Fully Homomorphic Encryption with relatively small key and ciphertext sizes

Another fully homomorphic paper recently on eprint: http://eprint.iacr.org/2009/616

It's also on HN: http://news.ycombinator.com/item?id=995203

We see that in practice our scheme appears to obtain a better depth of decryption circuit than theory predicts, although still not deep enough to enable fully homomorphic encryption; at least at practical key sizes.

Wow. I way misread that title at first glance.