1993-01-09 - Chaining addresses…

Header Data

From: Hal <74076.1041@CompuServe.COM>
Message Hash: 14a709297f7a17a3296e3afd42e20e5d280a34615b43113d3b4df5866bebf9fe
Message ID: <93010907390474076.1041_DHJ43-2@CompuServe.COM>
Reply To: _N/A

UTC Datetime: 1993-01-09 07:47:53 UTC
Raw Date: Fri, 8 Jan 93 23:47:53 PST

Raw message

From: Hal <74076.1041@CompuServe.COM>
Date: Fri, 8 Jan 93 23:47:53 PST
Subject: Chaining addresses...
Message-ID: <930109073904_74076.1041_DHJ43-2@CompuServe.COM>
MIME-Version: 1.0
Content-Type: text/plain


From: mjr@netcom.com (Matthew Rapaport)

> >I think the lesson is that this process of automatic alias assignment
> >may not be the best way to handle things... look at all the problems
> >Karl ran into.
> If I understand Karl right, he got this cascade of aliases because he
> tried to talk to HIMSELF through different accounts/aliases at alternate
> ends of the chain. Since no one would want to do that (other than to
> test things) normally, this wouldn't be a problem.

My understanding was that everyone who tried to talk to him would get
two aliases assigned automatically.  Karl made the problem worse by
talking to himself from two different addresses, but you're still
talking about a lot of aliases.

> >I still lean towards the idea of a "constructed" anonymous address,
> >where I decide ahead of time which remailers I'll use, and in what
> >order.
> But I already *do* control the order of use for MY mail, that means
> stuff I send out and stuff people send to me in DIRECT reply to my
> stuff. There is nothing to stop someone from sending to my id on pax say
> through a first remailer of their own choice, provided they originate
> the mail (i.e. a REPLY is not equivalent to ORIGINAL mail in this case).

OK, so you can set up an anonymous address which, say, goes through pax
and then penet and then to you.  If someone replies to that address,
they will be anonymous to you, by default; their anonymous address will
go through penet and then pax.

But if they didn't want that anonymous address, they could use one of
their own (say, rebma to soda to themselves) first, then go to your
address.  Now when you reply to them, I guess your message will go
through penet, then pax, then rebma, then soda, then to them.

My feeling was it would be better if they could put a Reply-To: into
the message that just meant to go to rebma then to soda to themselves, and
get that Reply-To: to go through the pax-to-penet chain to you.  Also,
they would not get anonymous ID's assigned by penet and pax, ideally.
Instead, you would reply to them using this Reply-To address and go through
just rebma and soda to get to them.  This will be simpler and faster
than having all messages go through the union of both communicant's
anonymous address chains.

> As for picking my own alias, this sounds appealing but is actually much
> weaker then a randomly assigned one. Besides that, it could be an
> administrative nightmare for the sysadmins on the aliasing systems.

I wasn't really talking about picking my own alias.  It is more a
matter of having a straightforward way to construct an anonymous
address that goes through the specific chain of systems that I choose.

Hal Finney

Version: 2.1