From: jim@bilbo.suite.com (Jim Miller)
Date: Tue, 25 Jan 94 17:06:53 PST
To: cypherpunks@toad.com
Subject: Re: REMAIL: Cover traffic
Message-ID: <9401260057.AA04497@bilbo.suite.com>
MIME-Version: 1.0
Content-Type: text/plain



Arsen Ray Arachelian writes:

> ...Even better, have each remailer send a specific
> number of messages to each of the other remailers on the
> network.  These messages would be bogus messages,
> however, there would be a fixed number of them.
> 

> If a real nym message arrives, it is sent to the next mailer
> up the chain, as part of the n (n-1 now) that are bogus.  That
> way a spook couldn't tell where a message was going since
> he couldn't count the number of messages going out of the
> mailer.  

> 

> Also if a target remailer has n real messages to be sent to,
> any messages over that assigned packet size of messages
> get spooled for the next round of bogus mail.  This way each
> remailer will send exactly n messages to every other
> remailer on the net every specified period of time. 

> 


I like this idea.  It seems to use fewer CPU resources than having a  
remailer route a bogus message through a random set of other remailers and  
back to itself.

If I understand the encrypted remailer scheme correctly, the "route  
through random set" mechanism requires a remailer to enclose a bogus  
message in a set of nested digital envelopes (one for each remailer in the  
random remailer set).  The "round-robin send to peers" mechanism only  
requires the remailer to create one envelope per bogus message.

I also like the idea because it seems easier to analyse, and therefore  
easier to describe/formalize the properties of the system as a whole.


Jim_Miller@suite.com