Apologies in advance for the length of the proposal. There is no TL;DR. Request for comment.
A large issue with widespread surveillance is that we are unable to trust even endpoints with some data, no matter the security of the container for that data in transit between clients, as endpoints may be pressured financially or politically into cooperation or may be compromised. Many stakeholders have called for the effort to involve not only engineering but some amount of legal and policy reform. It may at first seem unfortunate that all we have influence over are protocol standards. More hopeless still is to realize that passing the right policies and laws in North America won't fix our problem as agencies and companies under the jurisdiction of the United States are not the only players in the surveillance game, nor will they be in the future. Furthermore there will likely be escape clauses around policies and laws, especially where there may be international cooperation between countries with unique mandates. I'd like to bring up the fact that there *are* some things we can do to limit the damage that untrusted endpoints can do from a protocol perspective (far above and beyond authorization/authentication). Secure Multiparty Communication is now, and has been, a feasible technology - it's merely lacked a standardized protocol. I argue that its lack of adoption so far is due to said lack of a standard. With SMC, parties can interact to solve problems like: - Personalized advertisements without the advertising company getting raw access to preferences, browsing history or target demographics. - Search or database results without the service provider obtaining unencrypted access to the query. - Determine the winner of auction without revealing what price was paid (or compute a fair price of an auction market whilst keeping financial information of participants secure, as was done with sugar beets by the Danish) - Look up sex predators in an area without revealing an address. - Transfer money from a customer to a merchant without revealing the customer's credit card number to the merchant or the merchant's business ID to the customer. That is to say with SMC there can be a finer gauged level of control over what data gets shared with what endpoints. It is a way to engage in cooperative computation without giving up permanent control of personal data. Furthermore there are many protocols to consider that have been vetted by academics and peer review, some of which are *unconditionally* secure so that there is no need to worry at the possibility of cryptographic backdoors. Nothing comes for free: there is communication and computation overhead that is induced by participating in such a protocol. Thankfully the constants and communication rates involved in modern SMC are small enough to make many applications practical. We argue that now is the time to consider creating a standard for SMC, as underlying cryptographic gadgets will only become more efficient after standardization and because it's applications have become significantly more important. Soliciting feedback. Best, Ross Snider
_______________________________________________ perpass mailing list [email protected] https://www.ietf.org/mailman/listinfo/perpass
