Crude Oil Production Plateauing?
The latest data collected by Rembrandt on TOD shows indications of crude oil production plateauing.
In the comments, JonFreise referenced a potentially pertinent article to the Dispersive Discovery Model:
http://dieoff.org/page197.htm
Net Energy Analysis of the U.S. Oil and Gas Exploration Industry
"Why should yield per effort be related to effort? This makes sense for fish, for the fish can recover through reproduction and growth when not fished. Petroleum obviously cannot, at least on time scales of interest to our species. One possible explanation is that when drilling rates are low, the petroleum industry drills at locations where present information suggests that success is most likely. During years of high drilling rates, drilling is done there plus at other, less promising locations. Presumably the development of exploration theory, as well as seismic charting and interpretation, occurs at a more constant rate than drilling effort, so that when drilling effort (i.e., economic incentive) is low, it is concentrated in areas where success appears more likely. When drilling effort (and economic incentive) is high, much of that effort is directed at targets less likely to produce a large find. In a sense it is promising but untested geologic information that is depleted as wells are drilled and that accumulates in the absence of drilling."
I don't disagree with this as it basically says that a wide range of search rates get deployed over the years, and much of the high/fast effort gets expended near the margins of potential oil volume. The Dispersive Discovery model essentially takes a probabilistic range in search rates over a large potential volume of finds to come up with the more-or-less "bell-shaped" discovery curve.
posted by @whut at February 24, 2008
13 Comments:
How do you feel about the general comment that these are economic models?
The reason I ask is that while economic models are often useful, we know from common experience (financial TV) the forecasts tend to be recalculated and reissued ... daily.
I know I'm slow. I think because Hubbert was a geologist I kind of associated his model (and others) with the sciences. In fact, the inputs are economic data (units of production).
This reinforces my belief that while models may provide "indicators" they don't provide the sort of definite "predictions" so many in the peak oil community think they do.
Proved world oil reserves:
1.37 trillion barrels of oil.
Source: Oil and Gas Journal, Janurary 1, 2007.
Approximate numbers of years of consumption at current rate: 100 years, conservative estimate.
Not including future additions to proven reserves.
What is economics?
This is really the issue, and I am afraid I haven't been tainted by a particular belief system.
Turn the question back around, would you consider a basic pedator/prey model an economic model? If you assume that predators practiced economics then I suppose these then would also be economic models because some free will is involved. The fact that humans are animalistic and greedy and will not always work in their best interests means that in fact what we do as oil depletion analysts fall more into the biological predator/prey model category than economic models.
Cripes, do we think that extinction of a species like the passenger pigeon had anything to do with economics, but instead was an inexorable grind of a simple rate model destined to drive a population into the ground?
That was a twist I did not expect!
I was pretty convinced of the frailty of economic prediction, and was even going to throw this one out:
When someone says "I can predict the future," why isn't "no one can predict the future" sufficient answer.
I DO think economic predictions are frail, Black Swan, Fooled by Randomness, the whole bit ... and I had been thinking that "safe" scientific predictions, like the future positiosn of the planets were safe because they needn't model animal choices.
I had not thought of biological models, though now that I give it a moment, one of my other positions is that I support ocean reserves (no take zones) over managed fisheries, because in the latter I don't think we understand the systems as much as we think we do. I've argued in fact that fishing to the limit (of "sustainable yield" will always produce a crash.
So ultimately, no. I think simple predator/prey models are useful in teaching, and more complicated ones might help researchers reach for understanding
... but I don't think they (either) will ever become reliable predictors (in the real world).
Odograph challenges the information presented in the post and WHT responds with a non-responsive answer.
And has the pretension to say he hasn't been "tainted" by a particular belief system.
This guy has credibility?
Odograph, you make an insightful challenge, and what does WHT do?
He meanders off into a navel gazing exercise about preditor/prey economic models.
Was that a defense or explanation of why WHT presented that data and thought it was informative?
I appreciated WHT's answer, because it did make me think (early in the morning).
I'm interested (after a few years in and out of peak oil) in the relationship between risk assessment and prediction.
It is true that this bites in areas like fisheries management as well.
And so while I'm confident in my suspicion of economic prediction, I enjoyed the shift.
Greed is either constant or monotonically increasing which makes it a "nice" invariant to use for modeling.
Economics certainly has greed, but it is universally agreed that we cannot predict economics with any reasonable event horizon.
It's true that we can model things, and it's true that the models can sometimes increase our understanding ... but it's this leap from model to prediction that seems important.
BTW, I think this is a good related article at portfolio.com. It even has some outrageous Taleb quotes ;-)
Sorry, I linked you to page 3 above.
Of course Gaussian solutions and normal distributions are wrong for temporal series, and that's part of the rationale for my thinking a little differently.
I think I read that paper in a little broader context. It was a model that was accepted in a group think when it obviously, by superficial inspection, should not.
(Any model that requires a past time series violates the cardinal rule: past performance does not guarantee future returns.)
And so I think we are back to:
When someone says "I can predict the future," why isn't "no one can predict the future" sufficient answer?
Bayesian is better than nothing.
Post a Comment
<< Home