[sci.crypt] Blind signatures v. universal mail covers

chaum@mcvax.cwi.nl (David Chaum) (04/20/87)

     Around Christmas before last I was quite surprised to learn
that most paper letters in the U.S. go through one of a hundred
or so giant machines that try to automatically read the last line
of the address and translate this to a bar code, which is printed
on the envelopes before they are deposited in appropriate bins.
It occurred to me that extension of such techniques might have
serious implications, since in the U.S. there are no laws
limiting the postal service in its use of information written on
the outside of envelopes.  In fact, manual recording of such
information by the postal service on all mail sent or received by
certain individuals targeted by government agencies, an activity
called a "mail cover," apparently has been quite widespread. Of
course with computerized reading of complete addresses on a mass
basis, paper mail might also be subject to covert and possibly
retroactive traffic analysis, and even to further inferences
based on patterns of associations and timing.
     A possible solution to the problem occurred to me then.  The
idea is just a straight-forward adaptation of the blind signature
payment system already published (CACM Oct. 85).  The mailer uses
say a dot matrix to print an electronic bank note on each
envelope; when the envelope is processed by the post office, the
bank note is read and verified.  Just as in the basic untraceable
payment system, the payers (mailers) use their own computers to
obtain the signed blinded notes from the bank (post office);
these computers then check, unblind, and store the notes; and
ultimately the notes are provided as a form of payment (printed
on envelopes).  When the signed note numbers are received by the
payee/bank (post office), in addition to the checking of their
digital signatures, they are checked against a list of already
spent numbers and are rejected if they already appear on the
list.  Return addresses need not even appear on envelopes, but
the electronic note could be extended with a machine readable
indication of the zip code to which its letter should be
delivered.  If a piece of mail turns out to be undeliverable,
part of the electronic bank note printed on it is published,
which should give the sender the opportunity to resend or reclaim
it.
     In the last year or so, the U.S. Postal Service has been
trying to do away with postal meters by allowing a new way to pay
for mail. This involves delivery to the post office of a computer
generated "manifest" document indicating the number of pieces of
each weight and class along with a floppy disc containing a
complete list of the addresses of each piece of mail.  It seems
to me that such manifests: give the advantages of computerization
to big organizations at the expense of small ones and
individuals; give significant data to the postal service already
in computer readable form; and open the possibility of corruption
in accepting large shipments of mail without proper inspection or
charging. The alternative proposed above, blind signature stamps,
on the other hand, seems to solve these problems and reduce the
need for optical character recognition applied to addresses on
letters.