The news scene is in constant flux. It never looks exactly the same, in terms of which newsgroups and articles are available, ant any two points on the globe. It is tremendously succesfull, partly because it is completely decentralised and redundant.
Any article is allocated to one ro more news group, so news groups, as well as articles are objects for which names are useful. Newsgroup names are well defined,
comp.infosystems.www alt.hypertextmeaning in these examples, "Coimputing: Information Systems: World-Wide Web and "Uncontrolled Alternative Groups: Hypertext".
For the world-Wide Web, newsgroup names are simply preceded by "news":
news:comp.infosystems.www news:alt.hypertext
Here, then, is an example of a scheme which provides retrieval with no hints, from a name which is persistent, but for which an implementation exists which relies on the objects being transitory.
The world-wide web addresses articles using their message ids, again prefixed with "news". The fact that message-ids must contain at least one "@" character makes them distiunct from newsgroup names.
Unfortunately, some NNTP servers do not in fact support retrieval by message-id evn though it is mandatory in the NNTP specification. In these cases, articles may be retrieved by newsgroup name and article number. Unfortunately, as different NNTP servers get articles at different times, the article numbers vary from server to server. The newgroup/article-numebr approach can only be used as a reference very locally.
(up to existing schemes , on to gopher)
Tim BL