This post at The Oil Drum is a summary of a wide ranging critical review of various species of oil depletion models.

The author of the review, Dr. Adam Brandt, concludes that oil depletion models based on Hubbert’s logistic curve fitting can predict the decade of peak oil, but not the exact year, because in the real world some producers can and do hold back production for economic reasons. He also concludes that most models out there have not displayed good predictive ability. He calls for future oil depletion models to be based on system dynamics and to integrate both geological and economical dimensions into the model, but even then such models would only be useful for near-term forecasts because they become “brittle” over the longer term.

Dr. Brandt also emphasizes that the important question is not to try to pin down the exact date of peak oil, but to start planning for its consequences in a timely fashion, so as to have time to adapt. This is similar to the line taken by the German army think tank study on the economic and security implications of peak oil that was leaked yesterday.