But there is definitely a problem. A fundamental one. Overfishing is a classical "tragedy of commons" example. There are no easy solutions.
That's just P(t) - C(t) + K for P = production, C = consumption, and K = amount available at t = 0.
So for oil, we have small P, large C, large K. For tuna, we have small P, large C, small K. As such, projection estimates should rest on similar assumptions, and extinction models should be as realistic as you can get.
I agree that there are no easy solutions, but the problem is there; it's the elephant in the room, in all aspects of consumption.
(this was posted as more of a thought experiment than anything else).
I found a much better article on bluefin tuna troubles in Scientific American - indeed there are attempts to farm tuna:
http://www.sciam.com/article.cfm?id=bluefin-tuna-in-peril
BTW that's also one of the reasons why I think predicting populations 40 years in advance is hard. Even for a good model assumptions can change. And for systems of nonlinear differential equations even small differences in constants can lead to widely different outcomes.