As much as I admire and respect Wilson - and make no mistake, I consider him an intellectual giant, an eminently talented writer and a truly admirable human being - I think he's wrong on this one, and that so-called "multilevel selection" always reduces to gene-level selection. The latest evidence seems to confirm that:
See http://www.newscientist.com/channel/life/mg19826595.300-altruism--its-all-in-the-genes.html and http://www.sciencemag.org/cgi/content/abstract/320/5880/1213 for the original (unfortunately it appears you have to pay for the full paper).
Whatever the truth of the matter turns out to be, I love a good scientific debate and will continue to watch this with interest.
On a separate note, I was a little annoyed to read that Wilson and Hölldobler's book on ants was coming out in a revised edition, as I've just shelled out $80 for the original, but it appears that it's a new book rather than an update. So I'm happy again!