Re: (Fwd) Gov't run anon servers

At 10:52 AM 3/4/96 -0800, Hal wrote:

>However, if I were a computer-savvy law enforcement agent, and I wanted
>to track messages through one of my remailers, I would try a
>technological approach.  I would first break the key for my remailer.
>That is trivial.  The passphrase is in PLAINTEXT in the script file
>which runs the remailer!.  It has to be.  That is true of all automated
>remailers. 

Maybe I just don't know much about automated remailers, but I don't 
understand why you said that the passphrase "has to be" in plaintext in the 
script file.  I find this hard to believe.  While I am far from an expert on 
cryptographic matters, I would assume that any received attempt at a 
password could be securely hashed (128 bits?) and compared with a pre-stored 
hash value.   If it's the same, it's assumed that the password was correct.

What's wrong with this?

Hal writes:
# The passphrase is in PLAINTEXT in the script file
# which runs the remailer!.  It has to be.  That is true of all automated
# remailers. 

Jim Bell writes:
> Maybe I just don't know much about automated remailers, but I don't 
> understand why you said that the passphrase "has to be" in plaintext in the 
> script file.  I find this hard to believe.  While I am far from an expert on 
> cryptographic matters, I would assume that any received attempt at a 
> password could be securely hashed (128 bits?) and compared with a pre-stored 
> hash value.   If it's the same, it's assumed that the password was correct.
> 
> What's wrong with this?

For the less sophisticated remailer software that uses variable-size
messages and (optionally) PGP, the remailer script needs to feed the
plaintext passphrase into PGP to decrypt the remailer's private PGP key.

Mixmaster, which includes its own set of crypto routines (currently using
RSA with 3DES as I recall), allows you to compile the private key passphrase
into the executable, and wipe out the source code. This obscures the
passphrase plaintext from (very) casual observers.  

The fundamental problem AFAICS is the difficulty of getting a program to keep
a secret from an observer. If the program doesn't actually _use_ the secret
(in the way that the secret is useful, e.g. as the basis for a symmetric
key), then it seems you can attain an arbitrary level of "security through
obscurity", because you can encode the secret however you want in the code.

But if a program is capable of possessing and using the secret without
human intervention, then anyone with a copy of the program can do the same.  

Bottom line: if you can crack (say) the 8-character Unix passphrase for a
remailer account, you have full access to the remailer's secrets and all the
opportunities that presents. Good remailer account passphrases are
important.

-Lewis	"You're always disappointed, nothing seems to keep you high -- drive 
	your bargains, push your papers, win your medals, fuck your strangers;
	don't it leave you on the empty side ?"  (Joni Mitchell, 1972)