Re: "Supersedes" in e-mail versus Usenet News
Klaus Weide (kweide@xochi.tezcat.com)
Fri, 3 May 1996 01:49:45 -0500 (CDT)
Jacob Palme wrote:
>
> Question: Is it allowed in Usenet News to "Supersede" someone else's
> message. Or is it like "Cancel", which I believe you can legally
> only do on your own messages.
It is like "Cancel".
Although, what one can "legally" do is in fact not so well defined,
and opinions vary...
Roughly in order of decreasing acceptability:
1) Cancelling one's own message
2) Cancelling by a newsadmin or sysadmin, of a message originating
from the same system.
3) Cancelling of "gateway spew", i.e. messages regurgitated en masse
because of some misconfigured gateway.
4) Cancelling by a group's moderator, especially of messages with
forged "Approved" headers.
5) Cancelling of "SPAM", i.e. multiple copies of the same message
posted to many groups (not crossposting), without regard of
content.
6) Cancelling of "binary" messages to "non-binary" groups
7) "Retromoderation" - a specific newsgroup is technically
unmoderated, but someone has been given the authority to cancel
unwanted articles.
8) Cancelling by anybody else for any reason not mentioned, including
content based, censorship, copyright claims,...
I believe only 1) and 2) are covered by RFC1036, but 3) and 4) are
also universally accepted. 5) is generally accepted (and appreciated)
nowadays, or so it seems, by most newsadmins and users - at least
those who care enough to know about it. Except of course those who
want to use Usenet as a free advertising medium, and a minority of
people who object on reasons of principle. The sender of the cancel
is supposed to be clearly identified by a header and/or in the body
of the cancel message.
6) is happening, but there seems to be less support than for 5).
7) is even less accepted, but being discussed for some groups.
8) is a big nono.
All this applies to Usenet as the collection of the world-wide
newsgroups; groups in other hierarchies (local, regional,
commercial,...) may follow other rules - cancelling may be less
restricted, or delegated to some authority. Messages may be cancelled
because they are crossposted to some group or hierarchies where they
are cancellable. The newsgroup news.admin.net-abuse.misc discusses a
lot of this stuff.
So what does all this have to do with "Supersedes"?
Well mostly I wanted to show that it is not always clearly defined
whether a message can be canceled, and the answer may depend on who
you ask. The same would apply to "Supersedes". But in general,
to supersede a message you would have to be authorized to cancel it.
Some news servers and most clients that implement cancels do some
plausibility checking to verify that only the sender of a message
can cancel it. The checking is based on "From" and, if present,
"Sender" headers. For news servers, this is generally configurable.
At least for INN I believe that the checks apply to "cancel" and
"Supersedes" in the same way. They would usually only allow case 1)
from my list, so cancel messages from the other categories either
are sent with a (technically) forged "From" (and probably "Sender")
or will not be honored by many servers.
To return to your question, a message from person A superseding
person B's message without trying to "impersonate" B in the headers
could be slightly more acceptable than A cancelling B's message with
forged headers. The point of such a superseding message could be
to get the "Supersedes" header to the reader for interpretation by
the client, rather then effecting superseding at the server level.
(I am thinking of a hypothetical scenario where mail and news user
agents would do something useful with such headers.) But it would
rely on all news servers on the path doing the checking described
above, which can not be guaranteed, and is not currently being done.
So basically, only the sender of a message may supersede it -
for most of the cases other than 1) where *cancelling* might be
regarded as acceptable, using "Supersedes" instead doesn't make
much sense.
Klaus