[njkanjsd]

Having worked on a similar product and heard a very similar description of the 'proprietary' method, I'm guessing either security, speed or both are actually compromised.

[michwill]

There are many proprietary methods which are based on deterministic encryption + obfuscating word distribution, that's what most companies do.

We avoid doing that because of questionable security of such method. Also we tend to publish what we do (stay tuned for Hadoop paper :-)