenet complied with an actor model consisting of the concepts actor, roles, framework, areas and chains of action (aggregated and integrated courses of action)⁷.

The rules and procedures which make a Usenet article a technical object are constitutive for Usenet as an artefact. Usenet enables communication of a certain type: "multilateral asynchronous interactive communication" (Bins & Piwinger 1997, 38). Usenet participants send "articles" addressed to thematically-based newsgroups which are distributed to a more or less large number of Usenet "sites".⁸ There, they can be read by other participants and commented upon publicly (via Usenet) or bilaterally (via email). One of the main characteristics of a Usenet article is its hybrid character: On the one hand, it transports a message and is a technical object comprising specific rules and procedures on the other. Not all articles are directed towards human participants. Control messages take the same route as other articles and can be sent by all participants for host computers on the network to read and execute. Issues regarding network administration are dealt with like any other subject within newsgroups created for this purpose. Usenet is a particularly self-reflexive medium - its control, administration and communication occur within itself with the network's own means. Because of this, the "creation of a medium during its use" can actually be determined.

Methodologically, the analysis is based upon the observation of communication in selected newsgroups, especially those belonging to the news. hierarchy.⁹ These groups are strategically relevant locations of Usenet having itself as their theme, it being both the location and object of communication. At this point, the paths of the discursive (re)constructions of the envisaged order and the interested representation of Usenet's reality cross those of the synchronisation of intended actions directed towards its change. Thus condensed, the medium's structures gain plasticity and become "legible".

From:         "Xavier Lescalier" 
Newsgroups:   alt.culture.usenet
Subject:      Usenet needs a big cleanup
Date:         1998/05/25

More than 75000 newsgroups...

Who the hell is going to clean up that mess? All these newsgroup with no
message or with maximum one every other week. It's really boring.

And not to forget those newsgroup with the most ugly, vulgar, evil and
outrageous names and content (that even some very respectable and eminent
universities are hosting/managing). But I guess there's no way to avoid
having them on the Net.

Last but not least the Useless of Usenet: the newsgroups created by some
individuals to discuss with their friends and family! How interesting that
is for a guy from Paris, Milan or Amsterdam to read the conversations of
the members of the Atkinson family in North Dakota! Are you kidding?

But who is managing all that? (The alt hierarchy and all the others
outside the big 7 (comp, soc, rec, talk, misc, news, sci)) What are doing
the universities and the ISP's?

Isn't there a way to remove the newsgroups created by some brainless
idiots?

Or couldn't the ISP's find a system to no longer publish the newsgroups
that never had any message and/or that have such a ridiculous level of
usage that everybody ignore them? That should not be too complicated,
neither technically nor "ethically".

A good thing would be that the ISP's, the universities and other
institutions who are hosting and creating newsgroup create an
international control organization to keep the house in order.

Let's get serious with UseNet. Some guys should stop abusing or it will
become unmanagable and unusable.

It needs a big clean up, guys...

The claim in the article that Usenet is operating on the brink of chaos can not stand up to a thorough examination, no more than the associated idea of "the good old days", when one was part of an elite group and all was right in the world of Usenet. In retrospect, three periods with characteristic forms of medial (dis)order can be discerned which will subsequently be outlined. An interim synopsis will then follow, describing the typical modes of creating order on Usenet as seen from a systematic perspective.

"Usenet was a combination of several things. It goes back to late 1979 with the upgrade of UNIX from Sixth Edition to Seventh Edition. The particular version of Sixth Edition UNIX we were running had a facility for notifying people of operator messages, news items, in a way that you only saw them once at login time. We wanted something different in Seventh Edition when the code changed. Tom Truscott and Jim Ellis got the idea that Usenet should be distributed. UUCP came out in Seventh Edition UNIX so we got the idea to use this to distribute news items remotely. We saw several uses for this initially. One was administrating news when people weren't at their computers, we also saw it as something that could be used by our department for department news announcements and we also saw it being used for comparatively casual purposes. The main thing we saw it used for in a network sense was UNIX support. The way the original code worked was that any newsgroup whose name began with the word 'net' was distributed and anything else was local." (SB)

In 1979, the first Usenet articles are exchanged between tow universities on the east coast of the USA. Subsequently, Usenet was initially only able to establish itself in universities, research institutes, computer firms and phone companies. A first increase in growth arose in 1981/1982, when mailing lists propagated on ARPANET whose access was limited found their way into Usenet which was open from the beginning. The first graphical representations of the network which was still reasonably small at the time date from this period.¹⁰

Abbildung III/1: Usenet Sites im Juni 1981

Aucbvax.1745
NET.general
utzoo!duke!decvax!ucbvax!mark
Sun Jun 14 20:45:22 1981
current usenet map
After welcoming several new sites to Usenet, I'm enclosing the current
map.
Any sites which are missing or wrong please let me know.

                USENET Logical Map
                June 1, 1981
                !-      Uucp links
                :       Berknet links
                @       Arpanet links

                                  pdp
                    (Misc)         ! (NC)       (Misc)
           decvax sii  reed  phs--unc--grumpy  duke34  utzoo  cincy
           teklabs
              ! !  !    !     !    !             !       !      !      !
              ! +--+----+-----+-+--+-------------+-------+------+      !
              !                 !                                      !
              !               duke                                     !
              !                 !                                      !
              !          +------+---+-----------------------+--------+ !
              !          !          !                       !        ! !
     ucbopt   !  hocsr--mhtsa----research   mh135a        harpo-----chico
        :     !          !  !                  !
     ucbcory  !          ! eagle    ihnss   vax135  (Bell Labs)
  (UCB) :     !          !  !        !         !
     ucbvax--++----------+--+--+-----+--+------+--------+
        :    @                 !        !               ! (Silicon Valley)
     ucbarpa @      (UCSD) sdcsvax      !             menlo70--hao
        :    @     sdcattb-----+        !             !    !
     ucbonyx @                 +-----ucsfcgl        sytek sri-unix
             @     phonlab-----+
         cca-unix           sdcarl

The Usenet Oldnews Archive: Compilation Copyright© 1981, 1996
Bruce Jones, Henry Spencer, David Wiseman.

After a time, the possibility of sending control messages joins the basic functionality - posting (in several groups simultaneously), transporting and reading articles. Control messages consist of cancel messages, which enable articles to be removed subsequent to posting, and of group-related control messages used to add (newgroup)or remove (rmgroup = remove group)groups.

"The very first version [of Usenet Software] did have multiple newsgroups and crossposting between different newsgroups. We went through a couple of other versions. One of the more interesting things we considered and rejected was the idea of control messages. One of the reasons we rejected them was because we didn't think it was feasible to do authentication and the notion of control without authentication seemed to be a broken notion. (...) It wasn't very many years later before they became necessary because you have to have some means of control. (...) Control messages came in early 1982." (SB)

"There was net.news.groups, a newsgroup where people could discuss new newsgroups. And it was just a discussion-consensus process. And people would say, 'here's a new group', and everyone would say, 'that's no good', 'that's good', 'that sucks' - whatever they would say; and of course, my favourite non-issue: 'what name should it have?' But what became clear was that people would come up and once that process got established, it was sort of felt to be good manners that you would suggest and discuss things before you went and did it. You still just went and did it. It was a long time before you asked someone else to do it." (BT)

During the mid-Eighties, when Usenet grew to accommodate 1000 sites, not only increasing traffic becomes a problem, but also the behaviour of some users:¹¹

"There is strife and hostility over when and how to create and delete groups, and the sheer volume of postings is drowning sites. (...) Posters are becoming ruder, maliciousness abounds (...). Cancellation messages are being forged, 'rmgroup' messages have been forged, and articles and replies have been directed to the wrong places on purpose." (Spafford, Usenet-II, 10.11.85)

"At that point there were only two options to read the news. There was a programme called readnews and there was just a new thing called v-news. I only had readnews but was very frustrated with it for several reasons. One of them was that it would only show the newsgroups in the orders they were in the systems, the active file, and I thought that was really bogus because I wanted it to read it in the order of interest. (...) The other prime motivating factor was that every time I wanted to read an article with readnews, it would start up a new process and on that old machinery it took about five seconds to start up the process and I got tired of waiting for it. (...) The first prototype of rn was actually a shell script which invoked readnews on an individual newsgroup. To tackle the problem with the slow down I eventually wrote my own program that would actually do things on the fly. It would open the article you wanted to read and while it was still reading it in from the disc it would already be starting to display things on the screen and would also go off and do things on the background while you were reading things, so it actually was trying to be efficient psychologically. (...) Even back then there were people whose articles people wanted to avoid reading. (...) I put in kill files and they were very warmly received. Basically it took people up to another level being able to stay reading their interesting newsgroups without having to give up on them because there was just too much stuff. Eventually the kill files were refined to where you could not only select things that you didn't want to read but also say by default 'Show me stuff made by such and such an author or that contains such and such a keyword'. (...) I think they contributed to the notion of local control over the news." (LW)

"For users, the cost was effectively zero. For site admins, they largely became significant Usenet sites if they were able to bury the connection costs so it didn't come out of their budget or they were able to have someone else pick up the cost. When AT&T sites came on the Net, they were able to use their internal long-distance connections to connect sites together. When government agencies came online, they were able to use their long-distance government phone number system. When a lot of universities came online, they were able to bury the cost on trunk lines that the university owned and just paid bulk fees. With many big corporations the system admins buried the bills in the long-distance calls that were otherwise down for customer service. (...) It was because of that ability to hide the cost or bundle it in with something else that Usenet grew (...) and we were very hesitant to do anything that might endanger connectivity. The naming scheme and the scheme of creating groups grew out of the backbone collective desire to remain low-profile." (GS)

"Growth became an issue more around 1986-87. There were several reasons for that. In that time period, we saw that multi-user UNIX-based machines were becoming more affordable, outside network connections were becoming more affordable and because of initiatives such as the NSFnet initiative, many more places were getting computers in-house that students and faculty had access to. Whereas before, there were perhaps only 500-600 locations around the country that had machines and network connections where they could afford to participate in Usenet and had the population that was dynamic with new users. After 1984-85 that population began to explode. The VAX 750 and 730 became more widely available as major machines, SUN and Powell were out marketing their workstation and sever-based systems, so cheap UNIX was available. I don't know the dates when 1200 and 2400 baud modems became available, but those also became more common. There were a lot of issues that began to push that connectivity. Nationally, there was also a huge increase in the number of undergraduate majors in computer science, almost tripling each year for several years, so that had a major impact." (GS)

A groundbreaking element in Usenet's further development is the reorganisation of namespace. "The Great Renaming" ends in 1987 with the switch to the "Big Seven" hierarchies: the subject categories comp, misc, news, red, sci, soc and talk. ¹⁴In the history of Usenet, this renaming is the only case of a network-wide organisational change. Not only the implementation of a new name scheme, completed in 1987, but also its unintended side-effect gave shape to its character. The attempt at controlling the network gives rise to endeavours to withdraw from the "Big Seven" name scheme as well as from the new conditions imposed upon the creation of new groups introduced at the same time as the renaming.¹⁵ An effective way of doing so is by developing an alternative route of transportation circumventing the backbone sites and the Backbone Cabal's influence.

What starts out as an alternate backbone soon becomes part of the communications network. The alt. groups, introduced in 1986, form the heart of today's alt. hierarchy, in which every user can create groups without any formal procedure. The establishment of the alt. groups brought out the "interpretative flexibility" (Pinch & Bijker 1987) of Usenet technology.

"I mean, the central insight of all was that the software existed independently of the social structures around it, and that we could use the same software with an explicitly different set of social structures and social conventions, and that would be okay. There was almost no technical hacking involved. It was just a social hack." (JG)

"Alt. was sort of in retaliation, I guess, as much as anything else - as a way of creating newsgroups in an ad hoc fashion. I believe, even for larger companies like myself at Pacific Bell, that we really steered clear of a lot of the alt. groups. I just didn't want to deal with all the bandwidth of it, I didn't really want to deal with all the problems of managing it. I felt that most of the content that would be useful in a company was in the comp. and rec. and soc. ones. So I did a certain amount of censorship of news coming in." (DStP)

"In the mid' 80s they [die Backbone Cabal] pretty much controlled what newsgroups were created. A meeting that we had in the Bay Area was to discuss, 'Well, since they are doing this, let's have our ba. newsgroup hierarchy, and let's do our own thing, too!' And that was fairly big news at that time." (DStP)

n the field of network administration, hierarchically-specific rules and procedures develop. As an "institutional field", the creation of new groups undergoes further development in the "Big Eight" hierarchies. Here, integrated chains of action are formed over time along the lines of an increasingly formalised process with written rules, fixed points of joining and leaving, discussion and voting procedures and specific functional roles and group. The act of creating a new group - by sending a newgroup control message - lies in the hands of a person in a position of trust ("The Control"). The administrative model and also partially "Big Eight" namespace becomes a role model for rapidly-developing national hierarchies (such as de. Usenet, uk. or fr.). As one of Usenet's main communicative functions in its early days was bridging the gap between ARPANET (then still closed-off) and the world of Unix, so the "Big Eight" hierarchies that grew were principally those that made the network a commonplace medium, such as soc or rec.¹⁶

Abbildung III/2: Usenet-Konnektivität im Mai 1993

( ftp://gatekeeper.dec.com/pub/maps/letter/worldlinksfine.ps)

In summary, during the second period, Usenet takes on the shape of a segmentarily-structured organisation based upon the grounds laid down in the founding years. The formation of institutions occurs within individual hierarchies, but no longer on a network-wide basis. On the transport level, a radical innovation enables and facilitates the network's further expansion and, due to the speeding-up of communication, changes Usenet from a "correspondence" into a "conversation medium" (Rheingold 1993, 121). Were there around 1000 articles posted daily on Usenet in 1987 (Hauben & Hauben 1997, 44), the number of daily postings in July 1993 rose to around 27,000.¹⁷ Had the second period been one of advancement, the third announced a decline.

Had the second period brought about radical changes to the transport side, the third period mainly brings about innovations to the interface. Although the WWW is initially seen by some people as a threat and competitor, it actually helps in making Usenet more accessible and transparent: Web browsers with built-in newsreaders help even clueless newbies to start reading news; with search engines that also traverse and archive Usenet (http://www.altavista.com) and special interfaces that provide hitherto unknown possibilities of all-embracing newsgroup navigation and research (http://www.dejanews.com), accessing ongoing and past communication is furthered; a growing amount of resources on the network's technology, organisation, usage and contents is now only the click of a mouse away.¹⁸ Not everyone is pleased about this opening:

From:       jeremy@exit109.com (Jeremy)
Newsgroups: net.subculture.usenet
Date:       1998/07/10
Subject:    "Surf" Usenet? (was: nntp benchmark?)

>  This message sent via http://www.talkway.com/.  Surf Usenet!

Good God.  You don't "surf" Usenet!  The sky is falling.

(Actually, I just tried their site, and it doesn't suck all that much, at
first glance.  Presuming you don't actually want to do any serious
Usenet.  Which, if you're going to a site like this, you probably don't.)

Still, though.  "Through its service and interface, Talkway brings Usenet
to the Web masses..."  This is a *good* thing?  I'm all for newbies and
everything, but the *Web masses*?

The world of Unix, which was beating a retreat on the newsreader side, still holds fast on the server side, where Unix machines are used as Usenet workhorses, so to speak.

"The value of Unix is only to keep up with the volume in most respects of news out there. If you get a really high-performance, multiprocessor system that runs some other operating system and can keep up with five gigabytes a day, then I don't think anyone necessarily cares. But most of the development, at least up to now, for the better performing servers has I think it's still always been on Unix." (DStP)

From: centiped@xs4all.nl (Roelf Renkema)
Newsgroups: news.groups
Subject: Re: RFD: news.admin.nocem.policy
Date: Wed, 24 Dec 1997 08:32:37 GMT

(...)

>But we need to bear in mind that NoCeM works only on Unix systems
> right now, and it may be a while before it is included in non-Unix
>newsreaders.  So the only people who could evaluate the prospective
>moderators would be people who can run Unix readers.

Mwoh what do you need, what kind of machine are you talking 'bout?
As a matter a fact I'm currently generating NoCeM with a bot under
windows'95 and Forte's Agent, so don't give un(othing)ix all the credit
:-)

One of the things able to firmly establish itself was the hacker culture's legacy, the "open source" movement. The Internet Software Consortium (http://www.isc.org/) under whose roof the most popular news software InterNetNews (INN) is maintained and further developed, is part of this tradition. Server systems and reader programmes crucial to participating in Usenet have always been freely available on the net. The WWW has expanded access to reading news. Since the Linux operating system was developed, the threshold to server-side participation has sunk. Using Linux as "public domain" software, a Usenet site can be run from home; every user is in theory able to be their own administrator. One can - as does Brian Pfaffenberger (1996, 30) - view the "democratisation" of the means needed to run a Usenet site as a technologically-based deprivation of the power formerly wielded by news administrators ("technology is eroding the power of system administrators"). The new ways which enabled the unit of user and administrator can also be seen as an indication for a return to the roots. In the mid-Nineties, initial forms of use are currently experiencing a renaissance on Usenet.

Old traditions are also taken up as part of the "Usenet II" project. Ten years after the creation of the first Usenet II mailing list, a list of the same name is established in July 1995. Two years later, it develops into the net. Hierarchy (http://www.usenet2.org/). With strict rules ("Soundness Doctrine") for news administrators, the prohibition of posting anonymously, an expertocratic namespace management system (the naming of net. groups lies in the hands of so-called "hierarchy czars") and a "Steering Committee" which appoints and dismisses "hierarchy czars", the introduction of net. Groups gives rise to controversy, but does not create much traffic.²⁰ In a same vein, the "Mod Squad" adheres to old traditions, namely that of the mod. hierarchy which collected all moderated newsgroups in a separate hierarchy for a short time in 1986.²¹ The "Mod Squad's" intention is to revitalise the alt. groups by way of moderation, as their propagation within the network has deteriorated during the last couple of years.²² Such initiatives are not so much serious movements aimed at restoration but rather an indication for the "pluricultural" state of Usenet which cannot be circumnavigated. The pattern of diversity and doubling also appears when newsgroups with the same subjects and a different status become established:

From: rosalind@xs4all.nl (Rosalind Hengeveld)
Newsgroups: news.groups
Subject: Re: RFD: news.admin.nocem.policy
Date: Wed, 03 Dec 1997 14:01:24 GMT

A tendency lately is for 'useful' newsgroups to exist in: an unmoderated
version, a moderated version, a retromoderated version, and a Usenet II
(net.*) version, or at least a subset of the above. (...) It gives
newsgroups a chance to compete on a 'survival of the fittest' basis.  I
prefer that to endless arguments over these newsgroup statuses between
people who have their mind made up anyway.

What serves as a method for avoiding a lack of friction within the groups also pushes their number sky-high. Due to this context, Usenet more or less generates growth internally. This is nothing new in itself, but is helped by the "surplus economy" which took the place of the "shortage economy" of the founding years. With the group of the Internet service providers, a growing group of actors has arrived on the scene with a commercial interest in offering as broad a range of newsgroups as possible (mainly for marketing reasons and not because they make money out of it).

Abb. III/3: Topological Representation of Usenet in September 1997
( http://www.reference.com/usetop/maps/hosts_5000_0_30_0_0.jpg)

Usenet still continues to grow during the phase of its perceived decline. During early 1998, there are 1534 hierarchies recorded in the Internet Software Consortium's archive.²³ In the summer of 1998 200,000 articles are posted daily in over 70,000 groups.²⁴ A considerable amount of these articles is made up of a new type of unwanted mass posts. The first "Make Money Fast" article can be dated back to 1987, but spam does not become a widespread problem until the mid-Nineties. Dealing with such ill-usage leads to the collision of diverging regulative orientations once again and opens a new round in the struggle for "constitutive rules" (Höflich 1996) for its use.

Shadowed by the vocal debate between "net.control.freaks" and "net.kooks" on the boundaries of what is permitted, attempts at securing Usenet's unity move to the lower levels of network technology. In the years 1996/97 two working groups are formed under the roof of the IETF which deal with the norms of interoperability on Usenet. This protocol, which forms the base of the transport of news on Usenet, is subsequently examined under the microscope, as is the file format for Usenet articles, decided upon in 1983. ²⁵ After strenuous debates, a draft of the revised article format is available in the spring of 1998.²⁶ This document illustrates the price which the "instrumental medium" asks both from users and developers of network communication: "This draft defines the format of network news articles, and defines roles and responsibilities for humans and software" (our italics). As medial and communicative aspects merge when using computers for communicative purposes, so "technical" and "social" elements become inseparably interlinked during decisions pertaining to design. Via the localisation of possibilities of action, specifications belonging to Usenet's "technological core" informally evolve into a moral "geography of responsibilities" (Akrich 1992, 207).

Usenet is a medium which reproduces itself to a great extent by its usage and is intended for the renewed revitalisation of the constitutive rules of this usage under ever-changing conditions. This also includes scope for the extension and revision of existing practices. The interplay of design, use and interpretation brings about change. The resources of the creation of order are also subject to such change. One of these resources which was introduced in the first years of the network, are control messages. With a cancel message, posted articles can be removed, group-related control messages are used to create ("newgroup") or remove ("rmgroup") whole newsgroups. The flow of news can be edited using control messages. In principle, control messages can be sent by every Usenet participant, whether or not they are carried out lies in the hands of the individual local news administrators.

The observation that in the early days, April Fool's jokes on Usenet were often in the form of a (forged) control message and hardly ever nowadays not only points to the change in humorous practices on Usenet; the control messages themselves have changed. If a substantial number of control messages is forged before or after April 1st, a rmgroup announcement can no longer be regarded a good April Fool's joke - if only because of the fact that the majority of genuine control messages carry a "digital signature" and the message meant as a joke would merely be regarded as a primitive forgery attempt. Since the introduction of a regularly-posted list²⁷ who is an authorised sender of control messages in each hierarchy, the fun has ended. The joker could try to forge the digital signature in order to outwit even those server configurations that by default carry out authorised control messages completely and without any further ado and ignore those that are unauthorised. The joke might have then come off, but its author would be rather unpopular.

This example helps to illustrate various basic aspects of the creation of order on Usenet: There are a) miscellaneous article categories on Usenet, b) then as now, each article category can be sent by every participant, c) some articles have undergone supplementary operations (a "digital signature") in order to make forgery more difficult. There are d) further articles (here the "Config Files FAQ") communicating this which authenticate the instances authorised to send control messages. Furthermore, there are e) possibilities to send control messages "automatically" and also not to do so. Anyone forging a (group-related) control message must take the possibility of their article becoming the target of a f) cancel message into account which has g) been declared legitimate by a further document.

The creation of order on Usenet is based upon the assumption that the participants share a complex knowledge of which rules and procedures are "legitimately" linked to which sort of articles at a given point in time (and therefore which jokes are "allowed" and which are not). To a great extent, this knowledge is accessible to anyone and itself is in the form of an article or a document, it can partly be deduced by observing meta communication, for example on practices of network abuse, and it must partly be recognised when actually using the network. This knowledge defines the action frame for actors on the network. It describes and defines "what the artefact does".

Beyond all substantial change this knowledge has undergone over the years, there are staple types of texts and textual layers which enable communicative action ("do things with words"). This elementary stock is comprised of three modes of control, which can be labelled first, second and third-degree levels of control following a concept first developed in organisational research. However, these categories take on a different shape to their real-world counterparts in organisations.

"First-level control" on Usenet includes such textual forms of behaviour control (of human and non-human actors) implemented via programming and software options. These can include specifications in protocols; configuration files on news servers; "patches", subsequent extensions to the server software; control messages; digital signatures; moderator programmes or cancelbots.

"Second-level control"is realised by way of written rules or lists of a semi-official nature, such as: "Request for Discussion" articles, "Call for Votes", "List of New Groups", the "Config Files FAQ". They are all instructive texts posted regularly or used in a ritual manner on certain occasions.

"Third-level control" is effected in a narrative manner - by way of stories and episodes in which expectations as to what is "natural" and "appropriate" and why the Net is the way it is are communicated and discussed. Stories and episodes such as these can be woven around precedents, Net legends, "clueless newbies", and come in the shape of a cool argumentation or a flame.²⁸

During Usenet's development an apparent tendency towards an influx of the first-level control mode has become discernible. This is not to be taken as an indication that users and administrators are increasingly being bulldozed by technology but rather as an expression of the fact that an growing number of processes (can) run "in the background" and both users as well as administrators have greater possibilities of intervention (although they may not necessarily be symmetrical). The abundance of resources for forms of first-level control also means that there is an increase of human and non-human actors' power to act in this dimension. "Agency" increases, but remains distributed.

The second-level control mode by means of official documents also shows an expansion of certain fields of action but is on the whole less pronounced. This observation could support the thesis that - as an essential feature - on Usenet, solutions based on software technology are preferred to organisational ones. First-level control mechanisms are often accompanied by semi-official documents - as instructions for use or as a means of legitimitation.

Whether or not more stories are told, whether corrective episodes increase or decrease and develop differently, which changes occur in the third-level control mode is most difficult to monitor. It seems at least that the registers are broadening. On the one hand, the store of stories Usenet produces is ever-increasing - and with them the possibility of creating internal combinations which can be "instructive". On the other hand, new user groups bring with them real-world references, the art lying in the development of a feeling for which analogies on Usenet meet with a response. Over time, Usenetters become experts of analogy reasoning almost as a matter of course. It goes without saying that an appropriate image of the futility of all analogy reasoning is already in existence on Usenet:

From:       spaf@cs.purdue.edu
Newsgroups: news.announce.newusers,news.misc,news.admin.misc,news.groups,
            soc.net-people
Subject:    That's all, folks
Date:       29 Apr 1993 19:01:12 -0500

Axiom #1:
	"The Usenet is not the real world.  The Usenet usually does not even
	resemble the real world."
(...)
Corollary #2:
	"Arguing about the significance of newsgroup names and their
	relation to the way people really think is equivalent to arguing
	whether it is better to read tea leaves or chicken entrails to
	divine the future."
(...)

We will examine the meaning of newsgroup names mentioned in the article above ²⁹ in further detail in the following part. The aim is to shed light on why naming plays such a central role in the order of Usenet. A concrete example taken from recent network practice illustrated some of the problems which can arise when creating a new group. Naming is an established "institutional field" (Knoblauch 1995, 249-252) on Usenet. The institutionalisation of ways of dealing with Net abuse which will be discussed in the paragraph after this has not yet proved stable, although this field has brought about a software technical innovation.

This toponymic order is the heart of Usenet.³³ While structure and management of the Internet address space has reached a vehemence that goes far beyond the network itself, naming affairs have remained an internal matter on Usenet whose central significance can only be deduced after a closer examination. Its significance is derived from the face that the name of a group is (at least) coded fourfold:

1. (1) The naming scheme makes it possible to find addresses. As soon as the principle behind the location of subjects in a hierarchy is understood, finding an appropriate group should be easy.
2. (2) Has an electronic community been established as the result of continuous and substantial communication, the group name marks out this community's territory. Names function as a boundary by establishing identity (Kollock & Smith 1996). There are few groups with a moderator who actually wields his power as a gatekeeper to any extent. Group-specific forms of humour and a local netiquette are common forms of marking out and protecting boundaries between groups. Moving the boundary's posts, as is the case when splitting existing groups up into subgroups as the result of a reorganisation, often leads to conflicts.
3. (3) A newsgroup's name, indicated by the top level category, makes it possible to deduce which "self-governed area" within Usenet it belongs to. These areas can differ substantially when it comes to forms of cultivating their own namespace. Some, such as the "Big Eight" groups share a standardised procedure and participatory forms. New group creation, including the intended name, is voted upon. Other hierarchies thrive on simple principles, such as the alt. hierarchy ("use common sense") or the free. hierarchy ("absolutely NO rules"). Others have provided for special functional roles, for example the "hierarchy czars" in Usenet II, or supervisory organs.1 However a group's name should come about, individual news administrators are always free to decide whether or not to include newly-created groups on their server.³⁴
4. (4) The hierarchically structured namespace is a starting-point for technical operations carried out by news software. In accordance with the name tree's hierarchical pattern, a news server saves incoming articles: every newsgroup has its own corresponding subdirectory. On a Unix server, our first example taken from the group alt.culture.usenet would be saved in the directory /usr/spool/news/alt/culture/usenet/. The substitution of full stops by slashes signalises a change in materiality has taken place: the newsgroup tree's semantic space has been "translated" to the digital space of software.

Aucbarpa.484
net.news
utzoo!decvax!ucbvax!ARPAVAX:glickman
Fri Dec  4 15:31:15 1981
Re: heirarchical newsgroups, a warning

During the design of version B, I made and then un-made the '.' <-> '/'
change.  This was to preseve compatiblity with mhnews, part of RAND's
MH system.  However, compatibility with MH has yet to be used and the
change might be worth putting back in.  Once MH compatibility is
abandoned, there are a few data-base changes that can be made and would
make things faster.

Matt

The Usenet Oldnews Archive: Compilation Copyright(c) 1981, 1996 Bruce Jones, Henry Spencer, David Wiseman.

The multiple coding of a newsgroup's name lends a meaning to it of which the subject is only one aspect. In Usenet toponomy, taxonomical, regulative and operative processes of communication flow ordering interconnect. On the one hand, this makes namespace the network's "invisible glue" (John Gilmore). On the other hand, as several levels of order overlap within it, a newsgroup's name is a boundary object, and therefore a popular source of conflict. .

During the end of April 1998 a debate commenced in news.announce.newgroups with a "Request for Discussion" (RfD) on the creation of the unmoderated newsgroup rec.arts.superman. According to the RfD, the proposed group's purpose was for the discussion of Superman encompassing various types of media. The following "follow up" appeared in news.groups post-haste:

From:         Russ Allbery 
Newsgroups:   news.groups
Subject:      Re: RFD: rec.arts.superman
Date:         1998/04/26

In news.announce.newgroups, KalElFan  writes:

>                      REQUEST FOR DISCUSSION (RFD)
>                  unmoderated group rec.arts.superman

> Newsgroup line:
> rec.arts.superman     Discussion of Superman in all media.

Ugh.

While I agree that alt.comics.superman probably has enough traffic (along
with the Lois and Clark list and other such places) to support a Big Eight
group, I think the naming choice here is extremely bad.  "Superman" is not
an art and does not belong at the third level of the rec.* hierarchy.

Superman is, and has always been, a comic book character.  (...)  We don't
move discussion of Spiderman, Batman, and so forth out of
rec.arts.comics.* just because they're also in movies, cartoons, and on
children's lunchboxes.
In fact, there was enough discussion of other media in rec.arts.comics.*
to create a whole newsgroup for it, and while the success or failure (and
reasons for same) of that group are open for debate, the obvious fact that
media besides comics when relevant to comic book characters *are*
discussed in rec.arts.comics.* is not.

I think rec.arts.comics.dc.superman would make a great deal of sense,
along the lines of rec.arts.comics.dc.lsh.  I will definitely and without
question be voting against the proposal with the current name.

(Apart from that, I must say that this is a very well-written RFD.)

This article questions the name proposed by the proponent and gives rise to a controversy which came in several waves and came to a conclusion in November 1998.³⁸ For the most part, the controversy is divided between three groups of actors: (1) the new group's proponents and a few supporting voices, (2) the news.groups "regulars" (regular readers who participate in RfDs by contributing both comments and recommendations and (3) the opponents, who in this case are recruited to a large part from the circle of "news.groupies" as well as readers of the rec.arts.comics groups. In the course of their argumentation, supporters, "regulars" and opponents become spokespeople for further groups of actors who may not participate in the debate but are interested in how it ends or provide further input ("Superman fans", "Usenet newbies", "news administrators" or the "library world"). In the middle of the naming debate, a flame flares up, a heated and personal argument between one of the proponents and some "news.groupies", in which the spirit of Usenet is evoked:

From: bill@scconsult.com (Bill Cole)
Newsgroups: news.groups
Subject: Re: Re-opening the Superman newsgroup discussion
Date: Mon, 20 Jul 1998 22:02:26 -0500

(....)  You are a customer of one of the few entities that can
by fiat create a discussion forum with millions of potential participants,
yet you come here to ask a hundred thousand smaller systems to give you a
discussion forum to fit your precise specifications on all their systems,
for free. No one forces you to use the rec.* Usenet hierarchy as your
place of discussing Superman. You could ask AOL to make an area. You could
start a mailing list. You could set up a website with a bulletin board
area. You could even start a group in alt.* or start a whole new
superman.* hierarchy. What you are asking for when you ask for a group in
the Big 8 is for Other People to help you do that. Being an arrogant,
rude, insulting fool is not a good way to get Other People to HELP you.
(....)

Factually, the conflict over the naming of rec.arts.superman is centred around the question as to what the third location in the "Big Eight" naming scheme is for, whether a definition actually exists and who decides upon it. Thereby touching upon basic issues regarding namespace design. During the debate, the proponents concede to the news.groups regulars' argument agreed upon by all that the third-level hierarchy should be reserved for categorising things pertaining to the newsgroup's subject rather than singular phenomena. However, they do not initially take the suggestion to call the group rec.arts.sf.superman any further, which would have resulted in a mutual agreement, which eventually comes about after four months of debate. The fact that things categorised in rec.arts.* are subject to various principles of order (medium vs. genre) complicates the matter even more, as does the fact that supporters and opponents of rec.arts.superman are at cross purposes among themselves. The following article manages to epitomise the proponents' dilemma as well as the ambiguity of the matter quite well:

From:       sbhattac@u1.farm.idt.net (Shankar Bhattacharyya)
Newsgroups: news.groups
Subject:    Re: RFD: rec.arts.superman
Date:       3 May 1998 11:29:24 -0400

(...)

I do believe that the proponents have a set opinion on the namespace
and are more interested in deflecting other choices than in finding
a namespace choice that gets the newsgroup created. (...)

However, the proponents have a choice to make. Are they making the
argument that cross-media newsgroups should go in rec.arts.*? If so,
they should make that case and then stick to their guns. If
rec.arts.superman gets created it will reinforce that as part of the
rec.arts namespace architecture. That's a perfectly reasonable thing
to try to do, in general terms. I think it is the wrong design choice
here, but it is not out of line to say that they think this is a generally
applicable namespace choice and they intend to stick with it.

On the other hand, are the proponents trying to create a newsgroup with
a certain character, for discussion of superman-related topics? If so they
should design their group and then find reasonable namespace for it. (...)

It is time to choose: is this a newsgroup proposal or a namespace
proposal? Is it intended to be both? (...)

The heart of the conflict surrounding rec.arts.superman is comprised of the costs for opportunity concerning naming decisions. It is about namespace architecture and the art of making design-related decisions in such a manner that an extensible space for communication is created. And it is also about the question as to who the architects of this space are.

Amhuxa.314
net.news
utzoo!decvax!ucbvax!mhtsa!eagle!mhuxa!presley
Mon Nov 30 15:01:34 1981
net.* names
I personally prefer the specialized groups.  While we're discussing
this subject, I'd like to see a method established to name a new
newsgroup (if there is one, already, ignore the rest of this message).

If someone wants to start a new newsgroup, he should announce
his intentions over net.news or net.general, suggesting a name.
If there are no complaints after a decent interval (2-3 days?),
that's that.  It's possible that someone would come up with a
better name or point out that such a group already exists with
a different name.

The Usenet Oldnews Archive: Compilation Copyright(c) 1981, 1996 Bruce Jones, Henry Spencer, David Wiseman.

Early attempts at formalising the process go back to the mid-Eighties as part of the Backbone Cabal's endeavours "to put some control on newsgroup creation" (GS). It formed an "institutional field" in the second period, although another formalising surge can be registered after 1993. The gathering-place for naming debates and resolutions is traditionally the unmoderated group news.groups, whereas RFDs and other documents whose status is "official" are traditionally posted in the moderated group news.announce.newgroups (n.a.ng) The group's moderator ("The Control") also publishes the group-related control messages for the "Big Eight" hierarchies.

Fixed rules of procedure aside, a part of naming debates is often the question as to whether regulations are to be applied to a certain case and if so, which ones. The knowledge of these regulations is distributed and n.a.ng's moderator (in the following article "David L.") is reluctant to provide interpretations:

From:       Dmckeon@swcp.com (Denis McKeon)
Newsgroups: news.admin.hierarchies
Subject:    Re: Discussion of Mod Squad and Usenet II groups and designs
Date:       1997/12/03

(...)

It occurred to me last night that getting David L. to comment on n.a.ng
policy or practice is rather like getting Alan Greenspan to comment on
the future of interest rates - any response might cause turmoil, so
hardly anything is ever said that is publicly available. (...)

As an area for action, creating new groups is a vivid example for the basic pattern on Usenet that "constitutive rules" are simultaneously open to debate during ongoing procedures. Over the years there has been a number of attempts at exonerating naming debates from this kind of meta-communication. Proposals aimed at creating some kind of organisation have regularly led to nothing. Solutions which aim at the reorganisation of the procedural steps or at the revision of the voting process seem most promising. "news.groupies" are unanimous in their opinion that the prevailing method is in need of reformation, although views on what should be changed in what way diverge. In the mid-Nineties, the news.groups group expected an imminent revision of "Big Eight" group creation procedures. All eyes were upon the moderator of news.announce.newgroups who, contrary to a previous announcement, did not produce a suggested amendment. The reorganisation of news.groups was often brought up regardless and the group news.groups.questions actually created, so as to separate informational questions regarding the creation procedure from its execution. A further advance attempted in the second half of July 1996 towards splitting up news.groups into various groups along the lines of various procedural components, fell through.⁴⁰ A definitive proposal towards regulation of the group creation process was not again presented until the beginning of April 1998 and in a rewritten version in the end of June 1998.⁴¹ A brief debate, confined to questions of detail, follows. A reform of the procedure, however, does not come about.

The proposed new rules do not aim at a revision of existing forms of self-regulation, but most clearly their continuation and expansion.⁴² With this, a pattern of development is maintained which can be termed "change by integration" (Hoffmann 1997a, 24-25). The existing chain of events of (materially heterogeneous) elements of group creation is suggested to be supplemented by a further element consisting of a mechanism for removing such groups as have no or few traffic with the objective of relieving the news transport of "dead groups" and lending more clarity to the "Big Eight" hierarchies once again. The path leading up to a new "Big Eight" group consists of 27 steps according to the new/old regulations. The procedure in the "institutional field" of group creation seems by now to have become so chock-full of pre-conditions that fundamental revisions are opposed by its own gravity. The group news.groups itself develops conclusive powers of attraction. If toponymical ordering is the heart of Usenet, then it beats here:

From: abby@ucan.foad.org (Abby Franquemont)
Newsgroups: news.groups
Subject: Re: rec.arts.superman RFD: An open letter to news.groups and
KalElFan
Date: 20 Jul 1998 19:17:52 -0700

(...) But for the folks to whom USENET gives something good and
worthwhile, and who stick around and keep using it, eventually there
tends to come a point where you end up in news.groups.

It can be maddening, a frustrating ordeal, this is true -- but, in
many ways, it's the gateway to a sense of ownership of USENET. And
once you have that, I don't know if you ever look at it the same way.
The long-term denizens of news.groups are, for the most part (a few
kooks aside), people who'll fight tooth and nail to keep it going.
It's a passionate fight, these days, and tempers do run hot. (...)

Discussions on the limits between use and abuse and how to deal with usage perceived as being of an abusive nature have been a constant part of Usenet since its early days. As cited in the announcement of the first public presentation of Usenet in January 1980,: "In general, it will be straightforward to detect when abuse has occurred and who did it. (...) Experience will show what uses of the net are in fact abuses, and what should be done about them." (cit. in Bins & Piwinger 1997, 349) However, the initial optimism that abuse and the people responsible for it would be plain for all to see and counter-measures would be applied as a matter of course has been refuted.

Seen in light of this background the outbreak of network-wide culture-clashes between old and new user groups does not occur to the extent one might have expected. The invasion of "strangers", often vocally proclaimed on Usenet and referring to the users of online services, the infamous "clueless newbies" has proven to be a further and by no means dangerous episode in Usenet's history, seen from today's vantage point. The "culture clash" occurs during the struggle against net abuse in two different places: at the intersection of Usenet and "real life" instances and on the network between various groups of guardians.

A conflict zone in which the interior and exterior perspectives clearly separate is formed by the content control of the newsflow. When the online service CompuServe barred its users from accessing the alt.sex.* newsgroups at the Munich public prosecutor's behest around the beginning of 1996, there was a subsequent storm of protest on the network. Usenet, which was briefly wrongly stigmatised as the hub of pornographic activity on the net, soon handed its role as an outstanding object of governmental attempts at intervention over to the WWW.⁴³ What has remained is the attitude prevalent on Usenet, that contents per se do not constitute abuse. Net abuse is seen from within the net as being a "real" problem only if and when the net's ability to function could be harmed: "Cyberporn is not a real problem." (The Net Abuse FAQ). Not "content" but "conduct" is the object of regulation and control on Usenet. ⁴⁴

With the creation of the news.admin.net-abuse.* groups in the spring of 1995 a forum for the co-ordination of "spam cancel" measures has been established.⁴⁵ Spam detectors were developed and a limit decided upon according to which an article is classified as spam based upon how often it appears, regardless of its contents. If the limit is exceeded, scripts aimed at preventing the incriminated article from being propagated any further are run. Regular bulletins name the sites responsible for the most spam sent out and give further information on the numerical development of cancel measures.⁴⁶ Certain conventions require that the senders of cancel messages identify themselves. Cancel articles themselves are to be marked in such a way that they cannot only be carried out by large sites, but also ignored, as an "opt-out"-option geared toward promoting acceptance.⁴⁷

Sending cancel messages is supplemented by collective sanctions against the sources of spam articles. For example, unreasonable Usenet sites that are unwilling to undertake any measures prohibiting spam from being sent from their systems can be subjected to the "Usenet death penalty" (UDP). If this is imposed, individual news administrators are called upon to edit their newsfeed file in such a way that in future, no articles can be received from the incriminated site in question. Contrary to this passive form of the Usenet "death penalty", the active variant entails deleting all messages sent from a "bad site".⁴⁸

Drastic measures such as these call for strong legitimation. Spam is seen as theft of common resources. Accordingly, the IETF aspires to provide further information and help by issuing educational documents.⁴⁹ Spam is theft seen from the background of the Internet cost model because the senders impose the cost of transporting unwanted messages upon the recipients. As far as news is concerned, these costs also take on specific non-monetary forms (e.g. "Spam is theft because it pushes legitimate traffic into early expiration"). This view on things is, however, not generally shared on Usenet, so not only are "spammers" and "spam cancellers" at loggerheads, but also the various attempts at protecting the network's integrity which aim in different directions and can be seen as programme and anti-programme.

On the other hand, according to the stipulations of absolute freedom of speech, there are next to no legitimate reasons for impairing the flow of communication. Seen from this perspective, net abuse occurs when someone attempts to interfere with the relay function of Usenet's transport system: "What is net abuse? Any action that stops a properly configured transport system from performing its normal store and forward services."⁵¹ The aim of "True Free Speech" is mainly postulated by the "Freedom Knights of Usenet" ( http://www.jetcafe.org/~dave/usenet). In practice, their programme of action not only demands that they carry out all newgroup control messages, no matter from where they originate, or ignore rmgroup messages that are aimed at cancelling groups. Third-party cancel messages which are generally seen as abuse, are not only not carried out but deleted. On the other hand, spam is not regarded as net abuse.

"My newsgroup policy has always been 'I allow all newgroups and ignore rmgroups. If you want a new group, forge a new group, come on in.' Newsgroups names are becoming free, there's something called a cyclic file system which stores all news in one file. I think that's the way things should be handled in the first place. I don't think there's any way you can abuse the net if not technically. I could go and crash somebody's news server, but posting spam is not abuse. Posting spam is not abuse, given that I can ignore all those messages in that newsgroup. I must agree that 100.000 messages in the space of one hour constitutes abuse. What do I do? I wrote a posting limit on my news server that says 'you can't post more than this amount in one hour,' it's an exponential backup. You're sitting there trying to post and the server makes you wait. Given this, no-one can damage the system. It's content-blind and I took steps to ensure it's not unfairly applied. (...) We're working on a project to do anonymous news so you can post to a newsgroup and not be traced. The posting can't be traced either, it's a cloud of news servers. If there's enough of them and enough places in different countries, no-one can figure out where it came from, like the anonymous remailer system. My goal is to further the cause of people to post." (DH)

Net abuse forms a semantic and technical field in which programmes of action come together that are not involved in a common plan of preserving and reconstituting of the communication space called Usenet. Just as conventional filtering mechanisms are no longer effective in preventing spam, the mechanism commonly used on Usenet for neutralising or avoiding conflicts fails when dealing with net abuse: the separation of spheres of influence. .

On Usenet, third-party cancel messages are increasingly less seen as an efficient method against net abuse, but a kind of failure to establish order typical for the network. ⁵³ Their part in Usenet traffic is estimated at a third; cancel messages themselves are no small burden upon the network. A more economic system which presents fewer legitimation problems and rules out forgeries to a large extent would therefore be welcome. Such a system, which is recommended both as a "solution" to the spam and the spam cancel problem, is NoCeM (pronounced "No See 'Em).⁵⁴

This system introduces a new category of evaluative articles to Usenet. A NoCeM message includes a list of Usenet articles, which are in a certain format and evaluate other articles using a random principle that must be decided upon. NoCeM messages (NoCeMs)are not directed at human users but address programmes which react to the message's arrival with a locally preadjusted operation, whether they show the articles contained within to the reader at once as being worth reading, and exclude or delete unwanted ones. NoCeM, which has recently become the centre of attention as an alternative to spam cancels on Usenet, is still an incomplete technology and its further development is subjected to the ongoing conflict between both prevailing movements on Usenet as to the "proper" usage of the network.⁵⁵ The debates on NoCeM therefore exemplify the contingency of current technology development on Usenet. During the debate on NoCeM, old controversies on the localisation of the ability to act on Usenet are repeated and renewed.

These controversies reveal themselves as in a burning glass when the creation of the news.admin.nocem newsgroup is proposed during the beginning of December 1997.⁵⁶ Two rivalling views of what NoCeM "is" become apparent. One would like NoCeM to be developed and used as a supplementation for newsreader programmes as was originally envisioned by Cancelmoose. Here, NoCeM has the character of a "3rd party kill file", to which only the individual users have any access. NoCeM messages are regarded as "end to end" communication, which only pertain to the reading and posting software and not the intermediary distribution instances. NoCeM implementations at server software level with which news administrators can filter the newsfeed behind the users' backs (NoCeM on spool) is seen as abuse.

Subject:      Re: Is nanau under attack by HipCrime?
From:         fluffy@meow.org (Fluffy)
Date:         1998/07/14
Newsgroups:   news.admin.net-abuse.usenet

(...)

But NoCeM _isn't_ designed to eliminate messages, only to mark them
read at the newsreader level.  NoceM-on-spool is as much a misuse of
the NoCeM message format as a perl-based header filter is a misuse of
that header. (...)

From the other point of view, NoCeM on spool in the hand of news administrators is not only regarded as desirable but also - at least currently - as the only practicable use.

From:       rosalind@xs4all.nl (Rosalind Hengeveld)
Newsgroups: news.groups
Subject:    Re: RFD: news.admin.nocem.policy
Date:       Thu, 11 Dec 1997 09:48:30 GMT
At this point in time, NoCeM-compliant newsreaders hardly exist, and it
remains to be seen whether the big newsreader makers (Netscape,
Microsoft) will be interested in implementing NoCeM any time soon; also,
whether such newsreaders will have adequate performance in processing
thousands of NoCeM messages from news.lists.filters.

At this point in time, when we're talking application of NoCeM, we're
talking NoCeM on spool.  While still experimental in ways, that is at
least up and running at some sites (including mine). (...)

This marks out the positions between which the discussion on the semantics of cancels is raging: Is a NoCeM message the same as a cancel message or something completely different? It is about the usage of resources by NoCeM clients in regard to functionally equivalent solutions such as retromoderation, about intended effects vs. those that actually occur, and time and again about the actors to whom NoCeM presents new challenges. For news administrators, issuers are the critical mass:

Subject:      Re: Is nanau under attack by HipCrime?
From:         Rebecca Ore 
Date:         1998/07/15
Newsgroups:   news.admin.net-abuse.usenet

Dave Hayes  writes:

>
> Given that the on-spool software exists, and is maintained, it is
> reasonable to presume that it is in use. (In fact, I remember people
> admitting its use in this very group.)
>
> If it is in use at -one- site, then all NoCeM messages are
> effectively cancel messages.

Uh, Dave, you don't know how this works.  NoCeMs are not
across the board automatically applied.  To apply them on the
spool at one site, the sys admin has to put the various
issuers' PGP signatures in his accepted issuers file (on his
keyring).  Until cancels, this is a volitional act.  The
sysadmin has to decide which NoCeMs, whose NoCeMs and what
kinds of NoCeMs from each issuer he'll accept. (...)

On the users' side, even bigger hurdles have to be overcome before a wide-ranged use can be considered.

From: rosalind@xs4all.nl (Rosalind Hengeveld)
Newsgroups: news.groups
Subject: Re: RFD: news.admin.nocem.policy
Date: Tue, 09 Dec 1997 11:28:03 GMT

Henrietta Thomas  had written:

> > Who is going to teach everyone about PGP?

Russ Allbery  in :

> They can figure it out for themselves if they want to issue NoCeMs.  If
> they can't, they shouldn't be issuing NoCeMs in the first place.

I think Henrietta may have been speaking of the use of PGP on the
receiving side, especially on the end user side.  After all, one must (or
at least: should) use PGP to authenticate a NoCeM notice before applying
it.  So, end users must put together a PGP public keyring, and also a
NoCeM permission file (or its equivalent on the end user side). (...)  How
we are going to 'teach' utter newbies to have their newsreader act as a
third-party killfile by way of NoCeM is indeed a far from trivial problem.

Whether or not NoCeMs will make their mark upon Usenet and in which form is for the most part dependent upon which concept of use can find the better allies. NoCeM on spool and NoCeM for newsreaders attempt at creating a network consisting of users, administrators, issuers (of NoCeM messages) and clients. The easiest way to deal with them at present still remains to ignore them, as one of the participants in the discussion remarked.

The development of an institutional field is accompanied by temporary specialisation concerning models of action (for example the "Usenet Volunteer Votetakers" for the "Big 8" hierarchies, the "Spam Cancelers" or the "hierarchy czars" in Usenet II), the establishment of collaborative structures (such as "group advice" in the "Big 8" or the UK Usenet Committee) and by an automatisation of procedures. Especially the latter is an effective form of institutionalisation, as the news administrators have the possibility to delegate tasks that can hardly, or at least with difficulty, be carried out manually. This is depicted clearly in the following episode, using the administration of "Big 8" namespace as an example. The subject is "tale", who routinely sends out group-related control messages as "control", and the consolidation of his role of action by way of establishment in a newer vesion of the news software (INN).

From:       josephb@tezcat.com (Joe Bernstein)
Newsgroups: news.admin.hierarchies
Subject:    Bogus groups in the Big (was Re: mod.*)
Date:       29 Jul 1998 08:42:41 -0500

In article ,
Jim Kingdon   wrote:

> the Big 8 has gone down the "formalize it" path (as has net.* and us.*
> and various other national hierarchies although the formalities are
> rather different in each case).  Now, there are limits to this (for
> example, I suspect that people's adherence is to the Rules more so
> than to tale as a person, well except that the Rules give tale power
> intentionally or not).

Um, this is not what has changed things.  The distribution of INN with
tale defaulted in as the control for the Big 8 is the main thing that
has changed things. (...)

As newsgroups have proliferated, those admins who are selective at *all*
(other than by simple namespace masks, e.g. !alt.binaries.*) are obviously
increasingly pressed for time.  I'm not looking forward to the control
messages that'll start landing in *my* mailbox in a few hours, and I'm
only taking a few hierarchies.  So it's not just that the defaults make an
easy way out for admins, but that they make a *possible* way out for
someone who is not being paid to spend 60 hours a week reading control
messages. (...)

In net communication, delegating tasks to technology plays a more meaningful role in coordinating procedures than differentiated organisational structures and formal rulesets. The range of the delegation principle on Usenet is however limited to the hierarchy level. In self-governing areas such as these to whom certain forms of delegation are unknown, users themselves must lend a hand with technology. Whereas for administrators, the news server configuration files form a coordinating point, for users, the same function is carried out by the newsreader. The following example, which deals with the creation of an alt. group, exemplifies that the default settings of some newsreaders must be changed.

From:       katew@enteract.delete-me.com (Kate the Short -- Spamblocked!)
Newsgroups: news.groups
Subject:    Re: Control Codes for alt newsgroup creation?
Date:       Fri, 14 Aug 1998 22:02:35 GMT

In article <35d6af5d.75166083@news.supernews.com>,
mokane@gate.net sat on the sofa and said:

>Can someone tell me what they are or where to find them?  The FAQ says
>they are needed for alt. newsgroup creation but doesn't say what they
are.

Well, um, if you go to alt.config, they'll help you out with alllll the
info.  Some newsreaders can't put the codes in, though.  Agent *can*,
but you need to alter your agent .ini file with all the settings.  I
think it's ShowAllHeaders = 1 or something like that.  (Originally it's
0 but making it 1 gives you access to Control: and Approved: header lines.

As for the rest, I'd suggest discussing it in alt.config first, as they're
likely to send out a rmgroup message if you don't go there with the
proposal.

The case in point here is one in which the reader's "built-in" action roles obviously do not provide for users being able to send control messages. As can the server's configuration file, the reader's "operative text" can be edited by the user in such a way that the reader programme's settings are rendered obsolete. In addition to the institutionalised rules and procedures which are valid within individual hierarchies in certain areas of action, the - malleable - news software specifications that lay out the framework of action for Usenet participants. Behind them lies the third level, protocol specifications. As far as standard IETF categories are concerned, the status of news protocols is traditionally only an informal one. The interaction between servers when exchanging news between sites and the interaction between servers and news clients when receiving newsgroups and posting articles (RFC 977) as well as the article format (RFC 1036bis) is regulated on the protocol level. Protocol specifications themselves contain prerequisites concerning the developers of server and reader software according to the smallest possible demands on interoperability.

In the summer of 1996, a revision of Usenet's entire protocol world was begun. On the one hand, this work is aimed at refining RfC 977 and towards an extension to NNTP, adding a mechanism for attaching standard extensions and integrating extensions in common use. On the other hand, an effort is directed at the reworking of the Usenet article format. The subject is the "operative text" which itself is attached to the Usenet article, thereby (re-)determining the control messages' type and the conditions under which they are executed. During the course of this - hitherto unfinished - plan it has become apparent that, similar to the reordering of the creation of new "Big 8" groups, previous practices meet with approval and continuity, radical innovations have no chance - the "grandson of 1036" is true to type.

The authentication question has shown itself as not agreeable upon: is it advantageous for the authenticity of Usenet articles to be documented and how could this be best realised? This question not only leads to animated discussions on the proper method of encryption, but is connected to the problems surrounding cancel messages as a method of maintaining order from the start, itself a hot potato. For example, the chairman of the Usefor working group opens the discussion on the mailing list with the suggestion:⁵⁸

For a start I would suggest:

• Authenticated articles (perhaps a generalization of PGPcontrol)
• 3rd party cancels."

(Lyall, usefor, 10.4.1997)"

>My personal opinion is that unless the standard produced by this group
>contains at least an implementable proposal for how unauthenticated
>cancels of all forms will be done away with, it will have been a failure.

I agree. Without strong authentication, USEFOR may as well go home now.
(Franz, usefor, 28.6.1998)

The centre of the controversy is centred around the question as to whether a method of authentification system using a "public key" or a "private key" system is preferable. "Private key" advocates present the vision of a "clean" Usenet, within which all trustworthy participants will in future be more than glad to sign their articles, thus doing away with the worst evils ("spam, forgery and forged cancel") once and for all. Opponents of this proposal invariably mention (among other things) the practically unsolved problem of the distribution of public keys on a network the size of Usenet. It remains doubtful as to whether a network-wide delegation of trust in authorised places of certification and registratures inextricably connected to such a system are compatible with the medium's structures. In comparison, a system with "private keys", in which the user marks her articles with a code only known to her, seems to be less difficult.⁵⁹

H(H(secret+public) [private key cancel lock] is exceedingly easy to implement for 1st party cancels, requires no additional outside effort for the user besides defining a secret, and does not require anyone to trust anyone but themselves. (Cook, usefor, 10.7.1998)

Trust - as the debates on the Usenet article format show most clearly - is a rare resource on Usenet. Nevertheless, it is written in the beginning of the draft of the "grandson of 1036": "USENET is (...) an environment in which trust forms the basis of all agreements. It works."⁶⁰ This contradiction can partly be resolved if one recalls that trust - as well as agency - can be distributed. And communication protocols are a very important area in which their localisation is decided upon, sometimes in advance. And this turns the "technical centre" as defined in the protocol specifications into a "geography of responsibilities" (Akrich 1992). All around the Usenet article, the protrocol creates and defines a world of actors and delegates certain corresponding roles and responsibilities to human and non-human actors. The world of actors runs the gamut from "posters", "posting agents", "injecting agents", "relaying agents", "serving agents" to "readers" and "reading agents". These actors are equipped with a certain inscription: a narrative programme of action with prescriptions and permissions.

For example, it is defined which actors are legitimately authorised to run a certain programme ("Cancel messages MAY be issued by posters, posting agents, moderators, injecting agents on 'proto-articles' [other entities MUST NOT use this method to remove articles]"). The software is required to limitate actors' ability to act in such a way that only legitimate actors are given the possibility to use it ("Posting Agents meant for use by ordinary posters SHOULD reject any attempt to post an article which Cancels Supersedes or Replaces another article if the target article is not by the poster"). The final responsibility for an article's correct state is not held by the poster but the "injecting agent" ("An injecting agent is responsible for taking a proto-article from a posting agent and either forwarding it to a moderator or injecting it into the relaying system for access by readers. As such an Injecting Agent is considered responsible for ensuring that any article it injects conforms with the policies and rules of this document and any newsgroups that an article is posted to.")

The ongoing debates on the Usefor mailing list make it apparent that the regulative principle of the duty book is the actors' trustworthiness. Even more: some actors owe their mere existens to this principle.

"The whole point of distinguishing between injecting agents and others is that posting agents (of which there are many millions worldwide) are in general not to be trusted. They include all sorts of spammers (amateurish or professional) who may try to hide the true source of the article. We are placing a duty on injecting agents to be able to vouch for the correctness of the From (or Sender or whichever) line, and we are proposing to provide them with an Originator-Info header in which they can testify to what they are vouching for. So if something comes in that is suspicious (...) then it is the injecting agent that is supposed to work out what, if anything has gone wrong (since it is the last point where all the relevant information is available)." (Lindsey, usefor, 11.2.1998)

The programmes of action as defined in the communication protocols can be adhered to in daily network practice or not. Even if their interpretation is controversial, they are still acknowledged and used as a reference for moral guidelines of communicative behaviour on Usenet.

In article <3349924B.399D@idt.net>, boursy@idt.net plaintively meowed:

> J.D. Baldwin wrote:
>>
>> There is no "integrity" to the "From:" line.
>
>That's simply untrue.  In fact the cancel mechanism built
>in was specifically designed for posters to be able to
>cancel their own messages--someone else forging their
>email identity is clearly system abuse and morally >indefensible.

That is simply untrue.  In Fact the Cancel Mechanism built
in was specifically designed for Posters *and their site
administrators* to be able to cancel Messages.  What you
call «forged» Cancels are explicitly allowed by RFC 1036
and in RFC 850 before it. Posters have /never/ been granted
an implicit or explicit assurance of absolute Control over
their Articles.

While we're at it, what do you think RFC stands for?

Request Fluffy's Consent.

Meow!
Fluffy

205 Closing connection - good bye!

During the course of its use, Usenet has - contrary to its intention - changed from a primarily local information medium into a global communication service.⁶¹ Time and again, the network has gone beyond the horizon of past futures ("Imminent Death of the Net Predicted!"). Not only does its growth reflect expansion, but also upward transformation. According to the concept suggested by Kubicek (1997) of the "developmental space" of a medium with the coordinates degree of diffusion and types of institutionalisation, looking at the network-wide distribution of Usenet, by now it can be said that it is universally accessible.

The Usenet article has given rise to a separate new media genre, to which a complex knowledge of rules and procedures on Usenet that are passed on in many ways belongs. The resources of the creation of order and ordering patterns which are a consequence thereof take on a new and unusual shape which differs from old media. This text has already described some characteristics of the news medium as artefact and its organisation as a service in detail. Seen from this perspective, Usenet presents itself as well-structured in its own way, even if differentiated organisational structures in the traditional sense of the word have not yet become apparent.⁶² As a testimony to its original culture, Usenet still stubbornly seems to prefer software technical solutions ("first-level forms of control") to organizational ones. But also the "social hack" has its tradition on Usenet. Its constitutively decentral world is divided patchwork-like into a plethora of self-governed areas with their own rules and forms of collaborative management.

Usenet is an unfinished project. It does not present itself as a finished object, but rather as a delicate and confusable operation. Order fails especially when dealing with abusive usage practices such as "spam" as of late. The excess consumption of network resources goes hand in hand with internal regulation conflicts and leads to "informal entropy" situations. Collaborative filters (NoCeM) are a first possible way of coordinating translocal action, but their future is at present still undecided.⁶³ Though Usenet is in many respects a fast medium, yet institutional change is slow to come about.

In many ways, Usenet presents itself as a noisy medium. (1) This is reflected by its atmosphere, which has been described as a "mixture of noticeboard, newspaper and pub" (Bruchhaus 1994, 1), "ongoing commentary from everybody on everything" (Grassmuck 1995, 52), "stream of topical chatter" (Baym 1995a, 138), "narrative greenhouse" (Herz 1995, 80) or "a fair, a cocktail party, a town meeting, the notes of a secret cabal, the chatter in the hallway at a conference, the sounds of a Friday night fish fry, post-coital gossip, the conversations overheard in an airplane waiting lounge that launched a company, and a bunch of other things"⁶⁴ (2). On Usenet, the metaphor of "signal-to-noise ratio" has become the measure for the amount by which communication is impaired by unwanted or unqualified articles. (3) Noise includes the manifestation of the medium as medium, present in its use (Ellrich 1997, 208). On Usenet, users see themselves forced to deal with the medium's function and its materiality when communicating. If Usenet is judged by the fact that the better a medium is, the less it is noticed during the process of sending or distributing a message, then Usenet must be seen as not being particularly successful. However, as is pointed out in the present study, this form of noise is not only omnipresent on Usenet, but possibly a permanent side-effect of telematic interactivity in general.