My sentiments exactly. When you need to make several key assumptions for a model to work, and sensitivity analysis shows that varying all the assumptions leads to radically different results, you know it's time to stop basing conclusions - let alone policy - on these models. Even making 'business as usual' scenario's is dubious with this approach.

But the IEA just goes on and on and on, repeating the same mistakes from the past, without reflection.

You mean like this? (From page 3 of your link for the IEA's 2004 outlook.)

Mexican crude oil production is projected to peak at 4.2 mb/d around 2010, and then to remain almost flat during the 2010s. It is then projected to decline sharply, to 2.8 mb/d in 2030.

Mexico, total liquids, peaked in 2004 at 3.825 mb/d, the very year of that report. Last month their all liquids production was 3.080 mb/d. That is 1.12 mb/d less than the IEA's expected plateau in 2010. It should easily reach 2.8 mb/d in 2010 instead of 2030. That would mean their predictions are only 20 years off.

Ron Patterson