Megan McArdle addresses Laffer Curves and the marginal cost of public funds. I think this whole topic deserves much more attention but for now let me wax nerdy and just say that that Laffer Curves take average taxation on the X axis. This is really the only way to make sense of the typical shape. It is plausible to create utility functions such that maximum revenue occurs at 100% marginal taxation. More importantly there is really no reason to expect that revenues will go to zero at 100% marginal taxation. So to make the basic assumptions stick in general we have to be talking about average taxation.
Deadweight loss, however, is a purely marginal phenomenon. It is possible to have a 90% average tax rate and no deadweight loss if marginal taxes are zero along the relevant range.