Slashdot Mirror
Protecting Online Identity Through Cryptography
A new startup, Credentica, hopes to offer the ability for you to perform secure transactions using the smallest amount of personal information possible. Their goal is to both protect privacy and enhance security, which they hope will be a mutually inclusive process. "The technique employs secure multi-party computation, a branch of cryptography that can calculate meaningful answers about secret information by knowing only some non-revealing clues about that secret. The underlying theory was demonstrated in 1982 by Andrew Yao in the so-called Millionaire's Problem [...] U-Prove employs an ID token, a special kind of digital certificate that allows for minimal selective disclosure. The tokens can store all kinds of information, but users can disclose only the minimum amount of data required in any given transaction. They leave no unwanted data trails and permit both anonymity and pseudonymity."
Millionaire's Problem: Alice and Bob want to find out who has more money without disclosing the amount of their fortunes to each other, or even to a mutually trusted third party. By applying special functions to their information that disguised it, Yao proved that each could know who was richer without either revealing their true holdings.
No wonder Millionaires are so stupid... if this is what they consider a "Problem"...
For people who want background or just enjoy math, Brands's book is Rethinking Public Key Infrastructure.
This is not the first use of multi-party computation. MPC is probably the most advanced cryptographic tool theoretical crypto has produced in the last 35 years. (The strongest flavour being Universally Composable MPC). Also, though the intuitive concept of secure MPC was introduced by Yao the later results of Goldreich, Micali and Wigderson in their 1986 paper How to Play Any Mental Game is the one upon which modern MPC is based and the result which is usually cited in cryptographic literature. (My guess is the wired article author got the bit about Yao from wikipedia.) It is in this paper that the security requirements of such a protocol are first formally described using what is now called the ideal/real paradigm. Essentially a secure protocol computing some joint functionality of all players inputs should be as secure as if there where a totally honest trusted third party who would gather their input, compute the function and privately hand the outputs back to all players. (This paradigm is probably at least as important a contribution to modern crypto as the actual MPC protocol they presented in the paper.)
The problem with MPC protocols is that since they are so very general and powerful they tend to also be horribly inefficient (though polynomially bounded (i.e. in P). Never the less the constant are often horrible and could require on the order of n^2 rounds of communication. Another hurdle in their wider adoption in the field of security is that they represent a significantly more complicated concept then say encryption or a hash function and so tend to be a difficult sell to non-cryptographers.
However at least one company, Cryptomathics of Aarhus, Denmark are working on an implementation of MPC. The main client being the danish government which wants to use the product to setup an online market through which local farmers can to sell there goods. The idea being that by using an MPC protocol to do this rather then some central (government run) server no body needs to trust anyone else, not even the government; just their own implementation of the software on their computers. As long as that is correct and uncorrputed they are guarenteed all the security they could hope for.
Of course there is always the argument that you might well be better off trusting the government to host the entire show then your own computer, but on the other hand even IF the government runs some online auction server, you still need to connect to that remote system from your own computer. So a secure server is still not going to help you protect yourself from local corruptions. At least now that is the ONLY thing left to worry about.
To the asshole who tagged the article `terroristsdream': terrorism is not an excuse to erode our right to privacy. Fuck off.