The canonical reference for digital timestamping is the work of Stu Haber and Scott Stornetta, of Bellcore. Papers presented at various Crypto conferences.
More importantly, they have patented the plan. I've requested information on licensing and received no response. Who knows what they are up to.
Their work involves having the user compute a hash of the document he wishes to be stamped and sending the hash to them, where they merge this hash with other hashes (and all previous hashes, via a tree system) and then they *publish* the resultant hash in a very public and hard-to-alter forum, such as in an ad in the Sunday New York Times.
Does anyone know of any definitive prior art that reads against these patents? Hash functions are old news. Does anyone know of a published descriptions of a system that would report hash functions of large blocks of centralized data?
In their parlance, such an ad is a "widely witnessed event," and attempts to alter all or even many copies of the newspaper would be very difficult. (In a sense, this WWE is similar to the "beacon" term Eric Hughes used recently in connection with timed-release crypto.)
Haber and Stornetta plan some sort of commercial operation to do this, and, last I heard, Stornetta was moving to the Bay Area (where else?) to get it started.
This service has not yet been tested in court, so far as I know.
The MIT server is an experiment, and is probably useful for experimenting. But it is undoubtedly even less legally significant, of course.
--Tim May
-- .......................................................................... Timothy C. May | Crypto Anarchy: encryption, digital money, tcmay@netcom.com | anonymous networks, digital pseudonyms, zero 408-688-5409 | knowledge, reputations, information markets, W.A.S.T.E.: Aptos, CA | black markets, collapse of governments. Higher Power: 2^859433 | Public Key: PGP and MailSafe available. "National borders are just speed bumps on the information superhighway."