We all know about Gentry’s breakthrough this past year where he showed the first construction of a fully homomorphic encryption scheme. Gentry’s scheme is hard to understand and seems challenging to implement, and was/is viewed as a feasibility result only.
More recently, two new fully homomorphic encryption schemes have been proposed, by van Dijk, Gentry, Halevi, and Vaikuntanathan and Smart and Vercauteren. No paper is available for the first one, but the abstract explicitly claims conceptual simplicity. As for the second, a paper (with implementation results!) is available; I have not yet read the paper, but they claim that their scheme can be viewed as a generalization of Gentry’s scheme to algebraic number fields.
I was asked recently whether fully homomorphic encryption would become remotely practical within the next 10 years. While it’s still too early to say for sure, the fact that there are (at least) two relatively quick improvements to the original scheme gives hope.