Brad Templeton Home
How to create a newsgroup on USENET (trial group method)
The HistoryThe trial system was proposed around April 1990 as an alternate way to try out and create new groups inside the principle 7 (now 8) hierarchies of USENET groups. It was meant to be an alternative to the flame-filled "voting" system which many feel has gone awry.
In mid 1991 a "vote" was held to give the trial system the power to create groups under its rules without votes. This vote passed. However, at the same time I moved and my business took off and I no longer had the time to follow through the management of such a scheme. As nobody else took up the mantle, it has sat in limbo for several years.
I encourage some other party to take this up or something like it. I think this has already been approved, however there are some who feel that since the approval was years ago, it might need to re-seek the approval of people on the net at large.
It may seem odd to implement a system that uses readership measurement rather than votes by using a "vote" but that seemed the best way to move on to this next stage.
A description of the instructions on how to create a group are included below. Some of the document needs updating.
Here is how the trial system works, in brief:
Creating a group on USENET
To create a group on USENET, you must objectively demonstrate to most USENET admins that the group under consideration is worth carrying on their machines, and thus by default to all of USENET.
One common method of performing this demonstration is to conduct a survey/vote. Another regular posting in this group describes that method.
A new method now being tried involves giving the group a trial run in a smaller section of the to see how it does. This is described below.
(These are just guidelines. Other methods exist, and these guidelines have been written to be flexible. The real goal is that sentence above -- an objective demonstation that it's worth feeding the group, by default, to all of the many thousands of machines on USENET. What you see below is just one way that many people think is a good way of doing that. Because USENET has no government, neither method forces administrators of net machines to create or carry the groups on their machines. The better the demonstration of the value of the group, the more likely it is that most will carry it.)
The *.trial.* hierarchy exists for new newsgroups that are being tried out. A new group can be created there, and read by readers of that subnet. The readership and other forms of response are then measured. At the end of the trial (up to 5 months) the readership of the group is evaluated, and it is calculated where it would fit into the whole of USENET. If it meets the criterion, it is moved into the USENET mainstream. If it doesn't it is simply deleted with a few weeks notice. Readers of the group may elect to form a mailing list or find another method of distribution.
Judging the suitability of the topic
If you wish to create a group, you should ensure that you have a topic that is likely to experience varied discussion for a long and indefinite period of time. The purpose of the group should be clear -- not too general (like, "the IBM PC") and not too specific (like, "squid recipes"). The topic should not be short lived, unless all you want is a 5 month run in the trial hierarchy. A general topic should be broken down until you have something that is important and likely to generate a moderate volume of discussion.
There are exceptions to this which you can only learn fully through experience with USENET.
Next check to see if there isn't already a group on USENET that covers your topic. If you find a reasonable match, read that group for a while to see if your topic gets discussed there. Bring it up yourself it you don't see it discussed for a while. (This is a strong requirement. You should not propose a new group if you have no familiarity with groups that might well already contain discussion of your topic.)
If it turns out your topic is an offshoot of an existing group, and it is in fact already heavily discussed in that group, you may wish to split that group. The trial group system has no mechanism for splitting regular USENET groups. You must arrange another means to do that.
If you have a truly new topic and:
Make sure as well that the topic is appropriate for world wide distribution. You're going to be sending this discussion over more than 20,000 machines. (In part, this is what the trial will decide if you aren't sure yourself.)
Starting the trial
Write up a proposal, listing:
(See other postings to news.trial.newgroups for a model proposal.)
The name and USENET hierarchy won't be chosen by you, but you can make suggestions. A good suggestion that is consistent with other existing names is more likely to be chosen. Note that a good name must be meaningful to outsiders who are unfamiliar with the topic, so stay away from acronyms or terms known only to insiders. Unless you want to be ultra-specific about what takes place in the group, a well understood name is better than an extremely precise one.
E-mail the proposal to email@example.com. This will send it to all of the volunteer trial group moderators, or "judges." In addition, the software will randomly pick one of these people to be the judge for your trial group. The judge will be a person with long experience with USENET, and he or she will offer you help if there is room for improvement in your proposal. Some further names may also be suggested. Joke proposals will not be accepted -- a trial newsgroup involves a fair bit of effort and expense by a lot of people, and no matter how funny a joke is, it won't be as long lived as the effort in running a trial group.
[ Don't use the 'select-trial-judge' address other than to get a judge assigned to your group. That judge will reply to you -- correspond with him or her directly thereafter. If you really *have* to mail to all the trial judges, and not the one(s) involved in your group, you can mail to firstname.lastname@example.org. ]
Once a trial name has been decided upon and the proposal finalized, the trial group will be created by the trial hierarchy judges. You should then post the finalized proposal to news.announce.newgroups (or mail it to the moderator at email@example.com) and news.trial.newgroups (Due to the nature of moderated groups, you can't crosspost; you must post twice, once to each group.) If there is a group with a related topic, you should also post the announcement of the new group there.
Now start using your group. Write a more detailed description of the group, and post it there. Welcome new users and start discussion. While (unless you're moderator) you won't own or control the group after creating it, you might see fit to act as a sort of custodian for the group, helping new users, preparing lists of frequently asked questions or an introduction to the group that gets posted every month or two. If you stop reading your group sometime in the future, you should find another volunteer for that position.
Many sites in the trial hierarchy will send in readership reports. If your site does not do this, look for the "arbitron" program and instructions posted at the start of every month in news.admin, and get your site involved. The arbitron results will be posted on the 1st of every month.
For the first 1-2 months, the results for your group will be artificially low, since it takes time for results to come in from a wide enough range of sites. Don't be concerned about those early figures.
After the 5th readership report to include your group, the day of judgement is at hand. All mainstream USENET groups, and your group, will be ranked according to the number of readers per site which gets the group. (Your group will only go to the trial subnet, so both numbers will be reduced, keeping the ratio valid.) "Mainstream USENET groups" means all the accepted groups in the 8 main hierarchies of USENET, less those that have been truly dormant for several months, in the judge's opinion.
If your group ranks among the top 4/5 of mainstream USENET groups in readers per site, it gets promoted to a mainstream USENET group. Right now (May/90) this means a readership value of about 1 reader for every 3 sites getting the group, but that may change as USENET grows. If your group can't find a reader on 2/3 of the sites it goes to, it's probably not an appropriate topic for full-net distribution.
Either way, pass or fail, mail a reminder of the results to your judge.
An announcement will be made (by you, or failing that, a trial hierarchy judge) indicating that the group failed the test. This gives you time to wrap up affairs, or consider the creation of a mailing list devoted to the topic. After 2-3 weeks the group will be deleted (rmgrouped). Any attempts to post to it will result in mail to the poster and the USENET admin at the poster's site.
If the group passes the test, it gets renamed. That means that a USENET group with an appropriate name will be created. Many sites will alias the old name to the new name. You should tell all readers of the trial group about the new name, and get them to switch over. You may crosspost between the two groups for the first week, but after that, you should actively discourage any crossposting between the groups. After a few more weeks, the trial group will be deleted, with a few days notice.
Post once again to news.announce.newgroups, indicating that the group passed and has now been created within USENET proper.
Do the same thing in the USENET group that you did in the trial group. Tell people about the group, and post any standard introductory postings that you may have written. Welcome the new readers. Then participate in the group.
If a group ever ranks in the top 50% of USENET groups in readers/site, it can, at your discretion, move over to USENET immediately. Mail to your judge and request the immediate move. This can happen at any time, but the group must get this ranking in the top 50% with results from at least half the trial hierarchy. (ie. if you only get 6 sites reporting in the first month, and they all have 10 readers, it doesn't qualify.)
Refusal of a trial
In rather unusual cases, the judges can refuse a serious trial. For example, an illegal group might cause this to happen. If the judges can't convince the proponent of the group that it's not a good idea, they may decide to register opposition to the trial. At the start of the trial -- prior to the group's creation, a vote from 2/3 of the judges (5 of the 7) can refuse the trial. At the end of the trial, a vote from 3/4 of the judges (6 of the 7) can stop the group's promotion.
Is this a horrible autocratic power? Not at all. If a trial or promotion is refused by the above veto, the proponent of the group is still free to use the old discussion/survey method of group creation -- ie. things fall back to the method we had before the trial hierarchy was created. Refusing a trial is a very serious move that judges will do only very rarely -- not simply because they don't like a group, but rather because they feel it could cause serious damage to the net. To get 5 or 6 judges to agree to a refusal will mean there's a real problem, indeed.
As described above, a trial name will be decided by the trial judges in conjunction with you, the group's "champion." They get the final say. If there is real feeling that the name is inappropriate, you can discuss this in your trial group. Proposals for a new name can then be put to the trial judges. This may result in a new name if the group is promoted to full distribution. The final decision, however, remains up to the trial judges. If they really miss the boat on a couple of names, they won't be trusted as trial judges for much longer.
You can suggest a hierarchy to the trial judges, but the decision is up to them. Here are the existing hierarchies:
As noted, any serious group proposal will be created as a trial. You may get some advice not to do so from the trial group judges, but if you insist, it will be done. Some sites, however, will refuse to carry quasi-illegal material, such as groups relating to sex, drugs, porn and other activities that are illegal or discouraged in some parts of the net. Their machines are theirs to command, so it is considered polite to warn everybody about a group with potential danger so that they can make their own decisions.
If your group fails, it is advised that you not suggest it again for at least a year, unless something really unusual comes up. It would take very rare circumstances for the trial judges to restart a trial in less than six months.
You can, after a trial fails, go through another group creation process, such as a discussion/survey. Nobody knows what will happen the first time somebody tries this. I predict that people won't take kindly to this, but who knows. (Some people may hate the trial system and support you only for that reason.)
Even after your group has a good trial, news administrators on their own machines are not bound to create, carry or propagate your group. The fact that it does well makes it pretty likely, but not assured, particularly at the leafs. This is true in any group creation system.
Deliberate attempts to bias the arbitron statistics are likely to get discovered. If you're caught and exposed, it's likely to do you more harm than good. If there's proof, the judges will cancel your trial, and even if there isn't it is likely that you will anger site admins enough that they ignore the group even if it appears to pass the test. A suspect passing result can be worse than a failed one -- so don't even risk it.
Running around making sure reader sites send in their honest readership reports will bias the results somewhat. This is accounted for, and in the long run, it's good, because those reports will keep coming, and they report on all trial groups, not just yours.
The rules above are deliberately vague to give the judges room to breathe, and to stop control freaks from pointing out nitpicking technical violations. A technical violation of this procedure is tough, because things are left open.
If you don't like the judges and how they rule, you can always use the discussion/survey group creation method.
The only rules the judges must follow are:
Other "good ideas" are:
Who are the judges?
There are 7. I, Brad Templeton, am the first "chief justice." The chief justice has no special powers, other than having written this document, but hey, the title sounds neat. The other 6 volunteers are all system admins, from a variety of areas on USENET, who have had several years experience with USENET and USENET groups.
All judges will pass on any decision to other volunteer judges if they have a conflict of interest. (ie. plan to read the group, or participate actively in a similar group.)
The judges really aren't all that powerful. The *real* decision as to whether a group gets created belongs with the readers. If they like the group, it stays, if they don't, it goes. The judges only get final say on the name, and they get to interpret the minor guidelines.
Their primary goal is to give advice, as experienced netters, on how
to make a group creation go smoothly. They aren't there to hinder the
process, or fight against it. They get the title "judge" because it sounds
important, and it means that people will avoid arguing with them over silly