Theory vs Experiment
Working out theories about oil depletion through non-traditional channels such as a blog has some interesting side effects. I basically sit here and pontificate about stuff that I don't work on as part of my day job. This apparently gets some people upset, even though they pontificate about the same things I do and presumably don't do this as their day job either (actually I shouldn't pretend to know what they do).
So I have finally got someone beyond the tipping point:
I've explained to you about 10 times what you need to do to establish that you have a better model. You never do it - you just complain that people still choose to use the best predictive model we have right now If you had the first bit of sense, you'd realize that Hubbert was a first-rate scientist, and name-calling him takes away from your credibility rather than his.Which prompted me to reply with this:
What is wrong? I have a theory. Theorists don't always have to defend their position. That's what experimentalists are around for. It's a classic position in the scientific world. Theorists don't mind getting attacked for their ideas. And that's not to say I am any first rate theorist, just that I know how the game is played.I remember clearly the rallying cry my co-researchers and I would sing out when faced with challenging circumstances, man-made or not, "Nature's fighting us. We must be on the right track!"
No use getting worked up over this.
Hubbert never used any mathematical model; as far as I can tell he did his computations graphically. And why would I name call Hubbert anything? I think I dissed "Hubbert Linearization" because it is based on that ridiculous Logistic curve. It doesn't have any basis in reality.
In reality, I play the part of either Punch or Judy in a hand puppet dramatization of the world's oil predicament. Puppets sitting around theorizing versus neocons experimenting with their own puppets in toeholds of power, I actually feel part of some absurd War on Brains.
As Milfington would say, "whatever".
posted by @whut at December 05, 2005
8 Comments:
Came here after reading your comments on The Oil Drum. I think you've done some excellent work in coming up with a model that accounts for perturbations in extraction rate. The problem seems to be mostly the lack of data, rather than anything else. Just wanted to support the continuing development of this model, as you've gotten some harsh/dismissive responses.
The thing I would like to see most, as other people (particularly one Silent E) have mentioned, is the inclusion of price effects on the perturbations. That would seem to be the only element missing from an already groundbreaking model.
The idealized hubbert curve seems inapplicable -- one would have to discount so many variables. It just doesn't make sense that world depletion would occur at such a paltry rate.
Keep up the good work.
Thanks for the support. I find it odd that I get attacked even though my model gives the same depletion rate of around 5% for both lower-48 and world and of way over 10% for UK North Sea (same as the analysts experts).
The bonus to this is that it makes perfect physical and mathematical sense, which I consider extermely important for getting more people to understand the implications of depletion.
I just went back again through the history of the conflict between you and Stuart -- it seems to have started with the article A Nice Counterexample in TOD. What Stuart seems to be looking for is a predictive model rather than a model that retrofits after the fact of an oil shock or an extraction rate increase. Since it's a chaotic system, and since you can't really predict an oil shock or a sudden rise/drop in extraction rates, he is in effect asking you to do something impossible. What you can do in your model that the idealized Hubbert curve cannot is include the factor of later discoveries, and, say, a fairly constant slope (flat/increase/decrease) in extraction rate. Stuart acknowledges that "Hubbert linearization is a decent approximate model for oil production unless production is interrupted by a societal collapse, or production is very closely following on the tails of a very noisy discovery curve."
He asks you to provide a graph with better residuals, but you haven't claimed that you've gotten one. But you have. The data set that gives you lower residuals for your model is very clearly the "nice counterexample" that Stuart points out, of the UK. Even in your unshocked model, you still have a shoulder and a fatter tail end of the curve for your model than the Hubbert curve would have predicted. Coupled with extraction rate shifts (something that can't necessarily be predicted, but can be factored in as data remains current) your model gives a damn good curve, closely resembling the actual one. This should not be ignored. An entire range of predictions can be made, depending on factors such as discovery, price, and extraction rate.
If you're not tied down working on other blog posts, I'd be interested in seeing your model applied to world production with nonconventional oil sources included. I'm not sure how that would work, you'd probably have to model a curve for each type of oil (the fallow times etc. would be different for other sources) and then do an aggregate curve based on those individual graphs.
It would seem that this model would not just apply to oil, but could be applied to natural gas and coal and other alternatives. This is some pretty important stuff you've uncovered here, methinks. Thanks for the coverage.
"He asks you to provide a graph with better residuals, but you haven't claimed that you've gotten one. But you have."
What I mean is, you need to make the claim that you already do have a graph with better residuals than a Hubbert curve would have predicted... I haven't seen you make that claim yet, at least not to Stuart. Just wanted to clarify.
Understood. I have a natural gas peak model for New Zealand I worked on a couple of weeks ago.
http://mobjectivist.blogspot.com/2005/11/new-zealand-natural-gas-peak-model.html
I do admit that I haven't yet tried a multiple regression type of analysis where I sweep through ranges of parameters to try to get the best possible fit. I am still at the stage of using gut feel and intuition on what reasonable parameters would be.
Cool stuff... The insight that your model isn't dependent on geology is a big one. This model of yours seems to be something of a "unified depletion model" -- potentially applicable to all fossil fuels and other non-renewable resources. This is important work, as depletion rates seem to be the biggest unknown factor left. Way to go.
I like Prof Goose and Stepping out, but I think Stuart is an arrogant knob who should be wary of mythologizing Hubbert as evidence in favor of his position. Bit early for that.
Hubbert isn't Jesus; nor is he as popular as the beatles.
And I never did attack Hubbert the person. Like others have pointed out, he did most of his work by hand, graphically doing integrations on a piece of paper, probably using a square gridded chart. More power to him. It was only until later, via Deffeyes I believe, that the Hubbert Linearization technique gained some prominence. What bothers me is that with all this increase in computational power, we are still doing lots of this using old-fashioned, and perhaps quaint, algorithms.
Post a Comment
<< Home