I've put up a link to our recently accepted paper in the 10th International Conference on Distributed Applications and Interoperable Systems (DAIS 2010) here: Validating Evolutionary Algorithms on Volunteer Computing Grids. It describes our previous validation work (the newer validator is a bit different and more strict). We still do the same validation (usually optimistic) but now we're also validating other workunits selected at random to prevent awarding credit to bad hosts.

---