[entropyfails]

The Peak Oil debate that we have on here typically revolves around two theories. Colloquially, we call them “doomer theory” and “cornucopian theory”. However, given the diametric opposition of their proponents and their predicted results, we probably deal with only one theory and its negation. (Though feel free to add a theory and its computed probabilities to the thread!) Set in that framework, we should be able to apply some Bayesian analysis to these theories and perhaps get some real numbers behind the somewhat vague generalizations that we see posted far, far too often around here. I would like to use this thread as a starting point to put some real numbers behind these theories and compute the likelihood of each theory.

Starting Point

To begin a Bayesian analysis, we need to identify theories, their prior probabilities, and the probabilities of experimental results predicted by these theories. If you don’t know much about Bayes’ Theory, please do some research and come back when you have at least a general understanding of it. http://yudkowsky.net/bayes/bayes.html serves as a good introduction.

From that starting point, we can take the evidence gathered in the real world and posted daily at this site and compute the probability that a theory has validity. Bayes’ Theory works regardless of the order of the experiments given. So each new data point with statistical relevance either further establishes or detracts from the likelihood of the theory in question. Doing that across all available evidence should help us confirm or deny the theory in question.

Olduvai Theory

This seems to be the reigning scientific theory of how the “doomer” scenario will unfold and what causes it has. It maintains that energy use per capita serves as the fundamental unit of measure of industrial civilization and that industrial civilization can no longer maintain itself once that value falls below a critical threshold. It extrapolates the peak in energy use per capita in the 1970’s (though peak seems a bit strong, it pretty much has held itself flat for 30 years, with slight perturbations). It predicts specific events, such as large blackouts and other forms of industrial breakdown as the energy use per capita statistic drops. It also maintains that once the fall of energy use per capita begins in earnest, it will never return to the previous 1970 levels.

The prior probability of this theory should start out extremely low. Human civilization and industrial civilization has never had an unrecoverable collapse in its history. I think one thing that champions of Olduvai Theory need to understand is that they are presenting a theory with little historic probability, so any failures in the theory’s predictions will have catastrophic consequences for the theories likelihood. Though in nature, we do find many examples of this sort of energy use catastrophe, we probably shouldn’t set the initial probability of the theory any more than 1/10 of one percent.

Prior Probability of Olduvai Theory = 0.1%

Civilization Theory

This serves as the default premise of civilization. Namely that we will be able to overcome any obstacles to the long term growth of civilization and that baring extra-planetary intervention (be it asteroids or aliens), human civilization will continue and find a way to provide a better standard of living to all. This theory has been in action for the past 10,000 years and has proven itself a highly reliable indicator of human behavior, regardless of your feelings on civilization. It should have a very high level of confidence. Given that it seems like the diametric opposition to Olduvai Theory, setting its prior probability to the negation of Olduvai’s Theory seems reasonable.

Prior Probability of Civilization Theory = 99.9%

Experimental Predictions

Olduvai Theory presents several predictions at a 95% or greater

1) The average EUP will decline by .67% from the years 2000 to 2011
2) The average EUP will decline by 5.44% from the years 2012 to 2030
3) The EUP measured in boe/c/y will be 3 or less by the year 2030.
4) EUP will never rise again.
5) Blackouts will become more and more common until the 2011 range when worldwide blackouts will occur.

Given the language of the document, Dr Duncan feels 95% confidant in each of these predictions. Hence the probability of each of the first 4 predictions is 95%. Since the 2011 blackouts are the foundation of this theory, I will set the bar higher at 99% for this prediction.

Civilization Theory presents a much more complicated picture of energy use, splitting energy use into oil, natural gas, coal, and nuclear. The interactions between these energy systems has a regulation system known as the market. However, it does have things to say about the predictions above, even given the lack of rigor in its proponents claims. (Note that this criticism seems valid for some of the non-Olduvai theories of collapse, a subject I would like to expound upon at a later date.)

1) Industrial Civilization will never have catastrophic EUP incident which prevents the EUP from rising ever again.
2) Industrial Civilization will never have an EUP of less than 3 again.
3) EUP can decline in intra-year periods, but technology development spurred by this will allow it to rise after the infrastructure is in place.
4) The disconnected nature of energy networks among countries will ensure that no worldwide blackout ever occurs absent direct intervention via EMP or total war.

Given the complexity of the theories revolving around Civilization’s growth and progress, it can be somewhat difficult to determine probabilities for the data. I’ll give an outline of what probabilities I have seen for each of these predictions, but feel free to provide more information on any of them.

For prediction one, Civilization Theory predicts probably around a 99.5% chance of it being true. This seems like one of the fundamentals of civilization theory, namely that it cannot be stopped.

For prediction two, again, Industrial Civilization theory predicts somewhere around a 99% chance of an EUP above 3.

For prediction 3, for any given intra-year decline, this theory seems to predict a 80% chance of EUP growth. Any declines in EUP would be rated around 20% from the 100 years of industrial civilization. The theory states that external temporarily factors such as war and oil embargos caused the EUP decline, of which we’ve had around 20 to 25 of in the past 100 years.

For the fourth prediction, we find again that this theory presents a 99% or greater confidence in the “worldwide blackout” from occurring absent EMP/Nuclear war. Even in these cases, it asserts that the grid will be restored within a few year timeframe.

Blackouts in general are allowed in Civilization theory, though prolonged, frequent ones seems to go against the general idea of ever increasing standards of living. In the past 80 years or so, we have only seen a few years where large blackouts have occurred. I would say that it puts a 70% probability of any particular year being wide scale-blackout free. (I would like more numbers on this, however.) I would say that civilization theory predicts a 0.5% chance of a worldwide blackout in general absent EMP.

This analysis shows clearly that both theories seem diametrically opposed and one or the other will be falsified by the year 2030. To show this let us run Bayes’ Theorem for the 2030 predictions.

Bayes’ Theorem Run for the 2030 Era Predictions

If we have worldwide blackouts, and a EUP of < 3 in the year 2030, and no technological improvements do anything to change this (AKA shit hit the fan) we find that the revised probabilities of the theory are as follows. (Here likelihood takes its general meaning, not the statistical one)

Likelihood of Olduvai Theory = 98.9%
Likelihood of Civilization Theory = 1.1%

Now if we see no worldwide blackouts, EUP > 3 in 2030, and technology fixes for the above problems we find (rounding at the fourth decimal place)

Likelihood of Olduvai Theory = 0%
Likelihood of Civilization Theory = 100%

So, all the arguments about Olduvai theory being falsifiable have no merit. By 2030, we’ll know. Of course, that provides little comfort to us right now. So this becomes the real question, how will we know which theory predicts our future better.

In 5 years, of course, we will have the largest test of Olduvai theory, the worldwide blackout. If this doesn’t come to pass, we know that the theory has no merit. It predicts it at a 99% probability while civilization basically denies it could happen. So the real litmus test comes at this point. But of course, we don’t live 5 years in the future and we, for some strange reason, believe that we will be alive in 5 years. Hence we would like to know which theory to plan for.

From the numbers posted above, the best test will involve large-scale blackouts. Both theories do predict some of them, but Olduvai theory predicts them much more strongly than the Civilization Theory.

Bayes’ Theorem Run Based on Repeated Large-scale Blackouts

Since the year 2000, we have seen more and more large-scale blackouts. Some of these have hit the US, most seem confined to China and other resource poor countries. If in 2006 we see large-scale blackouts again, we should revise the Olduvai prior to 0.5% and reduce the Civilization Theory to 99.5%. If it happens in 2007, we should have Olduvai at 2.3% and Civilization at 97.7%. In 2008, if we see it again, we should set Olduvai at 10.1% and Civilization Theory at 89.9%. For 2009, seeing more blackouts should have us set Olduvai at 34.8% and Civilization Theory at 65.2%. For 2010, more blackouts means that Olduvai theory should get set at 71.7% and Civilization Theory at 28.3%. By 2011, we’ll know for certain.

Note that since Olduvai theory predicts blackouts so strongly, that any absence of them will very strongly disconfirm the theory. If we see blackouts from now until 2010 but no blackouts in 2010, we can be 96.8% certain that Civilization Theory trumps Olduvai theory.

Anyway, if you actually got to the bottom of this post that means you need to help me flesh out the various theories and put numbers behind the predictions. I have a feeling that if we use this approach we can avoid some of the flame war mentality that we often find here. And we will be developing useful information for people, instead of hyperbole.

Let me know what you think.

[Cynus]

Very nice post. I would just like to add the caveat that not all doomers are "Olduvai doomers." There's also what we might call "pre-industrial doomers," who don't think human civilization will end, but will be reduced to a lower energy use and standard of living. There's also what we might call "techo-ecotopian doomers," who think that society will reorganize itself and technology will not entirely collapse, and that renewables will allow a high standard of living with just lower mobility than we currently enjoy. But whether or not these are doomers-light or cornucopians-light is a matter of debate, albeit simply a semantic one. You say "Human civilization and industrial civilization has never had an unrecoverable collapse in its history." Industrial civilization may not have collapsed (yet), but just about every other human civilization has. I would say that the Mayans, Mesopotimians, Babalonians, Egyptians, Romans, and on and on, were unrecoverable collapses. In fact, I would say that simple induction over the history of past civilizations would warrant the belief that the chance of an eventual collapse are about 100%. You also write "Human civilization will continue and find a way to provide a better standard of living to all." But I need to remind you that centuries of darkness pass between the fall of one civilization and the rise of another.

[Ludi]

I don't know what you mean by "human civilization." Until globalization, there was no such thing as a worldwide "civilization." Civilizations have collapsed and not recovered.

[entropyfails]

$this->bbcode_second_pass_quote('Cynus', 'V')ery nice post.
There's also what we might call "techo-ecotopian doomers," who think that society will reorganize itself and technology will not entirely collapse, and that renewables will allow a high standard of living with just lower mobility than we currently enjoy. But whether or not these are doomers-light or cornucopians-light is a matter of debate, albeit simply a semantic one.
...
Industrial civilization may not have collapsed (yet), but just about every other human civilization has.

[entropyfails]

$this->bbcode_second_pass_quote('Ludi', 'I') don't know what you mean by "human civilization." Until globalization, there was no such thing as a worldwide "civilization." Civilizations have collapsed and not recovered.

[Ludi]

Ok, thanks entropyfails, just needed some clarification.

[aahala]

entropyfails, I haven't clue what your original post was about.

In order to use Bayes Theorem, one must know three of the following
four probablities: event "A", event "B", A given B, B given A. If we know
three, we can determine the probability of the fourth.

We don't know the probability of any of these four events. Out of thin
air, you have stated some probabilities, but the result in a calculation
can be no more accurate than the input. It's garbage in, garbage out.

[Ludi]

I agree, accurate predictions are impossible.

[MonteQuest]

$this->bbcode_second_pass_quote('', 'C')ivilization Theory

This serves as the default premise of civilization. Namely that we will be able to overcome any obstacles to the long term growth of civilization and that baring extra-planetary intervention (be it asteroids or aliens), human civilization will continue and find a way to provide a better standard of living to all. This theory has been in action for the past 10,000 years and has proven itself a highly reliable indicator of human behavior, regardless of your feelings on civilization. It should have a very high level of confidence. Given that it seems like the diametric opposition to Olduvai Theory, setting its prior probability to the negation of Olduvai’s Theory seems reasonable.

Prior Probability of Civilization Theory = 99.9%

[bobcousins]

$this->bbcode_second_pass_quote('MonteQuest', 'T')his serves as the false premise of civilization. It assumes exponential infinite growth in a finite world. Works just fine until you hit the finite limits.

Future probability of Civilization Theory= 0.00%

[Ludi]

$this->bbcode_second_pass_quote('MonteQuest', '')$this->bbcode_second_pass_quote('', 'C')ivilization Theory

This serves as the default premise of civilization. Namely that we will be able to overcome any obstacles to the long term growth of civilization and that baring extra-planetary intervention (be it asteroids or aliens), human civilization will continue and find a way to provide a better standard of living to all. This theory has been in action for the past 10,000 years and has proven itself a highly reliable indicator of human behavior, regardless of your feelings on civilization. It should have a very high level of confidence. Given that it seems like the diametric opposition to Olduvai Theory, setting its prior probability to the negation of Olduvai’s Theory seems reasonable.

Prior Probability of Civilization Theory = 99.9%

[MonteQuest]

$this->bbcode_second_pass_quote('bobcousins', '')$this->bbcode_second_pass_quote('MonteQuest', 'T')his serves as the false premise of civilization. It assumes exponential infinite growth in a finite world. Works just fine until you hit the finite limits.

Future probability of Civilization Theory= 0.00%

[entropyfails]

$this->bbcode_second_pass_quote('aahala', '
')In order to use Bayes Theorem, one must know three of the following
four probablities: event "A", event "B", A given B, B given A. If we know
three, we can determine the probability of the fourth.

We don't know the probability of any of these four events. Out of thin
air, you have stated some probabilities, but the result in a calculation
can be no more accurate than the input. It's garbage in, garbage out.

[entropyfails]

$this->bbcode_second_pass_quote('MonteQuest', '
')This serves as the false premise of civilization. It assumes exponential infinite growth in a finite world. Works just fine until you hit the finite limits.

Future probability of Civilization Theory= 0.00%

[Ludi]

Sorry, I'm not certain how this theorem is relevant.

What of situations which don't involve numbers?

[entropyfails]

$this->bbcode_second_pass_quote('Ludi', 'I') agree, accurate predictions are impossible.

[entropyfails]

$this->bbcode_second_pass_quote('Ludi', 'S')orry, I'm not certain how this theorem is relevant.

What of situations which don't involve numbers?

[Ludi]

No doubt I'm completely misunderstanding you.

How do you devise a probability for an event for which there is insufficient information? Wouldn't you just be pulling that probability out of your ass?

[entropyfails]

$this->bbcode_second_pass_quote('Ludi', 'N')o doubt I'm completely misunderstanding you.

How do you devise a probability for an event for which there is insufficient information? Wouldn't you just be pulling that probability out of your ass?

[Ludi]

$this->bbcode_second_pass_quote('entropyfails', '')$this->bbcode_second_pass_quote('Ludi', 'S')orry, I'm not certain how this theorem is relevant.

What of situations which don't involve numbers?