alt.support.ocd.moderated

Long-form reference · 7,080 words · about a 31-minute read

The newgroup message of 21 November 1999

Almost nothing of what was said in this newsgroup is available to this directory. What is available, in unusual completeness, is the paperwork that created it. The Internet Systems Consortium keeps the archive of Usenet control messages — the administrative traffic by which groups are created, removed and re-announced — and the file held there for this group, pub/usenet/control/alt/alt.support.ocd.moderated.gz, contains four distinct messages sent over the two months at the turn of 1999 into 2000. They are the group's birth certificate, a second creation message issued by a large American access provider, an attempt to delete it, and a further copy of the birth certificate sent eight weeks after the first. Read together they show how an alt.* newsgroup actually came into existence, which is a less orderly process than the phrase suggests.

The first message is dated Sunday 21 November 1999, 19:54:37 in the −0600 zone. It was posted to two groups at once, alt.config and the group it was asking for, and it carried the headers that make a control message a control message: a Subject line reading cmsg newgroup alt.support.ocd.moderated moderated, a Control header repeating the same instruction, and an Approved header without which no news server was supposed to act on it at all. The From line gives no personal name but a role, Newsgroup Creation Authority, at the domain algebra.com; the Organization header simply says alt.config. The message runs to ninety-one lines, and the great majority of them are not instructions to software but a document written for human readers.

It opens with the line that a news administrator would paste into a server's newsgroups file, the one-line description by which the group would be listed everywhere it was carried:

For your newsgroups file:

alt.support.ocd.moderated Obsessive-Compulsive Disorder. (Moderated)

That convention is older than the standard that eventually documented it. RFC 5537, the Netnews architecture standard published in November 2009, notes that in the absence of a properly typed description a news server may search the body of a newgroup message for the line For your newsgroups file: and take the following line as the group's description, and records that this was the convention for providing a description before the format was standardised. The same standard requires that a moderated group be marked by appending the case-sensitive text (Moderated) to the end of its description, which is exactly what this line does. It survives unchanged: the same description, with the same parenthetical, is still the entry for this group in the newsgroups file that the consortium distributes today, and the matching entry in the active file still carries the status flag m.

Two addresses follow, and they are the working parts of the whole arrangement rather than contact details. The host they name no longer resolves, and neither address should be read as reachable:

Group submission address: [email protected]

Moderator contact address: [email protected]

The distinction between the two is the entire administrative shape of the group. The first address is where articles went to be considered; the second is where a person wrote to the moderators about the group rather than posting to it. The second is plural by design — it is a board address, matching a clause further down the proposal that contemplates a vote among moderators when a decision is difficult. The proposal names three people as proponents and initial moderators, with personal e-mail addresses; this directory does not reproduce the names of participants in a support group, and does not do so here.

Then the request itself, in three sentences that contain a claim, a forecast and a favour asked:

Newsgroup alt.support.ocd.moderated has been successfully discussed in alt.config. We expect it to have large amount of traffic soon. Please create this moderated newsgroup.

The claim is that the group had been discussed where alt.* groups are conventionally discussed. The forecast is a forecast, and the archive contains nothing that would confirm or refute it. The favour is addressed to several thousand news administrators individually, because in this hierarchy that is the only audience there is.

The charter, as filed

Under a heading of CHARTER, and a line announcing that a proposal, charter and justification follow, the message sets out what the group was to be. This is the text quoted in the page's opening description, given here in full because it is short, specific and, unusually for a newsgroup charter, more concerned with housekeeping than with high purpose:

The newsgroup alt.support.ocd.moderated will provide a relatively troll and flame-free environment for the discussion of Obsessive-Compulsive Disorder (OCD).

People who have OCD, family members and friends of people with OCD and other people with an interest in OCD are invited to participate on the newsgroup

The newsgroup will be moderated.

No binaries will be permitted.

One advertisement per month per group/organization/individual will be permitted. Advertisements should be labeled by subject headers starting with “AD” and kept short

An alt.support.ocd.moderated FAQ will be created and stored on a web site with the URL to the FAQ posted on the newsgroup every couple of weeks

The first clause is the one the group is remembered by, and the qualifier in it is doing real work. Not a safe environment, not a supportive environment: a relatively troll and flame-free one. The promise is bounded, and it is bounded in the direction of what a moderator could actually deliver.

The second clause is broader than a peer-support room strictly needs. It invites people with the disorder, their family and friends, and — a distinct third category — other people with an interest in it. A room that admits interested outsiders is a room whose entry rule is about behaviour rather than membership, which is the position the justification goes on to take.

The remaining clauses are the domestic arrangements of a late-1990s newsgroup. The prohibition on binaries kept encoded attachments out: binary traffic was expensive to carry, and a hierarchy that had already grown a whole alt.binaries.* branch — one of the two large sections of alt.* that administrators were accustomed to accepting or refusing wholesale — had taught administrators to watch for it. The advertising clause does not ban advertising but rations it: one advertisement per month per group, organisation or individual, prefixed AD in the subject line and kept short. A support group for a specific disorder could expect approaches from clinics, book publishers and self-help ventures, and the charter made room for them under a label rather than assuming they would not come. The FAQ clause commits to the standard Usenet apparatus for a room with a constant intake of newcomers: a document kept elsewhere, with a pointer reposted at intervals so that a reader arriving mid-stream had somewhere to begin.

What no clause does is describe the conversation. There is no list of permitted topics, no statement about what members should or should not tell one another, and nothing resembling clinical policy. The charter legislates about conduct, format and advertising, and leaves the subject matter to the people who had it.

The justification, and the reason it gives

The longest part of the document is the justification, and it is the only surviving contemporaneous statement of why anybody wanted a second OCD newsgroup. It begins with the subject:

Obsessive-Compulsive Disorder is a mental illness, which is very troublesome and painful. We would like to provide its many sufferers and their loved ones a forum where they can ask for and give support and advice.

It then gives its reason, which is not a reason about the subject at all but about the traffic. The proposal states that a newsgroup for the discussion of OCD already existed, and that it had been, in the proposal's words, subject to recurrent invasions by trolls and abusive posters, with the result that a large proportion of the articles in it were flames written by those posters and by the people arguing with them. It adds that significant numbers of people who need emotional support have been scared back into lurking or quit the newsgroup altogether.

The proposers then do something newsgroup proposals very rarely do: they produce a number, and they show their working.

We have estimated that 352 [sampled from Midnight, Pacific Time, October 29th 1999, to the afternoon of Nov 4th at which point 1000 posts had been sorted ‘by hand’] of the sampled 1000 posts to alt.support.ocd were the result of a few trolls.

That sentence has to be handled carefully, and this page reports it strictly as what it is: an estimate made by the people asking for the group, from a sample they collected and sorted themselves over roughly a week in the autumn of 1999. The sorting criteria are not stated, the sorter is not identified, and no copy of the sample survives in the control archive or anywhere this directory can reach. What can be said is that the proposal was argued on measured volume rather than on assertion, and that it is unusual in the record for having tried. What cannot be said is that the number is right.

The same paragraph characterises the material it counted — the proposal states that the posts in question contained sexist and homophobic abuse aimed at individuals, and that some of the people targeted were, in its words, clinically depressed individuals who are trying to find some emotional support. Again: this is the account given by the group's proposers in a document written to persuade news administrators. It is not an independent finding, and this page does not adopt it as a description of alt.support.ocd, whose own record is not preserved here and whose story is told on its own page.

The final part of the justification is a method, and it is more detailed than most moderated groups ever committed to in public. The group was to use a combination of robo-moderation and human moderation. The criterion was stated in one line — posts may not contain personal insults against other posters — with an explicit carve-out: there would be no interference with strong and forceful debates and disagreements, but a line will be drawn when it comes to gratuitous personal insults and cruelty. Where the line fell was conceded to be a matter of judgement, and if necessary, vote among the moderators. Two operational rules follow: posters already known as trolls would be watched permanently should they join, and new posters would be watched for an initial period. The sanctions ran in a ladder — withholding an abusive post was to be the ordinary remedy, with suspension and banning reserved for extreme cases.

What discussion in alt.config actually meant

Everything above is a request rather than a decision, because the alt.* hierarchy does not decide. In the alt.* hierarchy there is no centralised control and anyone technically capable of creating a newsgroup can do so; in practice most creations follow an informal procedure of public discussion in alt.config first, and it is then up to each individual news administrator whether to add the group — some decline groups that were never discussed, and groups that skip the step tend to propagate badly. Removal is in theory the mirror image of creation, but in practice most administrators do not remove newsgroups at all. There is no Request for Discussion, no Call for Votes, no vote tally and no result posting; the apparatus that the Big Eight built around group creation is simply absent, and the hierarchy's own story is told on the alt.* page. A proposal's only real currency is that administrators find it credible, which is why a ninety-one line document with a charter, a method and a counted sample was sent where a single line would have been technically sufficient.

How genuinely decentralised this was can be read off the configuration file that ships with the standard news server software. INN's control.ctl, distributed from the same archive, sets the default handling for the hierarchy: accept newgroup messages from anyone except a short list of addresses associated with other hierarchies, from which an alt.* newgroup can only be a forgery, and silently ignore every rmgroup. Its accompanying comments are unusually candid for a configuration file — they warn that quantities of alt.* newgroups are sent out regularly with the intent of creating nonsense entries in active files rather than usable newsgroups, advise a small site to drop the lot and add groups only when a user asks, and conclude:

Be aware that there is no official, generally accepted alt.* policy and all information about alt.* groups available is essentially someone's opinion, including these comments. There are nearly as many different policies with regard to alt.* groups as there are Usenet sites.

The same comments tell an administrator how to check a doubtful group, and the recommended method is the one this page has used: fetch the control archive for the group name and read the first and last few control messages, partly to see whether the group is supposed to be moderated at all — because, the file notes, some alt.* groups that should be moderated are created unmoderated by hijackers to try to damage the newsgroup. For a support group whose entire proposition is a filter, that is not a theoretical hazard.

The rest of the file records what followed. On 21 December 1999 AT&T WorldNet Services, then a large American consumer access provider, issued its own newgroup for the group: a three-line automatic message saying only that it had been generated to add a newsgroup across that provider's news servers. It used the abbreviated flag m rather than the keyword moderated, which is the only newgroup flag the standards actually define. Two days later, on 24 December, a PGP-signed rmgroup was issued from an alt.* removal account, asking every server on the network to drop the group and giving its reasons as a two-item list: No Charter and No Traffic Rate Measurement. The message carried a pointer to the published documentation on signed control messages and a line an administrator could paste into control.ctl to have such removals verified and applied automatically.

The first of those reasons sits oddly beside a creating message that consisted almost entirely of charter, and the archive offers no explanation — whether the rmgroup issued from a batch process, whether the newgroup had not reached that site, or whether some other objection lay behind the label. What is certain is that it did not take effect anywhere the default policy was in force, because that policy ignores rmgroups in this hierarchy. On 19 January 2000 the Newsgroup Creation Authority sent the whole document again, ninety-three lines this time, the two extra lines being a marker inserted at the top of the charter announcing that this was a booster message and that the newsgroup had been created some time before. Boosters were routine housekeeping in alt.*: a second copy sent because propagation is a matter of individual servers, and some of them will have missed the first.

The Approved header, and what moderation is made of

On Usenet the word moderated names a mechanism, not a mood, and the mechanism is old enough to be specified in the same document that specified the rest of the medium. RFC 1036, the standard for the interchange of Usenet messages published in December 1987, defines a header field called Approved, states that it is required for any message posted to a moderated newsgroup, and says it should be added by the moderator. The same document defines the control message that creates a group, in the form newgroup <groupname> [moderated], notes that the optional second argument makes the group moderated rather than unmoderated, and adds that the message should be ignored unless there is an Approved line in its own header. The message of 21 November 1999 satisfies that specification exactly, nearly twelve years after it was written.

The mechanism is older still than the header. Before the Great Renaming of 1987, Usenet's groups sat in three hierarchies, and one of them, mod.*, existed specifically to hold the moderated discussions; a group was moderated by virtue of where its name sat. The B News release of 1986 removed the requirement that moderated groups carry the mod. prefix and allowed people to post to them from an ordinary newsreader rather than a separate mail program. Moderation stopped being a place in the namespace and became a property of a group recorded in a flag, which is why a proposal in 1999 could ask for a moderated group anywhere it liked and expect the software to do the routing.

RFC 5537, of November 2009, wrote down what the software had been doing in the meantime, and its rules describe the group's daily life better than any recollection could. A server that stores articles must keep a list of the moderated groups for which it accepts articles, and must reject any article without an Approved header posted to any group listed as moderated. An injecting agent — the software that accepts an article from a poster — must, if the article names a moderated group and carries no approval, forward it to the moderator of the leftmost moderated group in the header, customarily by e-mail, or else reject it outright. The moderator decides; if the answer is yes, they add the Approved header, identifying themselves and, so far as possible, any other moderators who have approved it, and the article is injected into the worldwide feed like any other.

This is why the submission address in the paperwork is a mechanism rather than a courtesy. A reader with a normally configured newsreader posted to the group in the ordinary way and never typed that address; their own server looked the group up, found it flagged as moderated, and mailed the article to the submission address on their behalf. The reader's article did not fail. It queued.

Two consequences follow, and they are the difference between moderation and cleaning up afterwards. The first is that nothing rejected was ever distributed: an article that stops in a moderator's mailbox is one that no reader opened, no server stored and no archive indexed. The second is that the alternatives available in an open group were markedly weaker. The cancel message, Usenet's mechanism for withdrawing an article after the fact, was so hard to authenticate and so widely abused that the 2009 standard simply records that many sites were by then ignoring all cancel control messages; and a reader's own defences — the kill file, a list of patterns matched against author, subject or other headers so that matching articles are never shown, first implemented in Larry Wall's rn and copied by every newsreader after it — operate one reader at a time, after the article has already arrived everywhere.

The mechanism's weak point was documented in the same standard. A malicious poster may simply add an Approved header to bypass the moderation process; injecting agents are advised to verify that articles approved for a moderated group are really being injected by the moderator, and the standard concedes that there is at present no standardised method of authenticating approval at all, only unstandardised schemes such as PGP Moose, described in a document of November 1998 and, the standard says, in common use. A group whose promise to its readers was a filter depended, in the end, on nobody bothering to forge the header.

Delay was designed in rather than apologised for. RFC 5537 encourages moderators to keep an article's original Date unless it appears stale — seventy-two hours or more in the past — for reasons understood by the moderator (such as delays in the moderation process). A standards document that budgets three days between writing and appearing is describing a medium in which replies arrive late by design, and the trade is stated rather than hidden.

What a moderator of a support group is deciding

The charter and justification together amount to a job description, and it is worth setting out what the job was not. It was not to decide whether a post was true, or wise, or clinically sound; nothing in the document gives a moderator any standing over the substance of what people said about their own illness. It was not to keep the room comfortable, either: the proposal explicitly preserves strong and forceful debates and disagreements. The single stated criterion is conduct between posters — no personal insults — with a line drawn at gratuitous cruelty, and an admission that the line is a matter of judgement, to be settled by a vote among the moderators if it comes to that.

That choice has a specific consequence for a room of this kind, because a support group is the sort of place where a content rule would be both tempting and unworkable. The material that would get an article filtered almost anywhere else — an unwanted violent image, an intrusive sexual thought, a description of a ritual that sounds alarming in isolation — is the material a room for obsessive-compulsive disorder exists to make sayable; the standard descriptions of the disorder list fears of harming others or oneself, sexual obsessions and fears of acting blasphemously among its common contents. Screening for distressing content would have removed the group's subject matter along with the distress. Filtering on how posters treat one another leaves the subject matter untouched and puts the burden of judgement on tone, motive and target: whether a sentence is an argument or an attack, and whether a new arrival's first posts are the ordinary awkwardness of a newcomer or the opening moves of somebody who has done this before. The watch provisions in the justification — permanent watch for known trolls, an initial watch on new posters — are that judgement written down as procedure.

The costs fell where they always fall. The work was unpaid and unavoidable: to keep abuse out of a room, somebody must read all of it, and moderation concentrates that exposure on the few people who volunteered rather than distributing it across everyone. The promise of robo-moderation in the charter suggests an intention to have software take the first pass — the control archive shows both group addresses sitting on a dedicated host rather than in anybody's personal mailbox, which is consistent with an automated front end, though the archive says nothing about what software ran there or whether it ever ran at all. And moderation of any kind concentrated the decision about what appeared in very few hands, which was a standing argument on Usenet long before this group and remained one after it. The paperwork's answer is procedural rather than philosophical: publish the criterion, name the board address, and settle hard cases by vote.

A relatively troll and flame-free environment

The charter's phrase is worth taking at face value, because both nouns in it had precise late-1990s referents. Troll had entered the vocabulary on Usenet itself: the first recorded usage is from the early 1990s, in the phrase trolling for newbies as used in the group alt.folklore.urban, where it described a comparatively gentle inside joke — a veteran posting an overdone topic that only a newcomer would answer earnestly. The earliest attestation recorded by the Oxford English Dictionary is from 1992. By the late 1990s the practice had been extended to playing a misinformed user in groups where one was not a regular, and the word had broadened from a game played on newcomers to a general term for provocation. A study of identity and deception in virtual communities, published in the same year as this charter, was analysing trolling as a game of identity deception played without the consent of the other players. When the proposers used the word without defining it, they were relying on their readers — news administrators — to know it already.

The other pressure on open groups was industrial rather than personal. Newsgroup spam was defined by Usenet convention as excessive multiple posting — the repeated posting of a message or of substantially similar messages — and it had arrived in volume in 1994, the year of both the first widely recognised Usenet spam and the advertisement for legal services that made the practice famous. By the late 1990s it had also become a weapon: sporgery, the forging of large numbers of articles in a targeted group, was used to make particular newsgroups all but unreadable without heavy filtering. The network answered with the Breidbart Index, an objective measure of a message's spamminess, and with organised volunteer campaigns to purge groups of it. A support group was not a special target; it was one of tens of thousands of rooms into which the same advertisements arrived. Both problems shared a property that made moderation attractive: they were about senders rather than subjects, and could therefore be filtered without anyone having to rule on what the group was allowed to talk about.

The response was a recognisable pattern across the network rather than an invention here. Once moderation had become a flag rather than a hierarchy, a moderated group could be created anywhere, and a common naming practice grew up of appending .moderated to the name of the group being shadowed. The scale of the practice can still be counted: the active file the consortium distributes today lists more than a hundred and fifty groups whose names end in .moderated, and about four-fifths of them are in alt.*. The suffix functioned as a public statement of mechanism: same subject, different door. Readers chose between the two by temperament — immediacy and openness on one side, a filter and a delay on the other.

One thing this page will not do is characterise the open group. The only account of it in the surviving paperwork was written by people arguing for a second group, and no message archive for either group is preserved in this directory. That alt.support.ocd was unmoderated is a matter of record; what living in it was like in 1999 is not something the record settles.

The subject, in the minimum a reader needs

The disorder itself is treated at length on the open group's page, and only the part that bears on why a moderated text room made sense belongs here. Obsessive-compulsive disorder is characterised by obsessions — persistent unwanted thoughts, mental images or urges that generate anxiety, disgust or discomfort — and by compulsions, repetitive actions performed in response to them in order to reduce that anxiety. Standard descriptions of the period and since list common obsessions as including fears of contamination, preoccupation with symmetry, fears of acting blasphemously, sexual obsessions and fears of harming others or oneself. The thoughts and behaviours are usually described as egodystonic — inconsistent with the sufferer's own image of themselves — and most adults with the disorder recognise the rituals as irrational while continuing them. Those are, almost by definition, the contents a person is least willing to attach to their own name, which is the whole reason a pseudonymous medium mattered: describing such a thought in text to strangers who recognise it is a materially different act from saying it aloud to somebody who knows you.

The clinical furniture of the group's era can be dated, and is set down here as history rather than as guidance about anything. Behaviour therapy for the disorder — exposure and response prevention, meeting the trigger while declining the response that usually follows — traces to the work of the British psychologist Vic Meyer at the Middlesex Hospital Medical School in London in the mid-1960s. Meyer derived the procedure from his analysis of the extinction of fear in animals by flooding and applied it to cases of obsessional ritual that were at the time considered intractable; his account was published in 1966. On the pharmacological side, the benefits of clomipramine in the disorder were first reported by the Spanish psychiatrist Juan José López-Ibor in 1967. It was the first drug to be investigated for and found effective in the condition; the first double-blind, placebo-controlled trial of it for that use was run in 1976, more rigorous studies followed in the 1980s, and it was approved in the United States for obsessive-compulsive disorder in 1989 and reached the American market in 1990. It remains the only tricyclic antidepressant approved there for the disorder, and the disorder remains its only labelled indication in that country. Fluvoxamine, approved by the Food and Drug Administration in 1994, was the first of the newer SSRI class licensed there specifically for it, and the SSRIs went on to displace clomipramine on tolerability and safety rather than on effectiveness. By the time the group was created, treatment existed, was recent, and was unevenly distributed.

Severity had acquired a number too. The Yale-Brown Obsessive-Compulsive Scale, designed in 1989 by Wayne Goodman, Steven Rasmussen, Carolyn Mazure and colleagues, rates ten items — the first five covering obsessions, the last five compulsions — and is normally scored as those two subscales; it measures severity without weighting the content of the symptoms one way or another, which is part of why it travelled. Its principal author later described it as having become the gold standard measure of symptom severity, and by 2018 the paper that introduced it was the most cited article on the disorder. The public standing of the subject had moved as well: Judith Rapoport's The Boy Who Couldn't Stop Washing, published in 1989, was a bestseller about a condition that had until recently been treated as a rarity. Classification moved later, and after the group's active life: obsessive-compulsive disorder sat among the anxiety disorders in the manuals current in the 1990s, and was given a chapter of its own, obsessive-compulsive and related disorders, only in the fifth edition of the American Psychiatric Association's diagnostic manual in 2013. Present-day estimates put the lifetime prevalence at about 2.3 per cent, with an annual rate around 1.2 per cent, affecting men and women about equally, and it is unusual for symptoms to begin after the age of thirty-five — figures that describe a condition common enough that any general-audience newsgroup would have had readers with it, whether or not it had a room of its own.

Reassurance, and what it does to a support group

One feature of the disorder bears directly on what a forum for it can be, and it is documented in the ordinary clinical description rather than in anything peculiar to Usenet. Compulsions are the repetitive actions performed in response to obsessions to reduce anxiety, and the standard lists of them — drawn in current summaries from the United States National Institute of Mental Health and the American Psychiatric Association — run: washing, checking, counting, reassurance seeking, situational avoidance. Requesting reassurance appears alongside hand washing and checking the locks, as an item of the same kind.

A support group's ordinary currency is the answering of worried questions. That is what the room is for, and it is what any newcomer arrives wanting. The awkwardness is that the same exchange — a question asked, a kind and definite answer given — may be plain human support or the completion of a ritual, and which of the two it is depends entirely on what the asking is doing for the person asking. The distinction is invisible from outside the exchange and, on the clinical account, frequently invisible from inside it as well.

The literature recorded the same problem in households rather than newsgroups, under the heading of family accommodation: the changes relatives make to their own behaviour to help someone avoid or reduce distress — providing repeated reassurance, taking part in the rituals, or arranging things so that the feared situation is avoided. Family accommodation was first studied systematically in obsessive-compulsive disorder during the 1990s, the decade this group belongs to, and greater accommodation has since been associated with greater symptom severity and impairment. A group that invited family members and friends by name in its charter was, at least in principle, a room where the people on both sides of that arrangement were present at once.

This page states the tension as the literature states it and stops there. It makes no claim about how any particular message in this group was written, answered or moderated; it does not say what the right reply to a frightened question is; and nothing here is advice. It is worth noting only that moderation does not touch this problem at all. The mechanism described above filters conduct between posters. It has no bearing on whether an answer given in perfectly good faith is helping, which is the one editorial question a support room cannot delegate to a moderator.

Peer support without a clinician in the room

The group belonged to a category with a longer history than the web. Support groups are conventionally divided into two kinds. The self-help support group is fully organised and managed by its own members, who are commonly volunteers with personal experience of the subject; the professionally operated group is facilitated by professionals who most often do not share the problem — social workers, psychologists or members of the clergy — who control the discussion and provide the managerial services, and such groups are often found in institutional settings, sometimes run for a fixed period and sometimes charge a fee. This was emphatically the first kind, and it cost nothing to join. The internet had been serving as a venue for support groups since at least 1982, well before anything a general reader would recognise as being online.

What the medium added has been studied since. Work published in 2000 by Walther and Boyd found a common set of reasons people found online support appealing: the social distance between members reduced embarrassment; participants valued the greater range of expertise available in a large network; anonymity both decreased embarrassment and increased people's confidence in providing support to others rather than only seeking it; and being able to reread and edit a message before sending it, at any hour of the day or night, was itself part of the appeal. Participation was asynchronous, so nobody needed to be present at the same time as anybody else. Every one of those properties describes a newsgroup, and the moderated variant added one more — a filter applied before distribution rather than after — at the price of the delay described above.

What the arrangement did not include was any clinical oversight whatsoever. Nobody screened participants; nobody followed anyone up; nobody noticed when a regular stopped posting. A Usenet moderator is an editor of a feed, with a stated criterion about personal insults and a mailbox, and the paperwork claims nothing more than that. Neighbouring corners of the network handled the neighbouring questions — the theory and the professional arguments belonged to groups such as sci.psychology.psychotherapy, and the general help-seeking traffic to rooms like alt.psychology.help — but none of that put a clinician in this room, and the group never suggested otherwise.

The company it kept, and how such a group ends

The moderated group was a small node in a network of things built in the same fifteen years by roughly the same impulse. The patient organisation of the period rhymes with it almost exactly. The OCD Foundation, founded in 1986 and today the International OCD Foundation, grew out of a self-help group formed at Yale from people with the disorder who were taking part in research studies there; it was assembled by Wayne Goodman, then on the Yale faculty, who went on to be a principal author of the severity scale described above and who chaired the foundation's scientific advisory board for its first ten years. A gathering of people with the disorder, on the premise that they had something to tell each other, arrived by two independent routes within about thirteen years — once in a university clinic, once in a control message.

Around the newsgroups ran mailing lists, early personal web pages and, by the end of the 1990s, web forums. This directory holds no archive of that material and makes no claim about how the various rooms referred to one another; what the surviving files do show is that the two group names have travelled together ever since, sitting on adjacent lines of the newsgroups file that the consortium still distributes. That transition, and what became of the open room, belong to the sibling page rather than this one.

How a moderated group ends is a mechanical question with an unusually clean answer. Where the moderated flag is honoured, servers must reject any article that lacks an Approved header; so when the moderators of such a group stop reading the submission mailbox, nothing further appears. It does not fill with spam and drift into noise, as an abandoned open group does — it goes quiet, and the entry stays in the active file with nothing behind it. The exception is the hazard that the standard configuration file warns about: a group recreated unmoderated by somebody who wants the filter gone. This page cannot say which, if either, happened here, because the record it has does not reach that far. It can say that both names are still in the newsgroups and active files that the consortium distributes today, one flagged y and one flagged m — the namespace outliving the traffic by a comfortable margin, which on Usenet is the usual order of events.

For completeness, the open group's own file in the same archive shows how differently the two were made. Its creation messages date from mid-November 1994 and are extremely terse: a two-line newgroup carrying the group name and a one-line description, a second one from another site whose entire body announced a two-week trial period after which the sender would remove the group if it failed to take off, and a booster at the end of that month from an account at Nova Southeastern University in Florida, noting that the group had been discussed in alt.config, that it had received support for creation there, and that quite a few sites still were not carrying it. That last message also supplied the newsgroups-file line the earlier pair had not settled, and a version of it — For overcoming Obsessive-Compulsive Disorder — is the description the open group still carries. Five years separate those scribbled notes from the ninety-one line document that created the moderated companion, and the difference between them is a fair measure of what had happened to Usenet in between.

What the record does not show

The honest inventory of this page's ignorance is long, and stating it is the point of a directory rather than an embarrassment to one. The control archive preserves the paperwork of creation and nothing else.

• No archive of the group's articles is held here, and this page quotes none. Nothing below the charter — no thread, no exchange, no phrasing used by any poster — is known to this directory.
• No participant is named. The proposal names three proponents and initial moderators with their addresses; this page deliberately does not reproduce them, and the addresses it does give are the group's submission and board addresses, quoted as mechanism and long defunct.
• No traffic figures exist here: no article count, no subscriber estimate, no measure of whether the forecast of large traffic was met. The proposal's own count of a thousand sorted articles refers to the other group, and is the proposers' estimate rather than an independent finding.
• Whether the promised FAQ was ever written, whether the automated first pass at moderation was ever built, how the moderation team changed, and when the group stopped being moderated in practice are all unknown here. The control archive ends in January 2000; it records creation, not life.
• The alt.config discussion that the message says took place is not preserved in this archive. That it happened is the proposal's claim, made in a document sent to the group where it would have happened.
• Why the rmgroup of 24 December 1999 gave No Charter as its first reason is not explained anywhere in the file, and this page does not guess.
• Nothing whatever is known about outcomes for any reader, and no such thing could be known from a newsfeed.

Two limits on how this page should be read follow from that, and they are not disclaimers bolted on at the end but the terms on which it was written. The first is that everything above about diagnosis, drugs and therapy is historical description, attributed and dated: it records what was documented in the period the group belonged to, and it is not medical or psychological advice, not a recommendation of any treatment, and not a substitute for care from someone qualified to give it. The second is that this is a page about a newsgroup created in November 1999, not a service a reader can join. What the paperwork preserves is a small, specific, closely argued attempt to build a room in which people with a concealment-driven illness could talk to one another without being hunted, filed in the only place such an attempt could be filed at the time: a control message, addressed to every news administrator on the network, asking politely.