Heheh, Stuart, you came perilously close to calling me an optimist there. I'll not demur, perhaps it was today's weather - a still and pleasant day in SE England with temperatures almost as high as 10 C (50 F) even a few hours sunshine. Can I presume you are tempted to concur with my other points?

Another couple of thoughts occured to me looking at your percentage of max production graph.

1. Both OPEC and non-OPEC production look a fairly close fit to a linear function recently running up to the Katrina blip. Non-OPEC from the beginning of 2004, OPEC from mid 2004, that might give enough datapoints to give a feel for slope and error bars.
   
2. There seems to be a fairly clear negative correlation between OPEC and non-OPEC production whenever non-OPEC production is abnormally high or low, sometimes with a 1 month delay on OPEC production. That smells of there still being an OPEC buffer, albeit small, though a multitude of other explanations are possible.

"No time now for contrition:
the time for that's long past.
The walls are thin as tissue and
if I talk I'll crack the glass."

The big question is: Are the errors consistent and /or counter balancing.

If country A over reports every month and country B under reports every month, the errors may cancel out. However, the extreme consitency of the statistics from places like Venuzuala has all the marks of the EIA or IEA saying, "the figures from Venezuela are obvious nonsense" (or even "we haven't anything from them this month") "so let's use last month".

So I'd set the error bars at 2mbd or so, certainly not less. We probably will never know the exact peak, only that we will be able to say, at some time, 'we have clearly passed the peak".