Should The Last Few Years Have Updated Your Idea of Peak Oil?

Nate Hagens draws my attention to this recent interview with Dennis Meadows, lead author of the famous Limits to Growth series of books.  Meadows is very pessimistic.  I was particularly interested in his views on oil production and peak oil, in which he states positively that peak oil is in the past (which is very arguable at best, given that oil production is still making new monthly highs), and also that

Oil production will be reduced approximately by half in the next 20 years, even with the exploitation of oil sands or shale oil.

For oil production to halve over the next twenty years, it would have to decline on average by 3.5%/yr throughout that time (possibly some years by more, some years by less).  Above I have posted the average annual change in oil production 1965-2012 (with data from BP except for 2012 from EIA).  I have also added a linear trend line out to 2040.  Obviously, this is a rather rough time series and the linear fit is not particularly strong and the extrapolation not particularly stable.  But it's not clear that anything else will work much better - global oil production is a very complex process that we understand poorly.  In that situation, we are probably best sticking to very simple models and acknowledging their severe limitations.  At any rate, the straight line implies that peak oil (in the sense of "average growth is zero") was in about 2009.  The straight line also implies that we would not reach average growth being -3.5% until almost 2040.

So Professor Meadows is asserting unequivocally that this line is going to make an abrupt turn downwards in the next decade or two.  Possible, but it's a strong claim that requires strong evidence, and the interview certainly doesn't lay out the basis for his views, or his certainty in them.

To give some idea of the issues with ignoring "oil sands or shale oil", let's stop the tape in 2007 before the US tight oil revolution came on stream (and before the great recession also, to avoid being too biassed by that).  That would give this graph:

You can see that on that basis, peak oil was projected to be in 2004.  So five years of additional data have moved the peak later by five years.  That should give anyone with an open mind very strong pause indeed about asserting that they have a very clear idea of when peak oil is any more.  And the date at which oil production would be -3.5% has gone from about 2030 to the late 2030s.  Five years' data have moved that estimate more than five years.

I think Professor Meadows is expressing an inappropriate level of certainty in an inappropriately negative scenario, given the available data.  I think all that we can say with any confidence at this point is that it seems in the last decade to have become much more difficult to raise oil production by large percentages than it used to be, even in the presence of sustained high oil prices.  Presumably this reflects a move closer to peak oil - I think it's reasonable to suggest that we are on a bumpy plateau around the oil peak.  But not more than that.