[raizcapoeira]

For all the talk about peak oil allegedly being dead, there's very little talk about the fact that, while the sensationalistic collapsenik predictions made by idiots like Michael Ruppert & perpetuated by people like Matt Savinar are bunk, the more level-headed peakists' predictions have been uncannily accurate. As pointed out on TOD, if we were to use mean absolute scaled error as a metric for accuracy, Jean Laherrere, Rembrandt Koppelaar, & the ASPO have been incredibly accurate compared to the likes of the utterly horrendous CERA forecasts & the IEA's mediocre forecasts. Not to mention, Deffeyes had the last laugh when it came to conventional production.

I swear, the only reason that the IEA can maintain its legitimacy is because it churns out forecasts faster than anyone can point out how mediocre they invariably are in the long-term, so, as a result, collective confidence in their permissibility is retained, regardless of how many times they and the EIA all the sudden downgrade reserves estimates by upwards of 40%, add caveats to their completely and utterly preposterous estimates regarding tight oil in Argentina & Europe, etc, etc.

Yes, I know. Economics 101 works, and I myself have never been a fan of the whole economics-bashing characteristic of the peak oil crowd (Hubbert & the technocrats believed that nuclear would replace oil 2 times over, obviously we all know how that turned out), but at some point, the IEA's predictions become worthless and they lose validity through time.

[ralfy]

From what I gathered, what the IEA gives are forecasts based on various conditions met, e.g., strong government policy needed to maximize oil and gas production, regulate businesses so that they use more renewable energy, and coordination between economies. FWIW, some of its people believe that these conditions won't be fulfilled.

I think similar applies to doomer forecasts, e.g., a war that escalates to worse, leading to economic collapse, etc.

For the middle ground, it's what has been taking place during the last decade.

Thus, we have an optimistic, worst-case, and moderate scenario.

[Arthur75]

The IEA "tried" to be more realistic, in 1998 in particular, but the political pressure are clearly there :
http://petrole.blog.lemonde.fr/how-the- ... ts-mission

As to Ruppert, what are you refering to exactly ? I don't remember any precisely dated scenario or forcasts from him.

[Pops]

Welcome RC. I agree but all predictions on all sides of the "argument" are either political or profit motivated.

Whether it's someone selling a book or OPECs paper barrels or tight oil land speculators or projections of the Tupi field, everyone "has a dog in the fight" as they say up here.

Can't really blame the IEA, though, for 30 years, about the length of an entire forecaster's career, the best formula to predict oil production has been:

Code: Select all

population*4bbls

Don't mess with success they say...

.

[dissident]

Welcome RC. I agree but all predictions on all sides of the "argument" are either political or profit motivated.

Whether it's someone selling a book or OPECs paper barrels or tight oil land speculators or projections of the Tupi field, everyone "has a dog in the fight" as they say up here.

Can't really blame the IEA, though, for 30 years, about the length of an entire forecaster's career, the best formula to predict oil production has been:

Code: Select all

population*4bbls

Don't mess with success they say...

.

That graph stops around 2005, just when things started going south in terms of global oil production. If the IEA is using 2005 and earlier correlations, then not only is their whole statistical black box modeling bad, it is also totally wrong. They have to update their correlation matrix on a continuous basis if they are not going to use a physically based model

[rollin]

The IEA has been generally feeding the policy makers what they want to hear, a great way to continue in the job. Oh, occasionally they feed out some warning or other but that seems to disappear from view and get shuffled off to the background. Oil production looks fairly rosy when viewed through corporate colored glasses.

Here is a mystery that even Sherlock Holmes could not solve:
"The Case of the Ever Expanding Oil Reserves and Oil Production"

Or maybe it is a case of what we don't see and what outrageous conclusion is left.

[Pops]

I just plucked that one out of the aether. I did this last year - 2011 data from BP I think... I can't remember the population source



The reason we aren't in worse shape is we've increased global per capita primary energy by "transitioning" to that breakthrough sustainable alternative source - coal. Per capita use up a third in the last few years. But oil is still about the same in the overall scheme.

[MD]

...the more level-headed peakists' predictions have been uncannily accurate.

yep.

[raizcapoeira]

The IEA has been generally feeding the policy makers what they want to hear, a great way to continue in the job. Oh, occasionally they feed out some warning or other but that seems to disappear from view and get shuffled off to the background. Oil production looks fairly rosy when viewed through corporate colored glasses.

Here is a mystery that even Sherlock Holmes could not solve:
"The Case of the Ever Expanding Oil Reserves and Oil Production"

Or maybe it is a case of what we don't see and what outrageous conclusion is left.

I don't know about the IEA and EIA simply telling the policy makers what they want to hear. I don't know of any incentive for them to do that. The IEA's population-based models have generally worked pretty well up until now.

[raizcapoeira]

I just plucked that one out of the aether. I did this last year - 2011 data from BP I think... I can't remember the population source



The reason we aren't in worse shape is we've increased global per capita primary energy by "transitioning" to that breakthrough sustainable alternative source - coal. Per capita use up a third in the last few years. But oil is still about the same in the overall scheme.

Definitely. I think this sheds light upon the fact that renewable energy is subject to just as many politically distortionary effects, because all I've been hearing from the renewable energy crowd is that coal is dead, despite its obvious uptick in relation to nuclear, renewables, oil, etc.

Coal won't die for a very long time, and any illusions about its being dead, as far as I've seen, have been the result of relentless propagandizing against coal by the AGW media blitz & the US' current administration's war against coal.

[raizcapoeira]

Welcome RC. I agree but all predictions on all sides of the "argument" are either political or profit motivated.

Whether it's someone selling a book or OPECs paper barrels or tight oil land speculators or projections of the Tupi field, everyone "has a dog in the fight" as they say up here.

Can't really blame the IEA, though, for 30 years, about the length of an entire forecaster's career, the best formula to predict oil production has been:

Code: Select all

population*4bbls

Don't mess with success they say...

.

That graph stops around 2005, just when things started going south in terms of global oil production. If the IEA is using 2005 and earlier correlations, then not only is their whole statistical black box modeling bad, it is also totally wrong. They have to update their correlation matrix on a continuous basis if they are not going to use a physically based model

What are you saying? That an algorithmic modeling culture should be used? That's an interesting idea.

Statistics for this purpose, I think in a lot of people's conceptions, always starts with data as being generated by a "black box" in which a vector of input variables x go in, and then out come the response variables, or y. I agree - IEA's data modeling culture (assuming a stochastic data model) is flawed. Drawing a data model through generating independent draws through

response variables = f(predictor variables, random noise, parameters)

and then filling in the black box through linear regression, logistic regression, or the Cox model, is probably too reductionistic for the purposes of forecasting such a complex issue.

[raizcapoeira]

...the more level-headed peakists' predictions have been uncannily accurate.

yep.

About the interstate highways thing, they could be made a lot better. Virtually 100% of roadbed damage is caused by heavy trucks. After repeated liberalization of maximum weight restrictions, far beyond the heaviest conceivable weight the interstate roadbeds were designed to support, fuel taxes fail miserably at capturing from big rig operators the cost of damage caused by higher axle loads. All the truckers have been successful at scrapping weight-distance user charges in all but a few Western states. So only about half of the highway trust fund comes from fees or fuel taxes on the trucking industry, and the rest is externalized on private autos.

All we really need to do is make truckers pay proportionally. There's no need for some ridiculous state-mandated overhaul of anything even remotely predicated upon this hip new "green tech" idea which has yet to materialize anywhere except for in fanciful conceptual art at geek-gatherings commenced by Amory Lovins.

[dissident]

Welcome RC. I agree but all predictions on all sides of the "argument" are either political or profit motivated.

Whether it's someone selling a book or OPECs paper barrels or tight oil land speculators or projections of the Tupi field, everyone "has a dog in the fight" as they say up here.

Can't really blame the IEA, though, for 30 years, about the length of an entire forecaster's career, the best formula to predict oil production has been:

Code: Select all

population*4bbls

Don't mess with success they say...

.

That graph stops around 2005, just when things started going south in terms of global oil production. If the IEA is using 2005 and earlier correlations, then not only is their whole statistical black box modeling bad, it is also totally wrong. They have to update their correlation matrix on a continuous basis if they are not going to use a physically based model

What are you saying? That an algorithmic modeling culture should be used? That's an interesting idea.

Statistics for this purpose, I think in a lot of people's conceptions, always starts with data as being generated by a "black box" in which a vector of input variables x go in, and then out come the response variables, or y. I agree - IEA's data modeling culture (assuming a stochastic data model) is flawed. Drawing a data model through generating independent draws through

response variables = f(predictor variables, random noise, parameters)

and then filling in the black box through linear regression, logistic regression, or the Cox model, is probably too reductionistic for the purposes of forecasting such a complex issue.

The statistical model has to be evolved to keep it marginally relevant for a complex time-dependent system. This is the price you pay for not modeling the actual physical processes but implicitly parameterizing them (with regression fits). The graph shows a parameter dependence that was valid up to 2005 and which is likely not valid today. Fixing your empirical model at 2005 conditions and trying to apply it today is not a valid approach.

But the IEA, as has been noted above, is tasked to feed its political masters what they want to hear. This IEA behaviour is an example why research should be arms length from the government. Ivory tower academics would not dance to the government's tune like the analysts at the IEA. Basically anyone who is not working for the IEA can see that its procedures are BS and so are its conclusions. And to think these clowns are being paid.

[raizcapoeira]

Welcome RC. I agree but all predictions on all sides of the "argument" are either political or profit motivated.

Whether it's someone selling a book or OPECs paper barrels or tight oil land speculators or projections of the Tupi field, everyone "has a dog in the fight" as they say up here.

Can't really blame the IEA, though, for 30 years, about the length of an entire forecaster's career, the best formula to predict oil production has been:

Code: Select all

population*4bbls

Don't mess with success they say...

.

That graph stops around 2005, just when things started going south in terms of global oil production. If the IEA is using 2005 and earlier correlations, then not only is their whole statistical black box modeling bad, it is also totally wrong. They have to update their correlation matrix on a continuous basis if they are not going to use a physically based model

What are you saying? That an algorithmic modeling culture should be used? That's an interesting idea.

Statistics for this purpose, I think in a lot of people's conceptions, always starts with data as being generated by a "black box" in which a vector of input variables x go in, and then out come the response variables, or y. I agree - IEA's data modeling culture (assuming a stochastic data model) is flawed. Drawing a data model through generating independent draws through

response variables = f(predictor variables, random noise, parameters)

and then filling in the black box through linear regression, logistic regression, or the Cox model, is probably too reductionistic for the purposes of forecasting such a complex issue.

The statistical model has to be evolved to keep it marginally relevant for a complex time-dependent system. This is the price you pay for not modeling the actual physical processes but implicitly parameterizing them (with regression fits). The graph shows a parameter dependence that was valid up to 2005 and which is likely not valid today. Fixing your empirical model at 2005 conditions and trying to apply it today is not a valid approach.

But the IEA, as has been noted above, is tasked to feed its political masters what they want to hear. This IEA behaviour is an example why research should be arms length from the government. Ivory tower academics would not dance to the government's tune like the analysts at the IEA. Basically anyone who is not working for the IEA can see that its procedures are BS and so are its conclusions. And to think these clowns are being paid.

Exactly, which is why an algorithmic modeling culture might be useful.

To be fair, though, Hubbert employed logistic regression. The only difference is that he does model the physical process. The economic models simply no longer have much predictive power. Curve-fitting & bottoms-up analysis are obviously the way forward.

[ROCKMAN]

r - Think one of the reasons Hubbert's model worked so well was the relative stability of oil prices between 1946 (when many of our big domestic oil fields began development) and 1973. Thanks much to the efforts of the Texas Rail Road Commission acting as a better cartel than OPEC ever did prices held in a very narrow range: $22 - $25 per bbl (inflation adjusted). If the TRRC had not forced oil companies to reduce production the market may have been flooded with cheap oil which would have slowed the exploration process. That could have moved the US peak a number of years different than it did occurs.

Again a simple question with an equally simple answer IMHO: had oil prices remained in the $20's per bbl range as they were in the late 90's would we be seeing the current aberration in the US oil production rate? Another question: had not the world slide into a recession and reduced oil down to around $10/bbl in 1986 what would the global oil production curve look like today? Not as easy an answer IMHO.

[Pops]

RC & Diss,
Honestly, my math skills are limited to about 7th grade level so I can't really argue models, my only point is the best predictor of production the last 30-something years was BAU; production meets demand; 4bbl/capita/year. Obviously 30 years of a flat line would make most any forecaster that forecast anything other than a flat line look kind of out there it seems to me.

--
ROCK, you keep saying "what would production be if prices were still $25?" as if price were some independent factor plucked out of thin air. Are you saying that we should be happy that Brent is at $114 because it means there are lots of beer and portable buildings being sold in N Dakota? LOL Aside from that little bit of tight oil in there and Texas, production of crude hasn't budged since '05 when oil was $40.

Price is a result of supply/demand, a higher price is supposed to raise production and lower the price, that is the rule. The fact that oil price has been at the highest average levels ever for going on 3 years now kind of indicates to me that something has changed. I'm pretty sure the price no longer being $25 indicates that we've reached the worldwide equivalent of the TRRC setting allowables to 100%

[dolanbaker]

All those global per capita figures look quite steady, but what about the regional variations, it would be interesting to see just how much of a variation there is between a mature western region with declining consumption (outsourcing of jobs, improved efficiency etc) and an emerging region (rapidly rising population increased production of goods for export etc).

[raizcapoeira]

RC & Diss,
Honestly, my math skills are limited to about 7th grade level so I can't really argue models, my only point is the best predictor of production the last 30-something years was BAU; production meets demand; 4bbl/capita/year. Obviously 30 years of a flat line would make most any forecaster that forecast anything other than a flat line look kind of out there it seems to me.

--
ROCK, you keep saying "what would production be if prices were still $25?" as if price were some independent factor plucked out of thin air. Are you saying that we should be happy that Brent is at $114 because it means there are lots of beer and portable buildings being sold in N Dakota? LOL Aside from that little bit of tight oil in there and Texas, production of crude hasn't budged since '05 when oil was $40.

Price is a result of supply/demand, a higher price is supposed to raise production and lower the price, that is the rule. The fact that oil price has been at the highest average levels ever for going on 3 years now kind of indicates to me that something has changed. I'm pretty sure the price no longer being $25 indicates that we've reached the worldwide equivalent of the TRRC setting allowables to 100%

Econ 101, works, for the most part, but as dissident pointed out, they're not working nearly as well anymore, because the IEA, EIA, CERA, etc. are simply projecting regressions derived from historical data onto the future, not modeling the actual physical process. But, IMO, simulation models which make assumptions about oil well geologies & economic optimal depletion fare best. The reality of the situation is that, while various curve-fitting techniques from Laharrere & Koppelaar & bottoms-up approaches from the ASPO have worked pretty well, & population-based models from the IEA have historically fared well (up until about 05, around when conventional peaked), there really are no models with any impressive degree of predictive power, we only have the best models. There are no truly accurate models.

[ralfy]

If no accurate models can be made, then the question of how organizations like the IEA can be taken seriously becomes irrelevant. Given that, the logical view will involve looking at a worst-case scenario, i.e., a slow decline with the possibility of collapse.

[Pops]

I think the biggest problem is not the models but the data.

Laherrere has been saying since way back that all data is political and realistically I don't see that changing anytime soon. Are all those barrels OPEC added P5 or P99? Have they actually not seen a decline in reserves in 30 years? Who knows?

The US under the leadership of Haliburton helped out by relaxing reserve reporting rules - with guidance from such as CHK. Who, coincidentally, immediately oversold the reserves held in their Granite Wash Trust - requiring a "Revision of Quantity Estimate" lowering reserves 26% only a year later.
(before you say anything rockdoc follow the link, lol)

Even the total amount extracted globally is obscured in the political and economic morass, over the last 8 years there is more variation between the different estimates of production than there is change in the value itself!