Didn't the USGS use a similar approach for new discoveries. I had no argument with the critiques of their reserves growth estimates but when it came to their estimates for new discoveries I couldn't see any problem with their methodology (using Monte Carlo simulations based on probability distributions). The only problem was the degree to which the resulting discover projections were at odds with the historic decline in disoveries (eg to meet their estimates would require a complete turnaround in this trend). This suggests that their probability curves may have been a tad optimistic!

Maybe it is this "unbiassed estimate" of the probabilities we are lacking in both cases.

The thing is, those reserve errors are not going to be independent from field to field; errors in estimating the size of a field are much more likely to be a result of methodology problems, misestimating current/future technology, etc, than just statistical error.  

So adding all of the individual field estimates together will not necessary reduce error dramatically, even if the field-by-field estimates are unbiased.

Even IF the probable reserve calculations have a higher level of accuracy than current conventional wisdom, one cannot escape the fact that a sizeable number of oil fields are on the brink of decline and even terminal decline. Once peak of an individual oil field has come, extraction during the second phase becomes increasingly expensive.

Combine this factor with ever increasing world demand and the result is predictable.  Arguing whether accurate estimates of probable reserves peg the brunt of the shock in 2 years or in 10 years is a matter of pedantry - as much of a waste of time as arguing over the ideal placement of furniture while the house is burning down.