Re: [freehaven-dev] formalizing anonymity

[dmolnar <dmolnar@hcs.harvard.edu>]

On Tue, 29 Feb 2000, Roger R Dingledine wrote:

> * Foo-anonymity: it is not provable that a document was ever stored on a
>   given host
> 
> clearly we'd need a better word for 'foo'. possibilities include history,
> path, trail, fence and perfect forward, none of which seems to be the best
> one imo. (tho i kind of like the phrase 'perfect forward anonymity'...)
> 
 
How about something along the lines of "plausible deniability" ?

> anyway, free haven offers the first three. freenet offers the first one.
> rewebbers offer #1 and #3.
> pretty cool to be able to quantify this a bit, eh?
> hey david: have people written papers about this, or should i write this
> up using capital letters and real grammar? :)

I don't know of anyone else in the data haven
area with good definitions (or much of any definition?).

There *are* some formal definitions or
struggling-towards-definitions out there for other settings...check out
Anna Lysyanskaya's (and Ron Rivest, and Amit Sahai, and Stefan Wolf)
"Pseudonym Systems" for a neat example, unified treatment, and survey of
unlinkable credentials, 
the MIX researcher I just posted to the list, 
Markus Jakobsson's "A Practical MIX" from Eurocrypt '98 (a break of this
scheme is to appear in Eurocrypt 2000), 
and Birgit Pfitzmann's Crypto '95 paper on "How to Break ANOTHER 'Provably
Secure' Payment Scheme",
Stuart Stubblebine and Paul Syverson's "Group Principals and the
Formalization of Anonymity,"
and there are probably some other definitions hiding in electronic
cash papers.

Thanks, 
-David