soc.culture.misc

Long-form reference · 8,526 words · about a 37-minute read

A group with no birth certificate

The Internet Systems Consortium mirrors the archive of news.announce.newgroups, the moderated group through which Big-Eight proposals, ballots and results were published, and it files the traffic by group name: one file per proposed name, holding the Request for Discussion, the Call for Votes, the result and any later administrative notice. Under the soc directory there are 193 files whose names begin soc.culture. Asking that archive for soc.culture.misc returns nothing at all — a plain 404. There is no proposal, no charter draft, no ballot and no result for this group anywhere in the record of how Big-Eight groups were made.

Part of that absence has a mundane explanation, and establishing it is the first job of this page. The archive’s running series of New Groups Being Created postings begins on 18 February 1991, and that first instalment lists four groups: rec.arts.bodyart, carried 171:59; rec.music.video, 157:52; sci.geo.geology, 266:11; and soc.culture.magyar, 164:36. The companion archive of Usenet control messages describes itself, in its own policy document, as holding messages “received since early 1991”, notes that it was originally kept at rpi.edu until the autumn of 1992 before moving to uu.net and later to isc.org, and admits that “a month or so of messages (around August 1992) are missing”. A group already in existence when those two series begin leaves no creation record in either of them.

The by-name files, however, reach back further than the announcement series does, and they carry the earliest documentary trace of this group. The oldest posting in any of the 193 soc.culture files is dated 21 January 1990: a result announcement for soc.culture.british, cross-posted to eleven groups, and one of the eleven named in its Newsgroups header is soc.culture.misc. Four more postings from the first three months of 1990 do the same — the Call for Votes for soc.culture.vietnamese on 8 February, the calls for soc.culture.french and soc.culture.german on 27 February, and the call for soc.culture.african-american on 27 March. A Newsgroups header is not by itself proof that a group existed on the servers that received it, but five independent posters routing traffic to the name in the first quarter of 1990 is as close to a birth announcement as this group has.

The control archive holds a file for the name, and it is short enough to describe in full. It contains four messages. The first is dated 23 May 1991 at 17:40:47 GMT, sent from a news account at Compass, Inc. of Wakefield, Massachusetts, posted to soc.culture.misc.ctl with a world distribution, and its entire body is one line:

Group for discussion about other cultures.

It would be easy, and wrong, to read that as the group’s creation. The same archive holds a message from the same account, fourteen seconds earlier, for soc.culture.magyar — the group whose 164:36 ballot had been announced three months before and which had been newgrouped from rpi.edu on 18 February 1991. The Message-IDs give the game away: soc.culture.british is 6044, soc.culture.canada 6045, soc.culture.indian 6054, soc.culture.japan 6056, soc.culture.jewish 6057, soc.culture.magyar 6061, soc.culture.misc 6062, soc.culture.nepal 6063. A contiguous run of numbers, in alphabetical order, emitted between 17:37:01 and 17:41:02 on a Thursday afternoon. This is a site sweeping its active file and re-announcing every group in it, not a person creating one. What the message proves is that the sentence quoted above was already this group’s description in May 1991, and it has never been amended since.

The remaining messages in the file are unremarkable and worth listing only because they are all there is. On 26 April 1992 a machine at the University of Maine’s computer science department issued a newgroup with the distribution umcs — local to that campus. On 21 October 1992 a system administrator at Colorado State University did the same with the distribution csu, the body reading chkgroup soc.culture.misc. Both are housekeeping. The fourth and last message, dated 4 November 2001, is an rmgroup issued under the name of Urs Janßen, who has maintained the tin newsreader since 1996, from a host in Quebec; its operative line reads “please remove the bogus newsgroup soc.culture.misc”. It did not take effect, and it was not aimed at this group in particular: rmgroups under the same name and with the same wording went out against soc.culture.ethiopia.misc at 00:32 GMT, against this group at 00:38, against soc.misc at 00:49 and against soc.singles.moderated at 00:51, a run of about twenty minutes that was not confined to .misc names. All three of the other groups are still in the master list today.

Nothing else. No moderator, no charter, no vote, no proponent, no date of creation. The group is older than the paperwork.

It is not alone in that, and the company it keeps is the point. Of the 139 groups the current master list carries under soc.culture, exactly 36 have no file in the proposal archive. Thirty-two of those 36 appear in the earliest checkgroups control message the Consortium’s archive holds, dated 10 January 1993, which names 54 soc.culture groups in all. That message is a checkgroups issued by a private site rather than an act of the Big-Eight authority, and it carries group names and moderation flags rather than descriptions; what it records is the branch as a well-connected server saw it at the start of 1993. In other words, the groups without paperwork are, almost without exception, the founding cohort: the rooms that were already standing when the record-keeping began. The other four have ordinary explanations, being companions and renamings whose paperwork is filed under a sibling name — soc.culture.ethiopia.misc under soc.culture.ethiopia, and the three Pakistani subgroups under soc.culture.pakistan-reorg.

The list of that cohort is a fair picture of which peoples had a standing room on Usenet at the start of 1993: Afghanistan, Africa as a continent, the Arab world, ASEAN, Asian America, Australia, Bangladesh, Brazil, Bulgaria, the Caribbean, the Celtic countries, China, the Philippines, Hong Kong, India, Italy, Japan, Jewish culture, Korea, Latin America, Lebanon, Hungary, Mexico, Nepal, the Netherlands, Portugal, Romania, Sri Lanka, Taiwan, Turkey, the United States — and the room for everybody else.

Today the group’s two lines in the Consortium’s configuration files, both read for this article on 25 August 2026, are as spare as they were in 1991. The active file carries soc.culture.misc 0000000000 0000000001 y — placeholder watermarks and a flag meaning unmoderated and open to posting, containing no information whatever about traffic. The newsgroups file carries the same description line, thirty-five years on.

The word that does the work

The line shown at the head of this page under a charter style is not a charter. It is the newsgroups-file description, the one-sentence gloss that the master list carries for every group, and for this group it is the whole of the written scope that has ever existed. There was no proposal to write a charter in, and none was ever supplied afterwards. The sentence has therefore had to do, alone, for thirty-five years, which makes its wording worth reading closely.

Its construction is the branch’s house style. soc.culture.indian is glossed “Group for discussion about India & things Indian.” The Greek room is “Group about Greeks.” The Filipino room is “Group about the Filipino culture.” The formula is identical; the noun changes. Where every other line names a country, a language or a people, this one names a relation: other.

That word is written from a particular vantage point — the point of view of a reader running an eye down an alphabetical list in which the named rooms are already named. It presupposes the list. A culture is “other” here not because of anything about the culture but because of an absence elsewhere in the same file. The description is, in effect, a pointer to the complement of the branch, and it is the only line in the branch that could not have been written before the branch existed.

The convention that puts a .misc at the foot of a Big-Eight family is a real and documented one, and it is set out at length on the page for soc.culture.ethiopia.misc, which is this directory’s worked example; it is not restated here. What matters at this address is the noun the sentence chose. The catch-all at the root of the social hierarchy, soc.misc, is described as “Socially-oriented topics not in other groups.” Topics. This one says cultures. The difference between those two sentences is the difference between a room for leftover subjects and a room for peoples who did not have one, and it is the subject of the rest of this page.

The branch built other rooms that were wider than a country, and every one of them is named for something rather than for the remainder. soc.culture.african is “Discussions about Africa & things African.” soc.culture.maghreb is “North African society and culture.” soc.culture.nordic is “Discussion about culture up north.” soc.culture.celtic enumerates: “Irish, Scottish, Breton, Cornish, Manx & Welsh.” soc.culture.native is “Aboriginal people around the world.” soc.culture.pacific-island is “Culture of the pacific islands (except Hawaii).” That parenthesis records a separate ballot rather than an opinion: soc.culture.hawaii had passed 337:68 on 21 October 1995, three days before soc.culture.pacific-island passed 123:13, and the description was written to keep the two apart. Each of those rooms is an umbrella with a name on it. soc.culture.misc is the one with no name on it, and it therefore sits at the top of a structure the branch built repeatedly and never quite finished: an umbrella above the umbrellas.

The arithmetic of getting a room

A Big-Eight newsgroup was created by a public ballot, and the ballot had a threshold that had to be cleared twice over. The wording that the volunteer votetakers put at the top of every result posting through the 1990s is the most useful statement of it, because it is the form in which proponents actually met the rule:

For group passage, YES votes must be at least 2/3 of all valid (YES and NO) votes. There also must be at least 100 more YES votes than NO votes.

The guidelines archived alongside the proposals, in the revision of 7 March 2003, put the same requirement in slightly different words: a proposal passes “if and only if it received at least 100 more YES than NO votes and received at least twice as many YES as NO votes”. Twice as many yes as no is the same arithmetic as two-thirds of the valid total, and the reader should not mistake the two formulations for two separate hurdles. There were two hurdles, but they were the margin and the ratio. The Big-Eight process as a whole — its origins, its committees and its afterlife — belongs to the soc.* hierarchy page, while what the machinery felt like from inside a scattered community belongs to the culture branch page. Neither is retold here.

What belongs here is the property of the rule that decided which peoples got a room: the hundred-vote margin was indifferent to population. It did not ask how many people spoke a language or lived under a flag. It asked how many of them held an e-mail account, could be reached during a voting window the guidelines fixed at twenty-one days, understood a ballot form, and cared enough to return one to a stranger. A diaspora with a few thousand graduate students in North American and European universities could clear it in a fortnight. A community of millions with a few dozen accounts in the whole country could not clear it at all.

The ratio requirement cut in a different direction. It could not be met by finding more supporters if the opposition was also organised, and it gave any sufficiently motivated body of objectors a veto that no amount of enthusiasm could overcome. Both conditions had to hold. A proposal could therefore fail with 115 votes in favour and 16 against, and it could fail with 1,732 votes in favour. The archive contains both.

Around the arithmetic sat a procedure with its own costs. A Request for Discussion had to be posted to the moderated announcement group and discussed in news.groups for a minimum of twenty-one days. A Call for Votes then had to be prepared and run by a neutral third party, in practice a member of the Usenet Volunteer Votetakers, who collected ballots by e-mail for a fixed twenty-one-day period and published the full roll of voters with the result. Five days were then allowed for objections. And the result postings carried a standing sanction that shaped everything a losing proponent could do next: “Unless serious allegations of voting irregularities are raised, the group may not be voted on again for six months.”

The asymmetry in that procedure is the reason so many names in the archive stop where they do. Writing a Request for Discussion costs an evening. Winning a Call for Votes costs a campaign — mailing lists to find, student associations to persuade, an address book to work through twice, and a fortnight of chasing people who meant to vote and forgot.

Counting the shelf

The shape of the resulting namespace can be measured rather than characterised, because the Consortium distributes the master newsgroups file — the canonical list of every group it tracks, rebuilt hourly — and it can simply be counted. The copy read for this article, distributed on 25 August 2026, lists 45,003 groups, of which 139 sit under soc.culture. The checkgroups message of 10 January 1993 listed 54. The branch therefore rather more than doubled across the decade of the mass ballot, and then stopped.

Counting how much of the world those 139 names cover requires a rule, and the rule used here is stated so that it can be checked. A group is counted as naming a state if any component of its name is that state’s name or the ordinary adjective for its people — so soc.culture.magyar counts for Hungary, soc.culture.filipino for the Philippines and soc.culture.malagasy for Madagascar — with soc.culture.czecho-slovak counted generously as covering two states and soc.culture.korean counted once. Groups named for peoples, territories or defunct states that are not present UN members do not count, which excludes soc.culture.taiwan, soc.culture.hongkong, soc.culture.palestine, soc.culture.puerto-rico, soc.culture.soviet and soc.culture.yugoslavia. On that basis the branch names 85 of the 193 member states of the United Nations. Rather more than half the world’s states have no room of their own in the largest branch of soc.*, and never did.

The unevenness is regional, and three of the regions have denominators that can be quoted exactly. The United Nations regional groups put 54 member states in the African Group and 33 in the Latin American and Caribbean Group; the sovereign states of Oceania number 14. For Europe and Asia no denominator is quoted here, because the two depend on where Georgia, Turkey and Russia are placed; the counts below place Georgia and Russia in Europe and Turkey in Asia, and a reader who moves any of the three will move one state from one line to the other.

• Africa: 12 of the 54 states in the African Group are named — Algeria, Egypt, Ethiopia, Guinea, Kenya, Liberia, Madagascar, Nigeria, Sierra Leone, Somalia, South Africa and Zimbabwe. Above them sit soc.culture.african and soc.culture.maghreb, and beside them soc.culture.berber and soc.culture.south-africa.afrikaans.
• Latin America and the Caribbean: 17 of 33 — Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Cuba, the Dominican Republic, Ecuador, El Salvador, Haiti, Honduras, Mexico, Nicaragua, Peru, Uruguay and Venezuela — plus soc.culture.latin-america, soc.culture.caribbean and soc.culture.puerto-rico.
• Oceania: 2 of 14 — Australia and New Zealand — plus the single regional room, soc.culture.pacific-island, and its parenthesis excepting Hawaii.
• Asia: 27 states named, from Afghanistan to Vietnam, plus a dense secondary layer of rooms for peoples and territories rather than states: Arabic, ASEAN, Assyrian, Bengali, Hong Kong, Kurdish, Palestine, Punjab, Taiwan, Tamil, Hmong, and six regional groups beneath soc.culture.indian.
• Europe: 25 states named, plus eleven further rooms for peoples, languages and regions that cross or sit inside borders — Asturian, Baltic, Breton, Catalan, Celtic, Cornish, Galician, Nordic, Occitan, Scottish and Welsh — a general room for the continent, soc.culture.europe, and two named for states that no longer exist, soc.culture.soviet and soc.culture.yugoslavia.
• North America: Canada and the United States, plus soc.culture.quebec, soc.culture.native, soc.culture.african.american, soc.culture.asian.american and soc.culture.mexican.american.

Read as a map, the density follows the location of well-connected universities in the early 1990s with some precision. Western Europe and North America are saturated to the level of the minority language. South and East Asia and the Middle East are well covered at state level and subdivided below it. Latin America is covered at roughly half. Africa is covered at rather more than a fifth, and the branch built almost all of it inside three and a half years: soc.culture.african was already standing in January 1993, and every other African room whose result posting survives was decided between June 1993, when soc.culture.maghreb passed 199:28, and November 1996, when soc.culture.guinea-conakry passed 154:20. The one gap is soc.culture.malagasy, whose second Call for Votes went out in May 1995 and whose result is not in the archive, though the group exists. Central Asia is absent entirely: there is no room for Kazakhstan, Uzbekistan, Turkmenistan, Tajikistan or Kyrgyzstan, and the nearest name in the branch is soc.culture.mongolian. The Pacific, fourteen states across a third of the planet’s surface, has two national rooms and one shared one.

The conclusion to draw from that is narrower than it looks, and drawing it carelessly would misdescribe what happened. The pattern is not a ranking. Nobody sat down and decided which peoples deserved a room; there was no committee with a map. What the pattern records is which communities contained, in a given three-week window between about 1992 and 1999, at least a hundred more people with working e-mail willing to say yes than there were people willing to say no, all of them found and persuaded inside three weeks. That is a fact about the distribution of university accounts, international student populations and academic network links in the 1990s, and about who had the time and the address book to run a campaign. It is not a fact about the cultures.

The point can be tested against the branch itself, and the branch supplies the counter-examples in both directions. Rooms for regions of a few hundred thousand or a few million people, well supplied with accounts, produced some of the branch’s largest turnouts: soc.culture.asturies passed 790:44 in February 1996, soc.culture.scottish 723:23 in May 1995, soc.culture.breton 400:25 in January 1996, soc.culture.cornish 255:31 the next day. In the other direction, countries with very little domestic connectivity but a large and well-organised student diaspora won comfortably: Nigeria’s room passed 267:17 in July 1994, South Africa’s 409:23 in November 1994, Zimbabwe’s 211:23 in August 1995. Connectivity at home was not the variable. Reachable, organised readers were.

The failures the archive keeps

The single most useful property of the news.announce.newgroups archive, for the question this page is about, is that it files by proposed name rather than by outcome. A proposal that lost still has a file. The result posting that announced the loss was published in the same moderated group, in the same format, with the same tally and the same roll of voters as a result that won, and it has sat in the archive ever since. The record of who did not get a room is therefore as complete as the record of who did.

Across the 193 soc.culture files there are 49 distinct result postings announcing a failure. Forty-five of them state a numeric tally; the other four report reorganisations or paired ballots in prose — the Macedonian ballot of May 1993, the Tibetan pair of October 1993, the Punjabi pair of March 1996 and the china reorganisation of June 1995. Of the 45, the arithmetic divides as follows.

• 25 cleared the two-thirds test and lost on the hundred-vote margin alone. These are proposals with clear, sometimes overwhelming majorities of those who voted, defeated by the number of people who did not.
• 3 cleared the hundred-vote margin and lost on the ratio. These are proposals that found their hundred supporters and met organised opposition.
• 17 failed both tests.

Set out in the order the results were posted, with the tally and the date each posting carries, the complete roll of soc.culture.* ballots that ended in a numeric defeat reads as follows. Every figure is transcribed from the result posting itself; where a posting used a date in North American local time, that is the date given. Three entries — the Mongolian ballot of May 1997, the New Zealand ballot of September 1997 and the Nordic ballot of October 1997 — were proposals to add moderation to a room that already existed rather than to create one; the archive marks each of them “moderates existing group”.

• soc.culture.indian.american — 85:173, 21 May 1992
• soc.culture.usa.appalachian — 104:44, 19 August 1992
• soc.culture.cambodia — 96:21, 10 August 1993
• soc.culture.muslim — 246:156, 14 April 1994
• soc.culture.hmong — 67:37, 16 May 1994
• soc.culture.indonesia.religion (moderated) — 148:111, 31 May 1994
• soc.culture.fijian — 124:28, 14 December 1994
• soc.culture.saudi — 78:21, 21 February 1995
• soc.culture.jewish.singles — 586:388, 7 March 1995
• soc.culture.indian.bihar — 125:33, 3 May 1995
• soc.culture.hungarian — 157:91, 23 May 1995
• soc.culture.saudi — 73:23, 18 October 1995
• soc.culture.gaucho — 136:37, 18 October 1995
• soc.culture.centralamerica — 97:22, 13 November 1995
• soc.culture.uganda — 98:23, 9 December 1995
• soc.culture.indian.jammu-kashmir — 82:50, 26 March 1996
• soc.culture.kashmir (moderated) — 453:1993, 20 June 1996
• soc.culture.sindhi — 348:197, 27 June 1996
• soc.culture.israel.moderated (moderated) — 130:51, 31 July 1996
• soc.culture.ghana (moderated) — 95:29, 16 August 1996
• soc.culture.indian.muslims — 132:1492, 8 October 1996
• soc.culture.laos.moderated (moderated) — 84:22, 19 October 1996
• soc.culture.uganda (moderated) — 115:16, 14 November 1996
• soc.culture.manx — 87:14, 12 December 1996
• soc.culture.azerbaijan — 1732:920, 12 January 1997
• soc.culture.punjab.moderated — 113:70, 12 March 1997
• soc.culture.kashmir — 169:312, 20 March 1997
• soc.culture.malta (moderated) — 92:23, 10 April 1997
• soc.culture.mongolian (proposal to moderate an existing group) — 83:30, 4 May 1997
• soc.culture.cameroon (moderated) — 98:20, 23 July 1997
• soc.culture.somaliland — 52:22, 23 July 1997
• soc.culture.new-zealand (proposal to moderate an existing group) — 95:24, 29 September 1997
• soc.culture.usa.south (moderated) — 120:37, 29 September 1997
• soc.culture.nordic (proposal to moderate an existing group) — 132:45, 22 October 1997
• soc.culture.charnego — 91:617, 18 February 1998
• soc.culture.uganda (moderated) — 51:15, 1 April 1998
• soc.culture.padania — 178:94, 6 May 1998
• soc.culture.iberica — 78:18, 24 July 1998
• soc.culture.italian.emigrants — 44:23, 17 December 1998
• soc.culture.manx — 114:19, 3 February 1999
• soc.culture.padania — 191:136, 18 March 1999
• soc.culture.english (moderated) — 77:87, 27 March 1999
• soc.culture.andalucia — 57:32, 23 April 1999
• soc.culture.padania — 219:207, 13 December 1999
• soc.culture.valencian — 109:21, 10 December 2002

Thirty-five distinct names under soc.culture carry at least one result posting recording a defeat and have no group in the master list today; two of those thirty-five, soc.culture.indian.american and soc.culture.tibet, also appear in a passing result, the first from an earlier ballot and the second from a paired vote in which the companion group passed. A further thirty-two names have one or more Requests for Discussion in the archive and nothing after them. Between them that is 67 soc.culture names for which somebody wrote a document, and which do not exist.

The other side of the ledger should be stated in the same breath, because it is what makes the failures legible. African national rooms did pass, and the ones that passed did so comfortably: soc.culture.maghreb 199:28 in June 1993, soc.culture.somalia 190:28 in May 1994, soc.culture.berber 247:30 in July 1994, soc.culture.sierra-leone 156:30 in May 1995, soc.culture.algeria 197:18 in August 1995, soc.culture.liberia 133:14 in October 1995, soc.culture.kenya 158:16 in December 1995, soc.culture.guinea-conakry 154:20 in November 1996. Compare Liberia’s 133 yes votes with Uganda’s 115. The distance between a room and no room, in that part of the branch, was eighteen people.

Ninety-nine

Two proposals in the culture branch failed by exactly one vote, and the coincidence is worth stating precisely because it is the clearest single illustration of what the hundred-vote margin actually did.

On 14 November 1996 the votetaker Jeremy Billones, writing for the Usenet Volunteer Votetakers, posted the result of the second ballot on a Ugandan newsgroup. The posting reads: “There were 115 YES votes and 16 NO votes, for a total of 131 valid votes. There were 6 invalid ballots.” Eighty-eight per cent of those who voted were in favour, which cleared the two-thirds test with room to spare. The margin was ninety-nine. The group was not created.

Thirteen months earlier, on 18 October 1995, the votetaker Michael Handler had posted the result of a ballot on soc.culture.gaucho, a proposal for a room devoted to the gaucho and the traditions of the southern cone. There were 136 yes votes and 37 no votes, for a total of 173. Seventy-nine per cent in favour; two-thirds cleared; margin ninety-nine. The group was not created.

Two more came within single figures of the line. soc.culture.fijian took 124 yes votes to 28 no on 14 December 1994 — a margin of ninety-six — and soc.culture.manx, on its second attempt, took 114 to 19 on 3 February 1999, a margin of ninety-five and a yes share of nearly eighty-six per cent. The Isle of Man had, and has, a room of sorts: it is listed inside the enumeration in soc.culture.celtic’s description. Fiji has soc.culture.pacific-island.

The Ugandan proposal is worth following through all three of its attempts, because the sequence shows the six-month bar and the exhaustion it produced. The first ballot closed on 9 December 1995 and lost 98:23, under a proponent named in the result as Itiakorit John, Osele. The second, a year later, lost 115:16. The third, announced on 1 April 1998, drew 51 yes votes and 15 no. Dr Frederick I Njuki is named as proponent on the later two, and is also named as the group’s prospective moderator. The rationale reproduced in the ballot papers makes a numerical claim about its own constituency that is as good a summary of the problem as anything written since:

There is over 3,000 Ugandans in the country and living in the Western world who have access to electronic mail and the internet who are keen to keep in touch using electronic media.

Three thousand reachable people, of whom a hundred and sixteen would have had to vote yes, against the sixteen who voted no, in a particular three-week window, for the group to exist. On the second attempt one hundred and fifteen did.

The Cameroonian proposal met the same wall with the same kind of evidence, and the evidence grew as the proposal aged. Its first Request for Discussion, in January 1997, cited a mailing list, CAMNET, with 400 subscribers; the second, in April, said over 500; the Calls for Votes that summer said over 700. The ballot closed on 23 July 1997 at 98:20. The Ghanaian proposal of August 1996 cited two existing lists, okyeame and ghana_think_tank, whose membership it described as “by and large, restricted to Ghanaians”, and lost 95:29. The Maltese proposal of April 1997 opened by observing that “Malta has had full internet access for the past year”, and lost 92:23. None of these communities was disorganised. All of them were, in the years the votes were held, simply not yet numerous enough online.

The other way to lose

The hundred-vote margin was not the only thing that could stop a group, and a page that reported only the near misses would give a false picture of the branch. A proposal could also be voted down, and the culture branch produced the largest defeats in the archive.

The largest ballot the branch ever ran was not one of them. On 24 April 1996 the moderator of news.announce.newgroups published the result of the vote on soc.culture.indian.jammu-kashmir: 20,470 yes votes to 1,642 no, from 22,112 votes the posting calls “(apparently) valid”, with 251 invalid ballots. The roll of voters was so long that the result could not be posted to the group at all; the moderator’s covering note says it “would have been nearly five times as large as any other article ever posted to the group”. An earlier announcement, on 16 April, had already set out the reasoning for honouring it: “By receiving over four times as many as votes as any proposal has seen before, and ten times as many votes as usual, this vote can clearly be declared, ‘extraordinary.’” The same name had lost a ballot 82:50 a month before, on 26 March 1996.

The largest ballot that ended in defeat, and the second largest in the branch, came nine months later. On 12 January 1997 the votetaker Jim Davis announced the result of a proposal for a moderated Azerbaijani newsgroup: 1,732 yes votes and 920 no votes, from 2,652 valid ballots. The margin was 812 — eight times what the rule required. The yes share was 65.3 per cent, and two-thirds is 66.7. The proposal failed by about one and a half percentage points after mobilising more voters than most Big-Eight groups will ever have readers.

Two others cleared the margin and lost the ratio: soc.culture.jewish.singles, 586:388 on 7 March 1995, and soc.culture.sindhi, 348:197 on 27 June 1996. Both had their hundred and more; neither had two thirds.

At the other extreme are the ballots where opposition simply outnumbered support. On 20 June 1996 a proposal for a moderated Kashmiri group drew 2,446 valid votes and failed 453:1,993; a second, unmoderated proposal failed 169:312 on 20 March 1997. A proposal for soc.culture.indian.muslims failed 132:1,492 on 8 October 1996. A proposal for soc.culture.charnego — the newsgroups line offered was “The Spanish-speaking community in Catalonia”, and the proposed name is a term that Spanish and Catalan usage records as pejorative or descriptive by period — failed 91:617 on 18 February 1998, in a ballot whose votetaker also recorded that “a total of 47 ballots were voided because closer examination showed that they were solicited by a mass mailing”, a practice the voting guidelines prohibited.

soc.culture.padania was put to the electorate three times, and the archive holds the tallies: 178:94 in May 1998, 191:136 in March 1999, and 219:207 in December 1999. Support rose by forty-one votes across nineteen months; opposition rose by a hundred and thirteen. Each successive campaign brought out more additional voters against the proposal than for it, which is a thing the two-thirds rule was designed to detect and did.

One ballot in the branch split the difference in an unusually legible way. On 29 October 1993 a single vote covering two proposed groups closed with 1,266 valid votes and 49 invalid. talk.politics.tibet passed 905:268. soc.culture.tibet, proposed as moderated, failed 626:630 — four votes on the wrong side of parity. The same electorate, on the same day, created a room for the politics and declined a room for the culture. The reasons were argued at the time in news.groups; this page records only the numbers.

These are the cases in which the namespace behaved as a franchise rather than a bottleneck. Nobody was short of voters. The question of whether a name should exist was put to whoever turned up, and whoever turned up answered it.

The proposals that never reached a ballot

Thirty-two names under soc.culture have a file in the proposal archive containing a Request for Discussion, sometimes a second or third revision of one, and nothing further: no Call for Votes, no result, and no group in the master list today. Two of the thirty-two are hierarchy reorganisation proposals rather than single new rooms. They are the quietest entries in the record and, for the question this page is about, among the most informative, because the point at which they stop is the point at which the process began to cost real effort.

The list includes soc.culture.corsican, soc.culture.icelandic, soc.culture.wales, soc.culture.hellenic, soc.culture.england, soc.culture.polynesia, soc.culture.virtual.oceania (proposed first as soc.culture.oceania in October 1994 and renamed a week later), soc.culture.china.shanghai, soc.culture.latgola, soc.culture.ventin, soc.culture.taino and soc.culture.indigenous.taino, soc.culture.lang.bengali, soc.culture.sign-language and soc.culture.hazara.

The last of those is the clearest specimen of the genre, and it is late enough to show the process near its end. On 3 June 2003 a proponent writing from an address at hazara.org posted a Request for Discussion for a moderated newsgroup for the Hazara people. The rationale runs to four sentences, and its first three do all the work of this article in miniature:

Hazaras are a tribal ethnic group of central Afghanistan, with a population between 6 to 8 million. There are no newsgroups dedicated to Hazaras, the closest match being soc.culture.afghanistan. However a large fraction of Hazaras with Internet access are from Pakistan, Iran and Denmark and their discussions are not directly related to the country of Afghanistan.

A population of several million; a diaspora spread across three countries in a pattern that cut across the branch’s national grid; a charter drafted, a moderation policy written, a newsgroups line proposed. The file ends there. No Call for Votes was ever issued, and there is no soc.culture.hazara. The document names precisely the situation the group at this address was, by its description, the room for.

A Request for Discussion is the one step in the Big-Eight procedure that a single person can complete alone in an evening. Everything after it requires a neutral votetaker’s time, a three-week campaign, and a hundred more yes votes than no votes. That thirty-two soc.culture.* proposals stopped at exactly that boundary is not a mystery; it is the shape of the boundary.

Persistence, and what it bought

The hundred-vote margin measured organisation rather than numbers, and organisation could be built. Several communities lost a ballot, went away, and came back with a bigger address book.

soc.culture.cambodia failed 96:21 on 10 August 1993. Thirteen months later, on 19 September 1994, the same name passed 312:22 — the yes vote had more than tripled while the no vote stood still. soc.culture.hmong failed 67:37 on 16 May 1994 and passed 179:23 on 31 October 1996. In both cases nothing about the community had changed except how many of its members could be found and asked.

The traffic went the other way too. Saudi Arabia’s room was proposed twice within eight months and did worse the second time, 78:21 in February 1995 and 73:23 in October. The Isle of Man improved from 87:14 to 114:19 across twenty-six months and still fell five votes short. Uganda’s three attempts ran 98, 115 and 51 yes votes; the third campaign, in 1998, could not reassemble even the electorate of the first.

The six-month bar on re-voting after a failure, stated at the foot of every result posting, made the rhythm of these campaigns a matter of the calendar as much as of enthusiasm, and by the turn of the century the calendar was working against everybody. soc.culture.valencian, in December 2002, took 109 yes votes to 21 — a margin of eighty-eight — and failed. The last soc.culture ballot the archive holds is soc.culture.indian.goa, which passed 460:16 on 9 April 2003. Every soc.culture result posting after that date is of a different kind: from May 2006 onward they record decisions to remove a group or to lift its moderation, taken without a general electorate at all.

Routes that did not go through a ballot

A community that could not win a Big-Eight ballot was not out of options, and the two alternatives left very different traces.

The first was to go to alt.*, where a group required no vote, no votetaker and no threshold — one correctly formatted control message and, by convention rather than rule, a mention in alt.config. The current master list carries 126 groups under alt.culture. This directory holds one of them, alt.culture.zambia, whose entire control-archive file is a single newgroup message of 22 August 1998; that page sets out how the mechanism worked and what the resulting paperwork looks like, and it is not repeated here. What is worth adding at this address is that proponents in the Big Eight were aware of the trade-off and wrote it into their own proposals. The Maltese Request for Discussion of 1997 mentions it in passing: “Recently another newsgroup, alt.culture.malta, was successfully created but the propagation of new alt groups remains a difficult obstacle.” A group nobody’s server carried was not much better than no group. alt.culture.malta is, as it happens, one of the alt.* rooms that did eventually acquire both a description in the master list and a moderator; alt.culture.zambia is one of the many that carry the words “No description.” to this day.

The second route was to send the newgroup message yourself and see whether it stuck, and the culture branch furnishes a pair of cases that are best read together, since they concern neighbouring communities and ended the same way.

On 1 April 1996 a newgroup control message for soc.culture.armenian reached the network purporting to come from [email protected] — David C. Lawrence, then the moderator of news.announce.newgroups and the person who actually issued Big-Eight newgroups. Its body carried a properly formatted description line, “Armenia, its culture and history”, and a paragraph stating that anyone could create a newsgroup in any Usenet hierarchy. Three days later a message from the real address said otherwise:

The recent newgroup for this group in my name was a forgery.

What followed is preserved in the control archive as a war of attrition. The file for that name holds 81 newgroup messages and 40 rmgroup messages, the earliest dated 1 April 1996 and the latest 29 November 1998: two and a half years of a name being created and removed on news servers by people who could not agree that it should exist. A Request for Discussion was eventually posted, on 23 September 1996, and it is the only posting in that name’s proposal file; no Call for Votes followed it.

The Azerbaijani case took the other order. The ballot came first and was lost on 12 January 1997 by the margin described above. Ten days later, on 22 January 1997, a newgroup control message for soc.culture.azerbaijan went out over the name and address of Adil Baguirov, one of the three proponents named in the result posting and the first of the five people the proposal had put forward as moderators. Lawrence issued rmgroups on 27 January, 3 February and 27 February 1997, and the control file holds nothing else.

Both names have pages in this directory — soc.culture.armenian and soc.culture.azerbaijan — and neither is in the master list today. This page records what the control messages say and on what dates; the arguments they belong to are not its subject.

The scale of that third route can be measured too. The control archive holds files for 296 distinct names beginning soc.culture. The master list carries 139. Every one of the 139 has a control file; 157 names have a control file and no group. Some of those 157 are groups that were properly created and later removed, and many are the residue of forgeries, sprays and disputes of the kind described above. The namespace was never a register of what existed. It was a record of what had been asserted, and of which assertions the world’s news administrators had chosen to honour.

What a general culture room was actually for

Two functions of this group can be documented from outside it, and both are worth setting out before the harder question of what was posted in it.

The first is that the branch used it as a notice board. Of the 193 soc.culture files in the proposal archive, 47 contain at least one posting whose Newsgroups header names soc.culture.misc, and across the archive well over a hundred postings do. These are Requests for Discussion, Calls for Votes and results, cross-posted to the general room because their proponents wanted the attention of people interested in culture rather than in any one culture. The Fijian result of December 1994 went there. So did the result that created soc.culture.intercultural in November 1995, which was addressed to news.announce.newgroups, news.groups and soc.culture.misc and to nothing else. The practice runs from the beginning of the archive’s soc.culture material in January 1990 through to 1995 and then stops almost completely: of the postings so addressed, all but a handful are dated 1995 or earlier. That is roughly the period in which the branch was growing fastest and a reader who wanted to know what new rooms were being proposed had somewhere general to look.

The second is that comparative and cross-cultural experience was recognised, on Usenet, as a subject in its own right, and it was eventually given rooms of its own. soc.culture.intercultural passed 148:26 on 24 November 1995. Its rationale opens with the argument for why such a subject cannot live in a national room, and puts it more crisply than a summary would:

Intercultural issues are separate from issues that pertain to any one culture. There exist certain experiences that a person who is half X, half Y undergoes that are: different from somebody who is X or somebody who is Y; similar to somebody who is half A, half B. Therefore intercultural issues do not belong to newsgroups that pertain to just one culture.

The same document notes, as evidence that a readership existed, that these questions were already being discussed in “most soc.culture.* newsgroups at one time or another”. Its charter listed the topics it expected: self-identification, conflicts within families, rejection by one side or the other, the law of official ethnic categories and dual citizenship, demography, research resources, and the business of finding local organisations that dealt with the same thing. Alongside it the hierarchy already carried soc.couples.intercultural — “Inter-cultural and inter-racial relationships” — and, at a different angle, soc.culture.native, “Aboriginal people around the world”, an umbrella defined by a shared position rather than a shared place.

Those documents establish that the subject was real and that people organised around it. They establish nothing about what was posted in soc.culture.misc, and this page is not going to pretend otherwise. What can be described, carefully and as a matter of genre rather than of evidence, is the kind of writing that had no other address in the branch’s architecture. A room whose description names cultures in general, in a namespace where every particular culture with enough voters had its own room, is by construction the place for four things.

• Comparison itself. A question of the form “how is this done where you are” is off-charter in every national room and on-topic in none of them, because it is addressed to all of them at once. Before the general room, the only way to ask it was to name four groups in the Newsgroups header and hope.
• Emigration and return. The mechanics and the aftermath of leaving — and, harder to place, of going back — belong to two countries at once and therefore to neither room. The culture branch’s national groups were written largely by people abroad, a geometry the culture branch page describes; the general room is where the condition rather than the country could be the subject.
• Third-culture upbringing. Children raised outside their parents’ country, or between two of them, share a great deal with each other and comparatively little with either national group, which is the argument the intercultural proposal made in 1995 and won.
• Belonging to a group too small or too dispersed to fill a room. This is the case the description line was written for, and by 1999 the arithmetic set out above had produced a great many such groups: the peoples of every failed ballot and every abandoned Request for Discussion, and the very much larger number who never got as far as writing one.

Translation is the fifth, and it sits slightly apart because it was a practical problem before it was a topic. Usenet was for most of its life a seven-bit medium in which a great many of the world’s writing systems could not be transmitted at all, and the improvised romanisations that resulted were argued out group by group. A question about how to render a name, a script or an idiom belongs to whoever has solved it, which is not usually the people in one’s own room.

All five are genres, offered as a description of what the branch’s structure left over. No thread, subject line, poster or date from inside this group is documented anywhere consulted for this article, and none is invented here.

What the record does not show

The gaps on this page are specific, and collecting them is more useful than hedging every sentence above.

No creation date. The group’s name is in the Newsgroups header of a posting dated 21 January 1990, it has a control message of 23 May 1991, and it appears in the checkgroups message of 10 January 1993. Nothing consulted here establishes when, by whom, or under what procedure it was created. It is older than the proposal archive’s announcement series and older than the control archive, and there is no reason to expect the date to be recoverable from those sources.

No charter, and no proponent. There is no charter document for this group, only the one-line description. There is no named proposer, no discussion period, no ballot and no roll of voters. The line at the head of this page is the description; it is not, and has never been, a charter in the sense the Big-Eight procedure used that word.

No traffic figure of any kind. No article count, no posts-per-day estimate, no subscriber number, no readership survey and no activity profile for any period. The active file’s watermarks are placeholders. Article numbers visible in the Xref headers of archived postings are server-local sequence numbers on one machine and are not counts of anything.

No content. No thread, subject line, poster, argument or FAQ from inside the group has been located. The standard box on this page links to a public web archive of Usenet; this article has not surveyed it and makes no claim about what it holds for this group.

No date of dormancy. The group is in the master list today and its name is documented from 1990. Nothing consulted shows when, or whether, it stopped carrying traffic.

No causal account of the failed ballots. This page reports tallies, dates and the rules under which the tallies were assessed. Why a particular hundred people did or did not vote in a particular fortnight in 1996 is not in the archive, and any explanation offered here would be invention. The correlation between the branch’s density and the distribution of network access in the 1990s is visible in the numbers; the individual case never is.

Scope, sources and limits

The primary sources for this article are four bodies of material mirrored by the Internet Systems Consortium: the archive of news.announce.newgroups, arranged by proposed group name, from which every tally, date, charter fragment and rationale quoted above is transcribed; the archive of Usenet control messages, from which this group’s own four messages and the Armenian and Azerbaijani sequences are taken; the archived checkgroups messages, from which the list of 10 January 1993 is read; and the master newsgroups and active files, in the copies distributed on 25 August 2026, from which every count in this article was computed — the 45,003 groups, the 139 under soc.culture, the 126 under alt.culture, the 85 states, the 193 proposal files, the 49 failed results, the 32 proposals that never reached a ballot, the 296 names in the control archive. None of those figures is taken from a secondary source, and all of them can be recomputed from the same files. The group-creation rule is quoted from the votetakers’ own result template and from the guidelines archived alongside the proposals in the revision of 7 March 2003.

Two of the counts depend on a rule that a recomputing reader would have to adopt in the same form, and both rules are stated where the count is made: the 85-state census turns on treating a group as naming a state when any component of its name is that state’s name or the ordinary adjective for its people, and the count of defeated and abandoned names turns on the presence or absence of a result posting in the file. The count of individual postings cross-posted to this group is sensitive to how repeated and superseded copies are treated, which is why the figure given in the text is the number of files — 47 — rather than the number of headers.

Quotations from proposals, charters, rationales and control messages are verbatim, including their punctuation and their occasional grammatical slips, or they are not made at all; the one exception is the intercultural rationale, whose original bullet points are run together here with semicolons. Where a proposal’s rationale states a figure — three thousand Ugandans with e-mail, four hundred and later seven hundred subscribers to a Cameroonian mailing list, six to eight million Hazaras — that figure is the proposer’s own claim, presented as the proposer presented it, and this page has not independently verified any of them.

The neighbouring pages of this directory own the neighbouring subjects and this article has left them there. The .misc convention across the Big Eight, its origins, its arithmetic and its doctrine belong to the page for soc.culture.ethiopia.misc. The culture branch as a story — its net.nlang ancestry, its naming fights, its wartime years and its decline — belongs to the soc.culture.* branch page. The Great Renaming, the group-creation machinery and the mechanics of moderation belong to the soc.* hierarchy page. The alt.* alternative belongs to the page for alt.culture.zambia.

On subject matter. Several of the ballots reported here concern peoples, borders and names that are the subject of serious and continuing disagreement, and in some cases the disagreement is exactly what the tally records. This page states what was proposed, on what date, in what words, and how many people voted which way. It takes no position on any question those proposals raised, offers no verdict on any argument conducted in news.groups or in the groups themselves, and is not a source on the merits of anything its subjects were arguing about. Where a proposal’s own language is quoted, it is quoted as evidence of what was said at the time and not as a description of the world.

On length and shape. This article is long for a group with no paperwork of its own because almost all of it is about the paperwork of everybody else — the ballots that decided who would need this room and who would not. Where the material ran out, as it does entirely in the two sections above on content and traffic, the article says so and stops rather than continuing.