At 11:59 PM 8/20/95, Nathan Loofbourrow wrote:
A couple of months back, someone (Adam Back, perhaps?) made mention of work on secure, reliable distributed computing on untrusted networks of untrusted machines. I seem to remember mention of sending out bogus computations in order to obscure the actual computation being performed.
I was never able to follow the reference, since all the info available was the name of a Japanese professor, with no mention of the university or of any publications I could look up.
Does this work ring a bell with anyone else?
Joan Feigenbaum, then of Stanford and now of Bell Labs, has worked for many years on what she calls "computing with encrypted instances." The canonical example is one where one wishes to send a problem, e.g., a routing problems a la the Travelling Salesman Problem, to a service for solving. But one doesn't wish to reveal the cities being considered, perhaps for business reasons. (The extension of this concern to crypto should be clear.) So, one transforms the data set, sends it to the compute services, gets back and answer, and reverses the transformation. I've always thought of her work as a kissing cousing to zero knowledge interactive proof systems, but it evolved along a different path. A check of her published papers should reveal several summaries. I know some of her stuff got published in "Crypto Proceedings." --Tim May ---------:---------:---------:---------:---------:---------:---------:---- Timothy C. May | Crypto Anarchy: encryption, digital money, tcmay@got.net (Got net?) | anonymous networks, digital pseudonyms, zero 408-728-0152 | knowledge, reputations, information markets, Corralitos, CA | black markets, collapse of governments. Higher Power: 2^756839 | Public Key: PGP and MailSafe available. "National borders are just speed bumps on the information superhighway."