> The main issue with this is that services may be reluctant to give out this information if they consider it useful to competitors. Well, if thats the issue i

> The main issue with this is that services may be reluctant to give out this information if they consider it useful to competitors. Well, if thats the issue it's possible to use more complex cryptographic techniques to make the entire process completely zero-knowledge. The idea is that you basically take the protocol I described but then execute it in an environment for zero-knowledge proof of general computation (e.g. http://www.scipr-lab.org/ http://www.scipr-lab.org/). You'd make a number of different performance tradeoffs to optimize for that environment, but thats basically the idea. There would need to be some development needed to turn that into a production system, but if the improved privacy is the _only_ holdup, it can certainly be fixed.