Re: (Another) alternative to

Reply to:   RE>(Another) alternative to remailer shutdowns

Where would these message parts be put back together/decrypted? Wouldn't this
require the receivers mail program to re-generate the original message?

--------------------------------------
Instead of having the last remailer in the chain store the plaintext of
an encrypted anonymous message, it might be more convenient to have the 
sender split the message into two messages and send these. The first 
message would contain random characters,  and the second would contain 
the xor of these random characters with the anonymous message. By themselves, 
each piece would, of course, be harmless random text, so remailer operators
greatest crime would be spamming. If the two pieces were sent through 
chains with different last remailers, no one operator could be held
accountable, and, of course, it would be ridiculous to suggest that
one operator could be held responsible for that fact that another
sent some random text which happened to be the xor of the random text
another had sent with a harassing message. (For instance, the
other operator could be trying to frame the first, with the
help of the receiver.) It seems to me that the only way to deal with a 
remailing scheme of this kind would be outlaw anonymous remalining in 
general.

Leonard

On 23 May 1996, Jay Haines wrote:

>         Reply to:   RE>(Another) alternative to remailer shutdowns
> 
> Where would these message parts be put back together/decrypted? Wouldn't this
> require the receivers mail program to re-generate the original message?
> 
> [my stuff]

Yes, it would require work on the part of the receiver to put the message
back  together. 

This shouldn't be too difficult, though. The receiver's software
could look through all random messages of the same length and xor them 
together to see if something non-random popped out. Alternatively, a
"standard" format for these split messages could include the 
a random message ID as the last 160 bits of one part of the message 
and the SHA hash of this ID in the last 160 bits of the other part.
(The idea is to keep the messages looking random  for legal reasons.)

Remailer operators may want to add a note to message to the effect of
"For legal reasons, this remailer only sends random looking text. For
information on the possible usefulness of random looking text visit
<joe neutral's home page>."

I would like to add an extension to my proposal for split message
anonymous remailing schemes. The current proposal counts on the good will
of the senders to split their messages into random pieces. If a remailer
operator receives non-random text I can think of at least three options
for what to do:

1) The operator can drop the message. 
2) The remailer operator can split the message him or herself and send 
   the pieces through new remailer chains.  
3) Suppose there is a computer named Moe that lives on the internet,
   is hidden behind a nym, and splits messages and enters the pieces into
   remailer chains. The remailer can then encrypt the non-random text
   along with a "please forward to" command to Moe, and  send the ciphertext
   anonymously to Moe. (Due to the encryption there is no need to split
   this message.)
   
The risk, in ascending terms of legal danger, for the operator would
be 1), 3), and 2). Nevertheless, all three alternatives should be safe 
within a secure (in the sense of Chaum, 1981) remailer network.

Leonard