Cypherpunks, pay per use remailers, and the good ol' days

[lcs Mixmaster Remailer]

Declan wrote, quoting himself:
> > Yet some form of PPU remailer could exist today: A remailer would find a 
> > cookie and an encrypted-to-PPU-public-key credit card in the body of the 
> > message it receives. It would then debit a credit card for, say, $3 and 
> [...]
> > The usual objection to such a system would be that the feds would impose 
> > pressure on the banking system (or credit card companies would do it 
> > themselves) and prevent remailer ops from securing merchant accounts. That 
> > may be true, but remailers at least today aren't seen as a serious threat. 
> > They could get away with it for a while.
>
> Thinking through this a little bit more, such a system wouldn't work
> well given today's technology. It would allow an attacker to know 
> with a high degree of certainty the truename (cardname) of someone 
> and link that with an encrypted message. By unwrapping it down the
> chain with subpoenas and court orders, it would be possible to 
> get at least the last To: line if not the final text.

An alternative is to pay the first remailer for the whole chain, and then
to have that remailer pay the second remailer, the second remailer pay
the third remailer, and so on.  This way the follow-on remailers don't
know who the original sender of the message was.  The remailers can also
batch up their payments when they are sending a bunch of messages to other
remailers, perhaps even just pay the net clearing amount on a daily basis.

Some discussion of this idea as a mechanism for anonymous payments is
in the archives at
http://www.inet-one.com/cypherpunks/dir.2000.02.28-2000.03.05/msg00302.html
and follow-ups.