Deffeyes' book is the source I was thinking of as well. I only know it through a review in Scientific American a few years back.
As others have noted, the Peak Oil hypothesis is not about running out of oil, but about a decline in output. IIRC, he concludes that that we'll hit peak production from applying the output production curve for individual oil fields, whose crest is known as "Hubbard's Peak" to world oil production, i.e. treating the world oil supply as if it were one huge oil field.
I'm not a petroleum geologist (one of the myriad things I'm not), so I'm not equipped to evaluate either the method or the conclusion (aside: Deffeyes himself puts Hubbard's Peak for world oil production: see http://www.princeton.edu/hubbert/current-events-01-16-04.html) I don't know of any direct critiques. It occurs to me that one could only apply the extraction curve of an oil field to total petroleum reserves only if the extent of total petroleum reserves are known, but perhaps there's a drop off in succesful petroleum prospecting?
Either way, it's not hard to see how the oil industry could exaggerate or misrepresent a state of affairs to plead for regulatory perks. If the production curve for fields doesn't apply to the global stores, that should be better known. If the curve does apply, an alternative to letting big oil run even wilder is needed.
Curtiss