[GTER] The Penny Black Project

[Gustavo Molina]

de http://www.research.microsoft.com/research/sv/PennyBlack/

The introduction of the Penny Black stamp played an important role in the reform
of  the  British Postal System during the 1830's. Before this time, postage fees
were  based on weight and on distance involved. Postage had to be calculated for
each  letter,  and  was typically paid by the addressee. The introduction of the
Penny  Black  shifted  the  cost  of  postage  to  the sender and eliminated the
complexity of postage computation by requiring a uniform, low rate.

As  with the British Post of the 1830's, Internet email is becoming increasingly
expensive  for  message  recipients.  In  the current case, the culprit is spam.
Although  spam  does  not  constitute a monetary expense for most users, it does
require time and attention (and hence productivity) to deal with spam. Moreover,
measurable  costs  associated  with  spam  are  incurred by providers of network
services, and these costs are increasing daily.

The  Penny  Black  project is investigating several techniques to reduce spam by
making  the  sender  pay.  We're considering several currencies for payment: CPU
cycles, memory cycles, Turing tests (proof that a human was involved), and plain
old cash. There are multiple system organizations that can support this: senders
can  pre-compute the appropriate function, tied to a particular message; senders
can  come up with the payment in response to a challenge after they've submitted
their  message;  senders  can  acquire  a  ticket  pre-authorizing  the message.
Recipients would aggressively white-list good senders.

The  ticket scheme involve s creating a ticket service that would issue tickets,
which can then be submitted with an email message. The recipient would then call
the ticket service to validate and cancel the ticket. There are some interesting
ramifications  to  the  ticket  server  idea. For example, 1000 pre-paid tickets
might  be  bundled with each new PC. Look here for some initial thoughts on this
scheme.

We have formal analyses of the CPU-based scheme.We have a plausible memory-based
function.  We  know  how  to  implement  Turing  tests (though there's no formal
analysis  of  how  good  they  are). We know how to build an extremely efficient
ticket server.

We're  working  on  prototy  ping some of this, fleshing out the design, arguing
about  the  merits of the various challenge schemes. It's also quite likely that
this  form  of  lightweight  cash-free  payment  scheme could be useful in other
arenas.


--- 
Gustavo Molina          mailto:gustavo at molina.com.br