Cascading aliases

I think Matthew Rapaport's point is good that much of the trouble with the
automatic assignment of aliases comes from the automatic mailing to the user
of a new alias.
 
Most of the remailing servers have a special address or command you can send
meaning "assign me a new alias, and tell me what it is".  Probably, as
Matthew says, they should only mail back the newly assigned alias when one
of these special commands is used.
 
I'm still not convinced that automatic alias assignment should always
be done when mail goes through a server from a new address.  It seems like
this might generate so many aliases that it would be too great a load on the
servers, especially if remailers become more widely used.  But it's hard to
say how bad a problem this is.
 
I feel that the main purpose of an anonymous address is to protect the
anonymity of the person being addressed, not people who send to him.  Just
because a person chooses to be anonymous is no reason to expect that
everyone who wants to talk to him also wants to be anonymous.  I think it
would be better to only provide anonymity when asked.  Systems that do too
much for people sometimes get in the way.
 
Hal Finney
74076.1041@compuserve.com
 



Distribution:
  Cypherpunks >INTERNET:cypherpunks@toad.com

Hal Finney writes:

> I feel that the main purpose of an anonymous address is to protect the
> anonymity of the person being addressed, not people who send to him.  Just
> because a person chooses to be anonymous is no reason to expect that
> everyone who wants to talk to him also wants to be anonymous.  I think it
> would be better to only provide anonymity when asked.  Systems that do too
> much for people sometimes get in the way.

Well, yeeeeesss... but....

It all depends on the intended target audience. If our users are pretty
sophisticated netfreaks, I agree that the philosophy of the system ought
to be "only do what the user asks for". But if the users are
non-computer-literate people, seeking a source of support and
understanding in this vast mess of e-mail and netnews, I feel they need
and deserve all the hand-holding and safety switches the software can
provide.

So it seems there is room and need for *different* remailers.

	Julf

I suspect that a ID creating forwarder should _never_ send the ID to
the user, as someone might be looking (both the current plain text
replys and traffic analsys are problems).  If the user wishes to know
their ID then they can send a message to themselvs, and read the ID
off of that, right?
		||ugh Daniel