[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: PGP-out-only vs. hashcash aware remailers?



On Sun, 4 Jan 1998, Bill Stewart wrote:

> If you modify your remailer to only _output_ PGP-encrypted messages,
> you get hashcash-equivalence, and cut abuse substantially.
> The cost is limiting recipients to pgp users (plus known exceptions),
> but it's tough to spam people when you need to look up their PGP key
> and encrypt to it (at least you'll only get spams for high-tech stuff),
> and it's tougher for random abusers to abuse people since most targets
> don't have PGP keys, and a mailbox full of PGP junk is less annoying
> to most people than a mailbox full of human-readable hate mail.
> In particular, it's harder to send death threats to politicians
> if they don't have published PGP keys.
> 
> Is this a feature that makes sense?

It makes some sense. It's similar to what I proposed a few weeks ago with
"casual" remailers. The smart middleman portion of coerce does something
similar: If it looks like a PGP message (has the "BEGIN PGP MESSAGE" 
line), it doesn't chain through a random remailer but delivers directly.
I'm not sure if anyone is actually using this, though (perhaps
tea/mccain). What you seem to be proposing is sending non-encrypted
messages to /dev/null. That may yet be an option if things get bad, but I
don't think they are that bad yet. It does seem to achieve, in part, the
goals of hashcash (although it generally takes longer to generate
hashcash, depending on the collision length required). 

> How would you implement it?

You are correct that there are easy ways to spoof PGP messages well enough
to fool a simple parser. One way around this would be to pipe any apparent
PGP messages (start and end easily detected) through PGP to de-armor only. 
A couple problems: PGP (2.6.x) doesn't seem to have an option to only
de-armor; a sophisticated spoofer could make the armor verify correctly
anyway by generating the correct CRC (trivial if you know what you're
doing). So it seems sensible to only consider some simple safeguards and
not worry about actually decoding the armor. 

Andy Dustman / Computational Center for Molecular Structure and Design
For a great anti-spam procmail recipe, send me mail with subject "spam".
Append "+spamsucks" to my username to ensure delivery.  KeyID=0xC72F3F1D
Encryption is too important to leave to the government. -- Bruce Schneier
http://www.athens.net/~dustman mailto:[email protected]   <}+++<