Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!decvax!bellcore!ulysses!mhuxr!mhuxn!ihnp4!houxm!whuxl!whuxlm!akgua!gatech!seismo!umcp-cs!cvl!umd5!zben
From: zben@umd5.UUCP
Newsgroups: net.news.adm
Subject: Re: Reading someone else's e-mail
Message-ID: <923@umd5.UUCP>
Date: Sat, 19-Apr-86 15:22:10 EST
Article-I.D.: umd5.923
Posted: Sat Apr 19 15:22:10 1986
Date-Received: Tue, 22-Apr-86 00:52:47 EST
References: <228@cord.UUCP> <2022@cbosgd.UUCP>
Reply-To: zben@umd5.UUCP (Ben Cranston)
Distribution: na
Organization: U of Md, CSC, College Park, Md
Lines: 57
Summary: Things an SA can do

In article <2022@cbosgd.UUCP> mark@cbosgd.UUCP (Mark Horton) writes:

>However, there are times when I have to read it.  Sometimes mail
>gets stuck here, and I have to read it to figure out who it's
>from and who it was supposed to be to, in order to try to deliver
>it.  (The postal service does this too, at the dead letter office.)

Clearly, and I don't think anyone can object to this.  If it had been
correctly addressed in the first place, it would have stayed secret.

>More often, a piece of mail is sent through cbosgd with an invalid
>To address AND an invalid From address.  (Happens a few times a week.)
>The To address is bounced by cbosgd, and a message is sent from
>MAILER-DAEMON@cbosgd back to the sender.  But since the sender address
>is also wrong, somebody else bounces this message, and it goes back
>to MAILER-DAEMON@cbosgd.  In order to avoid a loop, that's forwarded
>to root, which is forwarded to me.  So such mail gets dropped right
>into my personal mailbox.  I have to read it to try to deliver it or
>return it (if I can.)  Sometimes I can tell from a signature or a header
>what was intended.

In the ARPA Internet domain, advisories get sent with a null back-path,
so any error trying to deliver the advisory gets dropped on the floor.
So I have my advisory generator CC: postmaster.  If it's an obvious one
like "user not known at this site" I just delete it, but if it might
confuse I send an additional manual advisory.

>By the way, there is a logging mechanism in smail which logs each
>message passing through: the sender, destination, and length.  This
>log can be used to detect abusers of our phone bill.  I don't consider
>this logging unethical at all, I don't consider it to be "reading of
>other people's mail." ...

The analogous operation on physical mail (writing down the addresses to
which you send mail without actually opening the envelopes) is called a
"mail cover", and last I looked was a bit easier to get authorization to
do than for either a wiretap or a mail trace.

>...  Also, during debugging, sometimes I tee a copy
>of every message passing through into a short-term log file; this
>permits me to reproduce bugs that may appear when they are pointed
>out to me shortly thereafter.  I don't intentionally read this verbose
>log (which includes the entire message) but sometimes I see the message
>being complained about, and possibly some near it in the log.

One could process this log file and replace the body of the text with
one line saying (body removed here), but we both know that some bugs will
be sensitive to message length, "shape", or even actual contents.  If one
were really hard-core one could ROT13 the text.  This would preserve
enough of the shape for most debugging, while making sure that even an
inadvertant glimpse of the text would not reveal anything.  Plus, if the
bug DID turn out to be dependant on the actual contents, one could always
ROT it back...  But it hardly seems worth the work...
-- 
"We're taught to cherish what we have   |          Ben Cranston
 by what we have no longer..."          |          zben@umd2.umd.edu
                          ...{seismo!umcp-cs,ihnp4!rlgvax}!cvl!umd5!zben