soc.culture.ethiopia.moderated

Long-form reference · 8,237 words · about a 36-minute read

A room whose paperwork survives

soc.culture.ethiopia.moderated was the filtered member of a three-name family: an open group, a catch-all and a moderated room, all bearing the name of one country. Three addresses for one national subject is unusual in the culture branch, where the great majority of countries were given a single group and kept it. An arrangement like that invites an explanation. The intuitive one — that the moderated room was opened when an existing open group’s political heat made a calmer room necessary — is what this directory assumed for years, and the surviving documents do not support it.

It can now be tested, because this group’s paperwork survives, and survives more completely than that of almost any other group in this directory. The Internet Systems Consortium mirrors the archive of news.announce.newgroups, the moderated group in which every Big Eight proposal had to be published, and the file kept there under the name soc.culture.ethiopia holds the whole sequence: a Request for Discussion of 15 January 1996 and an identical reposting the following day, a second Request for Discussion of 14 March, a third of 4 June, a first Call for Votes of 14 August, a second of 26 August, and a result posted on 8 September 1996 carrying the name and address of every person who voted and how each of them voted. The matching archive of Usenet control messages holds the messages that created — and repeatedly failed to create — the names involved. A second file, kept under the moderated group’s own name, holds the proposal that removed it in 2006 and the board decision that granted the request.

What those documents show is more specific than that assumption, and different from it in one respect that matters. They do not describe an existing Ethiopian newsgroup boiling over. They describe a community that had no soc.culture group at all in January 1996, proposing to create one — moderated from the first line of the first document — and then spending eight months negotiating what the rooms it wanted should be called. The causal claim is therefore not established here, and this article does not restate it. What is established is set out below, document by document, with the dates those documents carry.

The documents, and what kind of evidence they are

A word on what this evidence is worth, since the rest of the article leans on it. The Big Eight group-creation procedure of the 1990s required a proposal to be published in a moderated announcement group, discussed in public for a set period, put to a postal ballot conducted by a neutral third party, and then — if it passed — implemented by a control message sent by the moderator of the announcement group. The mechanism is described on the soc.* hierarchy page of this directory and is not retold here. Its by-product is a paper trail: dated, publicly posted, archived by third parties, and impossible to revise after the fact.

That trail is authoritative for some things and not for others. It is authoritative for what was proposed, when, by whom, in what words, and with what result: those are the documents’ own content. It is not authoritative for the state of the world the documents describe. When a Request for Discussion asserts that Ethiopian mailing lists carried more than a thousand subscribers between them, that is the proponent’s claim, made in a document written to win a vote, and nobody audited it. This article distinguishes the two throughout, and where it reports a claim it says whose claim it is.

One further caution. Everything below concerns the creation and removal of newsgroups. It says very little about what was actually posted inside them, because on that the record here is close to silent, and inventing a substitute would be worse than admitting the gap.

January 1996: a moderated group under the plain name

The first Request for Discussion is dated 15 January 1996 and was posted by Christopher Stone, writing from a Princeton University address, in his capacity as the proposal’s group-mentor — an experienced volunteer assigned to walk a first-time proponent through the procedure. An identical text, superseding it, went out the following day over the name of the proponent himself, Belete K. Muturo, at the University of Akron. Both carry the archive name soc.culture.ethiopia. The same Chris Stone was mentoring a proposal for a filtered Kuwaiti room in the same month, which is documented on this directory’s page for soc.culture.kuwait.moderated; the two proposals took quite different routes and are not otherwise connected.

The heading of that first document is the fact this page turns on. It reads, in the layout of the period, REQUEST FOR DISCUSSION (RFD) — moderated group soc.culture.ethiopia, and the opening sentence is equally plain: this is a formal Request For Discussion on the issue of creating a moderated Usenet newsgroup called soc.culture.ethiopia. Not converting one. Not supplementing one. Creating one, under the plain unqualified name that every other country in the branch already had, and creating it moderated.

The distribution tells its own story about where the proponents thought their neighbours were. The document was posted to news.announce.newgroups and news.groups, as the rules required, and beyond them to rec.travel.africa, soc.culture.african, soc.culture.african.american, soc.culture.italian, soc.culture.kenya, soc.culture.nigeria and soc.culture.south-africa. It was additionally sent to five mailing lists, and the proponent undertook to post pointers in fourteen further newsgroups, among them soc.culture.somalia, soc.culture.egyptian, soc.culture.arabic, soc.genealogy.african, sci.lang, the moderated soc.religion.christian and soc.religion.islam, and talk.politics.mideast. A community with no room of its own was advertising in everybody else’s.

The rationale, in the proposal’s own words

The case the proposal made for itself is worth reporting in detail, because it is the closest thing that exists to a contemporaneous description of the Ethiopian diaspora’s online life in 1995, and because it is not the case the intuitive account assumes.

It opened with an argument from existing traffic. There were, the document states, several well-established mailing lists dealing with issues related to Ethiopia and Ethiopians; together they had more than a thousand subscribers; and the number of messages sent across them averaged over thirty a day during November 1995. That is a specific, dated measurement of a diaspora’s message volume, made by someone inside it, three months before it was published. It is also unverifiable now, and it was unverifiable then — no third party counted those lists — so it is recorded here as what the proponents said, not as a fact about the world.

It went on to an argument from general interest, listing Ethiopia’s traditions and the diversity of its peoples, the country’s part in Africa’s anti-colonial movement, its role in the formation and maintenance of the Organization of African Unity, and its athletes’ Olympic marathon golds, as reasons why Usenet readers at large might want to know more. The country itself, its history and that record are the subject of the parent page, soc.culture.ethiopia, and are not retold here.

The third argument was the one that dates the document most exactly. Internet access, it says, will soon be available in Ethiopia, and an increasing number of software programs for Ethiopic languages are appearing on the market; the new group would therefore serve as an introduction to Usenet for such users. Read in 1996 that is optimism about connectivity and about the script. Read now it is a forecast about a country whose domestic connection rate would remain a fraction of one per cent for more than a decade — the numbers are on the parent page — and about a script that did not enter the Unicode Standard until version 3.0, in 1999. Neither disappointment is a criticism of the proposal. It is simply what a group founded on an expectation looks like from the far side of the expectation.

What the January rationale conspicuously does not contain is any complaint about an existing Ethiopian newsgroup. There was none to complain about.

The moderation policy as written

The proposal set out its charter as twelve numbered topic headings, and they are a fair summary of what an Ethiopian group in the middle 1990s expected to be for: culture, history, politics, geography and language; human rights concerns in the aftermath of the Mengistu regime, political parties, and Ethiopia’s role in decolonisation; the concerns of Ethiopia’s various nationalities and the prospects for ethnic harmony; impediments, real or perceived, to democratisation and economic development, together with the role of international lending institutions, donor states and non-governmental organisations, trade, exchange rates and investment law; literacy and folklore, literature, arts and music; religion; tourism; cuisine; education; sports figures; science and technology, expressly including computer technology, internet access and word-processing programs for Ethiopic and Amharic script; and Ethiopians living outside Ethiopia together with non-Ethiopians wanting to visit, live or invest there. Ethiopians of all ethnic backgrounds, the charter says, are welcome to participate.

The moderation policy attached to that charter is more distinctive than the charter itself. Moderation was to be carried by a panel rather than by one volunteer. Anyone could join it with the consent of a two-thirds supermajority of the sitting moderators and could be removed on the same supermajority; resignation was always available. Five categories of submission were to be rejected: articles not pertaining to Ethiopia or falling outside the charter; unrelated commercial advertisements and spam; “me too” articles, excessive repetition, contributions adding nothing to a thread, and articles quoting an unreasonable amount of previous text; articles containing profanity or excessive rudeness, or lacking appropriate civility and tone; and — the category that most repays a second reading — excessively ungrammatical articles, articles containing excessive misspellings, and articles exceeding a line-length limit, set at seventy-two characters in January and raised to seventy-five in the later drafts.

The policy then anticipated the obvious objection and answered it in the document itself. Moderation, it says, is by no means intended to impede freedom of speech, but merely to help readers make the most of their time by ensuring conformity to the objectives outlined in the charter. A poster who believed a moderator had rejected a submission unfairly could appeal to the whole panel, which would reconsider and decide by majority; and in the event of an even split, the submission was to be posted. A tie went to publication.

Five people were named as the January panel, each with a short biography supplied by the proposal: a doctoral student in education administration holding degrees in economics and secondary education; a programmer analyst; a postgraduate in structural mechanics who had been a commercial pilot with Ethiopian Airlines; a computer scientist described as the president of a firm called Ethiopian Computers, holding copyrights in Ethiopic word-processing software; and a medical student. The list is reproduced here in the proposal’s own terms because those terms are the evidence; nothing further about any of these people is established by this record.

Spring 1996: the name arrives by forgery

Between the January proposal and the vote, something happened to the plain name that the proposal had asked for, and it is the strangest part of the record.

On 1 April 1996, a control message bearing the timestamp 11:11:11 GMT and the name of David C. Lawrence — the moderator of news.announce.newgroups, and therefore the only person entitled to create a Big Eight group — instructed every news server on the network to create soc.culture.ethiopia, with the newsgroups-file description “All things Ethiopian in the broadest sense”. No vote had been held. No result had been posted. The proposal was, at that moment, still in its discussion phase.

Three days later Lawrence issued the first of many corrections, in a sentence he would repeat almost verbatim for the next two and a half years: the recent newgroup for this group in my name was a forgery, soc.culture.ethiopia is widely considered a bogus newsgroup given that it has not been approved by accepted Usenet group creation procedures, please remove the group from your active file. The archive holds eleven further copies of that message over the following nine weeks — on 5, 8, 11, 12 and 15 April, on 4, 5, 6, 8 and 15 May, and on 6 June 1996.

The forger, or someone else, then dropped the pretence. On 4 and 5 May three further newgroup messages for the same name were posted from an address at Netcom, a large American commercial provider of the period. One of them appends the line “Please note that I’m not David Lawrence”. Another carries a cheerful open letter to news administrators advising them to reconfigure their servers to honour every newgroup message and silently discard every rmgroup message, which would have made removal impossible — an unusually candid statement of what a control-message forgery is for.

The mechanism being abused is worth stating once, because it explains why a single forged message mattered. Usenet had no central register of groups. A group existed on a server because that server’s administrator, or that server’s software acting under standing instructions, had accepted a control message saying it should. Most administrators configured their machines to accept Big Eight newgroup messages automatically, on the strength of the signature they carried; a convincing forgery therefore propagated as far and as fast as the genuine article, and undoing it required persuading thousands of independent operators to act.

In all, the control-message archive for the name soc.culture.ethiopia holds seventy-two newgroup messages and thirty-seven rmgroup messages, running from 1 April 1996 to 29 November 1998. Every one of the rmgroups was sent by the moderator of news.announce.newgroups, and every one of them describes the newgroup that provoked it as a forgery.

March and June 1996: two more drafts, and three names

While that was going on in the control channel, the proposal itself went through two more drafts, and each changed the shape of what was being asked for.

The second Request for Discussion, dated 14 March 1996, keeps the charter and the moderation policy word for word, notes that one moderator had been removed from the panel and two added — the new pair being described as a historian who had served as a spokesperson for the United Nations in Ethiopia from 1987 to 1991 and had published a bimonthly newsletter on relief and development, and a doctoral candidate in finance who had lectured at Addis Ababa University from 1991 to 1994 — and adds a companion group. Beside the moderated soc.culture.ethiopia there was now to be an unmoderated talk.politics.african.ethiopia, whose one-line description read “Politics in Ethiopia”.

The March draft also adds the sentence the intuitive account was reaching for, and it is not quite the sentence that account supposes. It reads: unmoderated soc.culture.* groups suffer from an unacceptably low signal-to-noise ratio, and soc.culture.ethiopia will therefore be moderated to ensure that submissions conform to the group charter. That is a proposition about the branch as a whole, offered as a reason to build a filtered room in the first place. It is not a report on the temperature of an existing Ethiopian group, because there was no such group whose temperature could be reported.

The companion group’s charter is candid about its intended function: an unmoderated forum in which to debate the politics and political economy of Ethiopia, in which discussion of ethnic issues is on-topic, as are rejected postings from the moderated group. Posters are asked — asked, not required — to avoid excessive flamewars, namecalling and crossposting. Two rooms, one with a door, and an explicit undertaking that whatever the door kept out would have somewhere else to go.

The third Request for Discussion, dated 4 June 1996, produced the family this directory preserves, and it did so for a reason the document states plainly. The unmoderated companion had been proposed in the talk.* hierarchy; several prominent members of important Ethiopian mailing lists, the draft says, disliked the placement in talk.*, and the proponents had therefore compromised and chosen the name soc.culture.ethiopia.misc. Renaming the companion forced a rename of its partner: the moderated room, which had been asking for the plain name since January, became soc.culture.ethiopia.moderated.

The three-name family, in other words, is not the residue of a schism. It is the residue of an argument about hierarchy placement — about whether a diaspora’s political argument belonged under the label soc or the label talk — settled by moving both rooms one level down the tree. The catch-all form that resulted has its own page at soc.culture.ethiopia.misc, which is where that side of the story belongs.

The June draft also rebuilt the panel, which grew from six to ten. The biographies attached to it describe, among others, a returned Peace Corps volunteer who had served in Harar and was by then a software engineer at Digital Equipment Corporation and librarian of the Ethiopian Philatelic Society; a legal academic with degrees from the University of London and Addis Ababa University, then lecturing at Nanyang Technological University in Singapore; a doctoral student in information and software systems at George Mason University; a graduate student in mechanical engineering at the University of Saskatchewan; and a doctoral student in economics at the University of California, Irvine, described as very much interested in promoting understanding, through constructive dialogue, among Ethiopians. Ten volunteers, writing from addresses in the United States, Canada, Finland and Singapore, were being lined up to moderate a newsgroup that did not yet exist.

The vote, and how it was counted

The first Call for Votes went out on 14 August 1996 under the banner of the Usenet Volunteer Votetakers, the standing pool of neutrals who ran Big Eight ballots. The votetaker was David Bostwick, writing from a chemistry address at the Georgia Institute of Technology; a second and final call followed on 26 August; and the voting period closed at 23:59:59 UTC on 4 September 1996. Ballots were cast by e-mail. Both groups were on the same ballot paper, and a voter could vote differently on each.

The result was posted on 8 September 1996. Four hundred and sixty-nine valid votes were counted. soc.culture.ethiopia.misc passed by 313 votes to 126; soc.culture.ethiopia.moderated passed by 331 to 127. Both cleared the two requirements of the period — a margin of at least a hundred more yes votes than no votes, and at least two-thirds of the votes cast on the question in favour — and the result records both tests as met for both groups.

Eighteen further ballots were set aside, and the result lists each of them by address with the reason. Seven were rejected as invalid votes, five arrived with no ballot attached, four contained no vote statement, one carried no name, and one was a duplicate. The tally block at the head of the posting counts all eighteen as votes in error, while the votetaker’s note immediately beneath it mentions only the seven he had declared invalid; the itemised list at the foot of the posting reconciles the two figures.

The result then set out the standard five-day discussion period, in which allegations of voting irregularity could be raised before the groups were created. None is recorded in this file.

One feature of that result is worth setting out carefully, because it is easy to misread. The moderated room recorded more yes votes than the open one, 331 against 313. That gap is not a measure of preference between them. The published roll shows that exactly twenty-five voters said yes to the catch-all and no to the moderated room, and exactly twenty-five said the reverse: among the voters who distinguished between the two rooms, the two rooms were tied. The difference in the totals comes from partial ballots — twenty-five voters said yes to the moderated room while expressing no view on the catch-all, against seven who did the opposite. What the electorate did not do, on the evidence of the roll, was choose between the two rooms.

What the ballot roll shows about the electorate

Because the procedure of the period required the votetaker to publish the complete roll — every voter’s e-mail address, every voter’s name as given, and both of their votes — the shape of the electorate that created this group can still be examined. What follows is counted from that roll and from nothing else.

The 469 valid ballots came from 298 distinct e-mail domains. Of the addresses, 194 were in .edu and a further eighteen in the United Kingdom’s ac.uk; 51 were in Canada’s .ca, most of them university addresses; 90 were in .com; 21 in .net. Beyond the English-speaking world the roll shows twelve ballots from Finland, ten from Singapore, ten from New Zealand, nine from Germany, seven from the Netherlands, four each from Norway, Australia and Sweden, three each from South Africa, Spain and Austria, two each from France and Italy, and single ballots from Denmark, Japan, Hong Kong, Poland and Croatia. Two came from United States government addresses and one from a military one.

Two features of that distribution are worth stating carefully. The first is how institutional it was. The large consumer online services already carried Usenet by September 1996, yet seventeen of these 469 ballots came from AOL, CompuServe, Netcom and Prodigy combined, while ten came from community free-net services in Canada. The Ethiopian-interest electorate on Usenet was, overwhelmingly, people with a university or employer account.

The second is how concentrated it was. Forty-six ballots carried addresses in the University of Kansas system, thirty-three in the University of Akron’s, nineteen at the University of Saskatchewan, nine at Nanyang Technological University in Singapore. The proponent wrote from Akron and one of the proposed moderators from the Kansas medical centre. That pattern — a proposal carried by the institutional networks of the people proposing it — was entirely ordinary in Usenet ballots of the period, and the reason it can be observed at all is that the rules obliged the votetaker to publish the roll so that anyone could observe it. That was the point of publishing it.

One absence is worth recording. No ballot in the roll carries an address in the .et domain. The proposals themselves are the reason to expect that rather than to be surprised by it: their third argument for the group was that internet access will soon be available in Ethiopia — future tense, in 1996. The connectivity figures are set out on the parent page and are not repeated here.

Finally, the roll shows how the two questions were answered together. Of the 469 valid ballots, 281 voted yes to both groups and 99 voted no to both; fifty split their votes, twenty-five favouring the catch-all over the moderated room and twenty-five the reverse; thirty-seven voted on one question and abstained on the other; and two recorded no preference on either. Three hundred and eighty ballots, then, answered both questions the same way. The electorate was, in the main, voting for an arrangement rather than choosing between two rooms.

Creation, and the second forgery wave

The control-message archive records that soc.culture.ethiopia.misc was created on 16 September 1996 at 14:30 GMT, by a PGP-signed newgroup message from the moderator of news.announce.newgroups quoting the 313:126 result and reproducing the charter; the message was reissued on 23 September and again on 16 October, as was normal practice for servers that had missed it. For soc.culture.ethiopia.moderated the same archive preserves no creation message at all: the file kept under that name opens in November 2001. The group plainly was created — it is the subject of a formal removal by the Big-8 Management Board ten years later — but the message that created it is not in this archive, and this page will not supply a date it cannot show.

The plain name, meanwhile, was forged into existence a second time and on an industrial scale. Between 27 August and 18 November 1998, sixty-eight further newgroup messages for soc.culture.ethiopia were posted in the name of the moderator of news.announce.newgroups. Each carried the header line Organization: The Cabal. Each ended with the single word “Meow!” above the moderator’s name. And each cited a vote result — sixty-eight messages, sixty-eight different tallies, ranging from 56 votes to 11 up to 346 to 57, and including one implausibly unanimous 145 to nothing. Each attributes its tally to a report in news.announce.newgroups, on dates spread between 26 August and 19 November 1998. No such report exists. A genuine result has one tally.

The sign-off dates the episode. The Meow Wars were a Usenet harassment campaign that ran from about 1996 to about 1998, whose participants flooded newsgroups with high volumes of nonsense to swamp the conversation, and whose calling card was that word. This page records what the messages contain and when they were sent; it makes no claim about who sent them, and the archive does not establish it.

The consequence is still visible in the namespace. The master active file distributed from the Internet Systems Consortium’s Usenet configuration archive — a list of roughly forty-five thousand groups across every hierarchy it tracks — carries exactly one member of this family: soc.culture.ethiopia.misc. The plain name, the one the community asked for in January 1996 and never received by any procedure the Big Eight recognised, is not in it. Neither is the moderated room, which was removed on request in 2006. Servers carried soc.culture.ethiopia for years, archives hold its traffic, and this directory preserves a page for it; what it never acquired was a valid creation record.

The three-group family, seen from the record

Set against the documents, the unusual three-name shape resolves into something less dramatic and rather more interesting than a community splitting in two.

Two of the three names were voted into existence together, on one ballot, on one day, as a deliberate pair: a moderated room with a written constitution, and an unmoderated room whose charter promised in advance to accept whatever the moderated room turned away. Both were proposed by the same person, mentored by the same volunteer, and argued for in the same rationale. The third name — the plain one at the head of the family, the one this directory treats as the open group — is the one that was never voted at all. It arrived by forged control message in April 1996, was repudiated thirty-seven times by the moderator of news.announce.newgroups, and was forged into place again sixty-eight times in 1998. It is the only member of the family whose creation the Big Eight never recognised, and the only one for which no valid creation date exists to be recorded.

The branch’s other filtered rooms took other routes. Kuwait’s community, in the same season, chartered a moderated second room beside an existing unmoderated group and left that group untouched; the account of that pattern, and of the remedies the culture branch reached for generally, belongs to the page for soc.culture.kuwait.moderated. Singapore’s took a third route later the same year, described on the page for soc.culture.singapore.moderated. The Ethiopian case is distinctive within that small set for having tried to make the flagship name itself moderated, and for having ended up with a family whose senior member was an administrative ghost.

What the record does not establish about this group

Against all that documentation, it is worth being blunt about the size of the remaining hole.

• No traffic figure of any kind is established. There is no article count, no posts-per-day estimate, no subscriber number, and no readership survey for soc.culture.ethiopia.moderated at any point in its life.
• No moderator roster after 1996 is established except by inference. Ten people were named in the June proposal and reprinted in the September result; the only later name in the record is that of the volunteer who proposed the group’s removal in 2006 and described himself as its moderator.
• No creation date for the moderated group is established here, for the reason given above: the control message is missing from the archive that holds its sibling’s.
• Nothing about the content of the group is established. Not one thread, subject line, poster or argument from inside it is documented in the material consulted for this page.
• The directory’s editorial summary — that the room was created because the open group’s political heat made it necessary — is not supported. The record shows a moderated room proposed before any Ethiopian group existed, justified by a general statement about unmoderated groups in the branch.

What is established is the list at the head of this article: the dates, the drafts, the charter, the moderation policy, the panel as proposed, the vote, the roll, the forgeries, the removal. For a group of this size that is a great deal. It is not everything, and the difference is not padded over here.

The diaspora’s own network, as the proposal records it

The most valuable thing the 1996 file contains, apart from the vote, is an inventory of where the Ethiopian diaspora was already talking before Usenet gave it a room. The proposals were required to state where they had been distributed, and the distribution lists name the community’s own infrastructure.

Five mailing lists appear in the January and March drafts: a list called simply Ethiopia, at a host in the vm.its.rpi.edu domain at Rensselaer Polytechnic Institute; Ethiolist and Ethioforum, both at Netcom; Oromonet, reachable through a Netcom address; and AFRICA-L at a Virginia Tech address. The June draft adds three more: Afrique at the Université Claude Bernard in Lyon, an Ethiopia-politics list at a host in the apc.org domain, and the Nigerian network Naijanet at MIT.

The host names are themselves dating evidence. Addresses of the form vm.its.rpi.edu, vtvm1.cc.vt.edu and mitvma.mit.edu are IBM mainframes of the kind that carried the academic mailing-list software of the 1980s and early 1990s, when a scholarly list was likelier to live on a university VM system than on anything resembling a modern server. The apc.org domain belonged to the member networks of the Association for Progressive Communications, the international network of civil-society service providers founded in 1990, whose members were in many countries the first providers of internet access to non-governmental organisations. And Netcom was a commercial American provider selling dial-up accounts to individuals. Three generations of network provision, in one distribution list, all being used at once by the same community in the same year.

The one quantitative claim attached to that inventory has been given above and is repeated here with its warning label: more than a thousand subscribers across the lists collectively, and an average of over thirty messages a day in November 1995. It is the proponents’ figure. Nothing in the record corroborates it. It is nonetheless the only contemporaneous number of its kind that this page has been able to find for Ethiopian diaspora list traffic in the mid-1990s, which says something about how thinly that period was measured.

What researchers documented about diaspora media in the Horn

The Usenet groups themselves have attracted no scholarly attention that this page has been able to locate. The wider phenomenon they belonged to — a dispersed population conducting its argument about home through media it built and paid for itself — has attracted a good deal, and the findings can be reported without being adopted.

The most directly relevant study is Terje S. Skjerdal’s “Journalists or activists? Self-identity in the Ethiopian diaspora online community”, published in the journal Journalism in 2011 (volume 12, issue 6, pages 727 to 744). Skjerdal combined content analysis of selected Ethiopian diaspora websites with interviews of their editors, and asked how the sites related to conventional journalistic ideals. He reports that the sites were highly politicised and displayed a marked critical attitude towards the Ethiopian government through what he calls an activist journalism approach; that their editors differed among themselves as to whether activism conflicted with professional norms, some justifying it by conditions in the homeland and others treating the combination as a forward-looking ideology; and that the diaspora’s online initiatives functioned to prolong media contestations in the homeland. Those are Skjerdal’s findings about the sites he studied. They are set out here as a researcher’s account and neither endorsed nor disputed.

One strand of that ecology can be dated precisely because it began in print. The Ethiopian Review, a news and opinion magazine published in English and Amharic, was launched in 1991 by Hailu Indashaw as publisher and Elias Kifle as editor; the editorship changed in 1995; and in 2000 the print edition was discontinued and the title became an online magazine. A diaspora periodical founded in the year the Derg fell, still on paper through the years this newsgroup was being chartered, and web-only in the year the news2mail gateway began operating: the sequence is a reasonable proxy for where diaspora argument was living at each point.

For the period after Usenet, the standing academic reference is Iginio Gagliardone’s The Politics of Technology in Africa: Communication, Development, and Nation-Building in Ethiopia, published by Cambridge University Press in 2016, together with the Mechachal reports of the same year — “Mechachal: Online Debates and Elections in Ethiopia”, by Gagliardone with Matti Pohjonen, Abdissa Zerai and Zenebe Beyene — which analysed Ethiopian online political debate at scale. More recently the edited volume The Global Ethiopian Diaspora (2024), edited by Shimelis Bonsa Gulema, Hewan Girma and Mulugeta F. Dinbabo, includes a chapter by Gulema on diaspora, social media and partisan discourses. The literature exists, it is substantial, and it begins after the newsgroups it might have described had already emptied.

The neighbouring case in the literature: Eritrea online

The best-documented case of a Horn of Africa diaspora building its own public sphere online belongs not to Ethiopia but to Eritrea, and it is worth setting out because it is the nearest thing to a controlled comparison for what a filtered Ethiopian room might have been for.

The anthropologist Victoria Bernal has studied Eritrean diaspora websites for two decades. Her account appears in “Eritrea Goes Global: Reflections on Nationalism in a Transnational Era” (Cultural Anthropology, 2004), “Eritrea on-line: Diaspora, cyberspace, and the public sphere” (American Ethnologist, 2005), “Diaspora, cyberspace and political imagination: the Eritrean diaspora online” (Global Networks, 2006), the book Nation as Network: Diaspora, Cyberspace, and Citizenship (University of Chicago Press, 2014), and “Civil Society and Cyberspace: Reflections on Dehai, Asmarino, and Awate”, published in Africa Today in 2013 and appearing again in the collection Postliberation Eritrea in 2018.

Her findings, as she states them, are that the Eritrean struggle for national independence coincided with the rise of the internet; that Eritreans abroad used it as a transnational public sphere in which to produce and debate narratives of history, culture, democracy and identity; that through the web the diaspora mobilised demonstrators, raised funds during war, debated the drafting of the constitution and sought to influence the government at home; and that what she calls internet intellectuals used their postings to interpret national crises and construct community, making the medium an emotionally charged and creative space rather than merely an informational one. Diasporas online, she concludes, may invent new forms of citizenship, community and political practice.

Two cautions attach to using that here. Bernal writes about Eritreans, not Ethiopians, and about websites, not newsgroups; the two diasporas are distinct, their politics are distinct, and the relationship between them during the years in question was the subject of a war whose account belongs to the parent page. What the work establishes for present purposes is narrower and sufficient: that in exactly the region, exactly the decade and exactly the medium this newsgroup occupied, a diaspora public sphere is a documented phenomenon rather than a supposition, and one whose participants took it seriously enough to argue in it about constitutional drafting and to raise money in it during a war.

Why a filtered room was a plausible thing to want

Nothing in the record says why any individual voted for soc.culture.ethiopia.moderated, and this page will not guess. What can be set out is why, in 1996, a diaspora community proposing its first newsgroup might reasonably decide to put a door on it before ever opening it — and that can be anchored in the documents and in dated public events rather than in imagination.

Begin with the documents, which are unusually explicit. The proponents’ own stated premise was that unmoderated groups in this branch suffered from an unacceptably low signal-to-noise ratio. Their charter’s list of admissible subjects put human rights concerns in the aftermath of the Mengistu regime at number two, the concerns of Ethiopia’s various nationalities and the prospects for ethnic harmony at number three, and impediments to democratisation at number four — that is, the three most contested questions in the country’s recent history were on-topic by design, in a room whose whole purpose was to be calm. The companion group existed precisely so that what the panel rejected would still be published somewhere. This was not a community proposing to avoid its arguments. It was a community proposing to hold them in two places at once.

Second, the timing. The June 1996 draft was written between the promulgation of Ethiopia’s federal constitution in 1995 and the outbreak of the border war with Eritrea in 1998; the vote was taken in the middle of that interval. The dated chronology of those years is the parent page’s subject and is not reproduced here. What matters for this page is only that the questions the charter listed as on-topic were, at the moment of chartering, live and unresolved, argued by people who disagreed about them profoundly and who had no other public forum in common.

Third, the structural point, which the literature supports and which is general rather than Ethiopian. A diaspora forum concentrates in one room people who are geographically dispersed, politically divided about the same distant events, unusually motivated, and mostly unable to meet. Bernal’s Eritrean material documents such a forum being used for fundraising during a war and for constitutional argument; Skjerdal’s Ethiopian material documents diaspora outlets that were, in his account, highly politicised and understood themselves as prolonging contestations at home. A community that expected to be described in those terms, and asked for a moderator before opening its first room, was acting on that expectation. Whether moderation would have answered it is not something this record can show.

Fourth, and least often noticed, the ordinary reasons. Three of the five rejection categories in the Ethiopian moderation policy have nothing to do with politics at all: spam, “me too” articles, and over-quoting. The middle 1990s were the years in which industrial-scale commercial spam arrived on Usenet, and an unmoderated national group with a small, dispersed readership was among the least defensible targets on the network. Part of what a filtered room bought, in 1996, was simply an inbox somebody was watching.

The general account of the culture branch’s conflicts and the four remedies available to it — the reader’s kill file, the charter, the namespace and the moderator’s inbox — is given on soc.culture.kuwait.moderated and deliberately not repeated here.

What moderation meant here

The mechanism — submissions mailed to a moderator, returned to the network only with an Approved header, and propagated by servers only if that header is present — is described on the soc.* hierarchy page, and the moderator’s working day, the procedure and the spam argument belong to soc.culture.singapore.moderated. Only what is peculiar to this room is added here.

Only one thing about this room could not have been done from outside it. The charter put word-processing programs for Ethiopic and Amharic script among its own listed topics, and the script in question had no place in the standard character sets of the period: Ethiopic entered the Unicode Standard only with version 3.0, in 1999, three years after the vote. The parent page sets out that problem in full. A moderation policy that rejected excessive misspellings and off-charter argument, applied to a readership writing about a politics whose terms were themselves contested, asked for judgements a volunteer from elsewhere in the hierarchy would have been poorly placed to make. That is the readiest explanation for a panel of ten people drawn from inside the community, governed by a supermajority rule and a right of appeal, rather than one caretaker with a mail folder; the proposals themselves do not give a reason.

1999 to 2006: the last decade of the room

What happened to the group after 1996 is known from one document, and it is candid to the point of bleakness.

On 22 May 2006 a Request for Discussion was published proposing to unmoderate or remove soc.culture.ethiopia.moderated. It was formatted by Tim Skirvin for the Big-8 Management Board, the body that had taken over group creation and removal from the mass ballot the previous year, and the change history at its foot records that the original e-mail had reached the board on 1 May. The proponent is named as Martin Tierney, described as the group’s moderator, and the rationale is given in his own words.

He had agreed to moderate and administer the group, he writes, after being asked to do so by the person responsible for it when the hosting arrangements changed; he set it up on a colleague’s machine “around 1999 I think”, with that colleague’s permission. It had never been a very successful group. The person who had asked for his help had been difficult to contact even in the early days. There had not, he says, been more than a handful of active posts in all that time, and the only postings were an automated mailing list that had carried no content for some while, together with autoposted spam. He had removed the mailing-list feed in the hope that a group falling to zero posts would prompt the original organiser to make contact. It had not. This, he concludes, is most definitely a dead group.

Under the heading EXISTING CHARTER the document says: none listed. Ten years after a 469-vote ballot had approved a charter running to twelve numbered topics and a five-part moderation policy, the board tidying up the group had no copy of it on file.

The board agreed on 26 May 2006, the result was posted the same day, and the removal control message was issued on 1 June 2006. One incidental figure survives in the headers of those postings: the cross-reference line of the May 2006 proposal, as filed by a Stanford news server, shows that server having assigned article number 3,796 in soc.culture.ethiopia.moderated. That is a local counter for one machine across the group’s whole life, including spam, cancelled articles and everything else; it is not a measure of readership or of genuine traffic, and it is quoted here only because it is the single quantity of that kind anywhere in the record.

Two earlier removal attempts sit in the same archive and were not the board’s. On 4 November 2001 rmgroup messages for both soc.culture.ethiopia.moderated and soc.culture.ethiopia.misc were posted from a private address, each asking administrators to remove “the bogus newsgroup”. Neither was issued by the moderator of news.announce.newgroups, and neither succeeded: the moderated group was still there to be removed by due process five years later, and the catch-all is in the master newsgroups list to this day.

One coincidence of dates is worth recording without labouring. The news2mail gateway operated between 2000 and 2004 — which is to say, across the period its own moderator would later describe as containing an automated feed, some spam and almost nothing else. The gateway carried what the group roster said existed. So does this directory.

Afterlife: what became of the family

Of the three names, one survives in the Big Eight namespace. The master active file distributed from the Internet Systems Consortium lists soc.culture.ethiopia.misc and neither of the others. The moderated room was removed in June 2006 by the board, at its own moderator’s request. The plain name was never validly created and cannot therefore be validly removed; it simply persists on such servers as still carry it, and in the archives.

The argument the family was chartered to hold did not stop; it moved, in stages that can be dated. The first stage was the web, and the Ethiopian Review’s abandonment of print in 2000 marks it as neatly as anything: a diaspora title founded on paper in 1991 becoming a website in the year Usenet’s decline was already well advanced. Skjerdal’s 2011 study is a survey of what that stage had become by the end of the following decade — a set of diaspora-run news sites with editors, readerships and an explicit sense of their own position relative to the media at home.

The second stage was satellite television, which put the diaspora’s output back into households inside the country in a way no newsgroup ever could. Ethiopian Satellite Television dates its own establishment to 24 April 2010 and is headquartered in Washington, DC; it was founded by exiled journalists and is funded largely by donations from the diaspora. The Oromia Media Network, launched in 2013 and headquartered in Minneapolis, was formally inaugurated on 1 March 2014, was founded by Jawar Mohammed, and is funded on the same model. Both were among the outlets the Ethiopian government charged in absentia, and the charges against both were dropped in 2018 — those against the Oromia Media Network and its executive director on 29 May of that year. Both are described in the sources cited here as independent outlets funded by diaspora donors; this page takes no view on any of the disputes in which either has been involved.

The third stage was the social platforms, and it is the stage the research finally caught up with: the Mechachal reports of 2016 analysed Ethiopian online political debate on a scale no study of the 1990s ever attempted, precisely because by then the debate was happening somewhere with an accessible interface and a great many more participants.

Across all three stages the constraint that shaped the newsgroup persisted longer than the newsgroup did. Ethiopia’s telecommunications remained a state monopoly, and the OpenNet Initiative, the academic filtering-measurement partnership, classified the country in October 2012 as engaging in pervasive filtering in the political sphere. Those are the findings of that project as reported; the parent page carries the connectivity figures. The practical effect for the story told here is that a diaspora medium remained, for a quarter of a century after 1996, a substantially different thing from a domestic one — which is what a soc.culture group about a country with almost no domestic internet access was in the first place.

Scope, sources and limits

This article is about a newsgroup and its paperwork. Its primary sources are the archive of news.announce.newgroups and the archive of Usenet control messages, both as mirrored on the Internet Systems Consortium’s Usenet file server, together with the master newsgroups and active files distributed from the same place. Every date, tally, charter clause and quoted phrase attributed to a proposal, a call for votes, a result or a control message above was read in those files. The ballot statistics — the domain counts, the split votes, the institutional concentrations — were recounted from the published roll in the 8 September 1996 result and are stated as counts, not as estimates.

The secondary literature cited is named in full in the text so that it can be checked: Skjerdal in Journalism (2011); Bernal in Cultural Anthropology (2004), American Ethnologist (2005), Global Networks (2006), Africa Today (2013) and Nation as Network (2014); Gagliardone (2016) and the Mechachal reports (2016); and The Global Ethiopian Diaspora (2024). Where their conclusions appear here they are attributed to their authors and are not adopted as this page’s own.

The limits have been stated as they arose and are collected here. Nothing is established about the group’s traffic, its readership, its moderators after 1996, or any article ever posted in it. Its creation control message is missing from the archive that holds its sibling’s. The causal account this directory formerly gave — that the moderated room answered an existing group’s political heat — is not supported by the record, has not been repeated as fact, and the description now standing at the head of this page has been corrected to match the documents.

Finally, the subject matter. The charter of this group made the constitutional settlement of the 1990s, the position of Ethiopia’s nationalities, the aftermath of the Derg and the prospects for democratisation explicitly on-topic, and every one of those remains a matter of live and serious political disagreement. Several of the people named in the documents above took positions on them; the sources cited advance interpretations of them. This page documents that the arguments happened, in what forum, under what rules and on what dates. It takes no side in any of them, offers no verdict, and is not a source on the merits of any question its subjects were arguing about.