Henry Kissinger once quipped, “University politics are vicious precisely because the stakes are so small.” While Kissinger has achieved a level of amorality not conceivable to most of us mere mortals, to my knowledge he has never been accused of being stupid. The recent debate over Peak Oil has certainly been laced with plenty of bitterness, but in this case the stakes are incredibly high. The reality, or lack thereof, of imminent Peak Oil directly affects nearly every person currently drawing breath on this planet. If Kissinger is right and the level of hostility is in any way indicative of what is at stake, surely we have sunk to the level of a “tempest in a teapot.”

In the interest of focusing this discussion and opening it as far as possible from the confines of a teapot, it might be helpful to evaluate the Peak Oil issue as a testable hypothesis. My hope is that a clearer understanding of the nature of this hypothesis and how the data must be evaluated can give us some much needed perspective that will be useful as we work through the debates about Peak Oil now appearing in both the alternative and mainstream media. We’ll finish by applying these analytical tools to the recent discussion about abiotic oil.


Framing Peak Oil as an hypothesis requires a little background. In science (and this should be a scientific discussion), hypotheses come in different flavors and are tested in different ways. The most relevant contrast for our purposes is the difference between tests based on class membership versus those based on properties of “individuals” [1]. Simply put, classes of things have distinct boundaries that are set either by natural laws or by arbitrary convention. Helium is a natural class of atoms with a particular nuclear configuration of two protons and two neutrons. Energy-efficient cars are a class set by convention, perhaps those that achieve 30 mpg. Any atom having an atomic weight of more than two or any car that doesn’t get 30 mpg are not members of the class “helium” or “energy-efficient cars,” respectively. Importantly, failure of any member of the class to meet even one of the specifications of that class is grounds for being kicked out. For example, a car that gets 25 mpg is not a member the class “energy-efficient cars,” even though it may be a Honda with a 1600cc engine and minimal emissions.

An individual is quite different. With respect to hypothesis testing, the “individuals” we are talking about can be any single thing that has a unique beginning, ending, and space that it occupies during that time. A single organism, car, helium atom, or event qualifies as an individual. For example, an individual human has a unique beginning (it is born) and end (it also dies), and as far as we know it is limited to the domain of Planet Earth (astronauts are less restricted than others). Likewise, an individual event, like the bombing of September 11, has similar restrictions in time and space. Even if a seemingly “identical” event occurred in the future it would not be the same simply because it would have happened in a different time. Whereas class boundaries are sharply defined, the boundaries for individuals have “fuzzy edges” either in time or space. The rancorous discussions over the beginning and ending of human life will always be tinged with arbitrariness, precisely because the interval of time within which an individual human exists is indistinct. This is why the question “when does a human life begin?” is impervious to boundary-related definitions.

This distinction between testing hypotheses based on class membership versus the reality of an individual impacts how we evaluate data. Hypothetical statements about classes of things are relatively easy to blow out of the water, that is, to falsify, because it takes only one instance of falsification to make the entire statement false. For example, “all planets are somewhat round” is falsifiable if we find a single square planet. The alternatives are that the square thing we just discovered is not a planet or that the class “planet” needs some revision

Testing a hypothetical statement about an individual is not as straightforward because we test the reality of individuals by establishing a balance sheet between data that support the recognition of that individual and data that refute that reality. For example, if I had failed to show up at a lab meeting last Wednesday afternoon people would likely have wondered where I was, particularly since I was supposed to be presenting research results. However, I seriously doubt anyone was going to suggest that I never existed. Why? Because there is a vast amount of other data that show that I do in fact exist, and that the simplest explanation is that I spaced out and forgot to come.

Another important concept is that the data ledger that we will establish is not a static, unchanging account. There is always room for more data and/or better approximations of the existing data. This ongoing process of data accumulation and refinement implies another important feature of testing for individuals: you can always change your mind depending on what the available data are telling you. Unfortunately, our egos are real and it is easy to forget that we ought to be in the business of searching for answers rather than defending our reputations.


The distinction between tests of classes and those of individuals is the basis for establishing the terms of our discussion of current and future oil reserves. “Peak Oil,” if it exists, certainly has attributes of an individual, in this case an individual event. It would be a unique instance in the geological history of the earth, or for that matter the known universe. Moreover, it has a beginning and ending that is (or will be) extremely fuzzy. This fuzziness would be true regardless of whether we had the means to measure it precisely (as far as I know, we do not). In fact, many depletionists suggest that we will only recognize Peak Oil in hindsight.

Recall that for individual events such as Peak Oil, we accept or reject their existence based on the balance of all the data that support or refute their existence. In the simplest of terms we could weigh the accumulated evidence by counting up the individual data points on either side of the issue and make up our minds accordingly. Unfortunately, things are usually considerably more complex. What is perhaps different about the individuality of Peak Oil as a world event is that, perhaps up until now, it has been a prediction rather than historical fact. Thus, the data must be placed into a predictive model and a critique of the hypothesis will also include a critique of the model.


A detailed analysis of oil data, monitoring, and modeling are well beyond my expertise and the scope this short essay. Nevertheless, to place the current debate into the logical context of described above, a few general comments are in order.

If we are going to be in the business of tabulating data, we must decide on the nature of the individual data points. For hypothesis testing these should be reduced to the smallest possible units that are still fully independent of one another. I’ll leave this to those with expertise in petroleum, but from the outside looking in it seems that current depletion models that consider oil production in individual fields (e.g. [2]) are the fundamental starting point and consistent with the logical structure for testing Peak Oil. These data incorporate a weighting scheme based on production in each field (see Data Quality below).

Various other “data” have been dragged into this discussion that supposedly refute Peak Oil. These include unproven reserves that supposedly need only the “market” to encourage development, vast oil-containing regions of Iraq and Saudi Arabia that remain unexplored, fusion reactors that are “oh so close,” and a plethora of other under-developed alternative energy technologies. These issues are irrelevant for evaluating the current status of Peak Oil. This does not mean that we shouldn’t discuss them or that we should write them off as possible energy sources. When these possibilities are available as concrete energy resources they can be incorporated into our analysis. This new model might be a more general hypothesis of Peak Energy such as the one proposed by Richard Duncan [3] in which he accounts for net energy loss in the conversion of hydrocarbons to electricity.

Other data that have been touted as supporting Peak Oil are equally irrelevant with respect to our scientific approach. Examples are the “the War on Terror,” downsizing of proven reserves by Royal-Dutch Shell, September 11, and the Patriot Act. These issues are economically, sociologically, and geopolitically interesting and worthy of discussion, but they are at best derivatives of oil data and are not independent of the amount left in the ground. Thus, while this circumstantial evidence is fascinating and may even be a predicted outcome of Peak Oil modeling [3], it has no place in a fundamental test of the Peak Oil hypothesis.


In the simplest of all possible worlds, we would simply tabulate all of the data related to oil production and depletion. However, the reality is that the information content of the data matters. Various factors impinge on this issue of quality.

The credentials of the person presenting the data can be important. I’m an evolutionary biologist, not a petroleum geologist. I am (marginally) qualified to spout off about hypothesis testing but not at all competent to expound on the geology, monitoring, or modeling of petroleum resources. Ultimately, however, it is not sufficient to rely on one’s credentials to sell data (“I know more than you, therefore I’m right”). The data must be supported in peer review. Although most of us are largely at the mercy of the petroleum geologists in this regard, we should at least be able to access their argumentation.

Another important feature of data quality is relative importance. Drilling the Arctic National Wildlife Refuge has been touted as a means to “energy independence” for the U.S. However, there is relatively little oil there as compared with Ghawar Field in Saudi Arabia, the largest in the world and the health of which is currently under discussion [4]. Relative importance should also be standardized with regard to the energy profit ratio (EPR), which is the amount of energy invested divided by the energy returned (see [5] for an excellent discussion of EPR). A large conventional oil field at the peak of production will have an EPR as high as 50:1, but the EPR falls precipitously as the field declines [5].

Unconventional oil resources such Canadian tar sands have a considerably lower EPR of less than 2:1 [6]. A useful but daunting task might be to refine oil production models that reflect the amount of recoverable energy rather than simply the amount of recoverable liquids, thus bringing the model more into line with the concept of net energy [7]. The division of reserves into conventional (energetically easy to access and process) versus unconventional (energetically intensive to access and process) in current models [2] is certainly moving in the right direction.

While data quality is an important issue in evaluating the relative support for Peak Oil, it often leads down a slippery slope toward arbitrariness, subjectivity, and appeal to “expert” status on the part of the presenter. Thus, deciding how to weight various data points can lead to endless and sometimes pointless debate. One possible solution is to test a variety of models under different assumptions about the data to see how the outcome is affected, a process sometimes referred to as sensitivity analysis. An example of this type of analysis is the modeling Peak Oil forecasts under different scenarios of oil discovery (e.g. [8]).

Perhaps the crucial issue is the inherent accuracy of the original data themselves and the honesty of the data analyst. Unfortunately, with respect to oil it appears also to be the subject to a great deal of disinformation. In the end, what we are trying to achieve is data transparency. This transparency has been sorely lacking on most fronts, both from the oil-producing entities and those attempting to model production and decline. In a recent critique of the “Hubbert Curve,” Michael Lynch cited a number of instances in which depletionists have been inconsistent in their treatment of production data, and he criticized the use of a proprietary database, which violates the principle of data transparency [9]. Likewise, energy banker Matthew Simmons has made an appeal for transparency in reserve estimates with his “13 Points of Light” proposal [10].


Now that we have some of the logical framework sketched out for testing the Peak Oil hypothesis, we can better evaluate the recent debate around abiotic oil.

For those not familiar with the issue, here’s a thumbnail sketch. The textbook explanation for oil formation is that it is formed from previously living matter that ultimately used sunlight to fix carbon. Dead organisms accumulate, are buried in a process called sedimentation, and are then “cooked” at the right temperature and pressure to produce a variety of hydrocarbons, including oil. This geological process of accumulation, sedimentation, and cooking is very slow and cannot possibly occur on a timescale rapid enough to replace the oil being pumped out of the ground. Note that it is the discrepancy between the timescale of formation relative to depletion that is fundamental to the notion that oil is a nonrenewable resource.

The abiotic theory of oil formation postulates that oil is formed by chemical processes deep in the earth’s crust that do not require living material as a carbon source. This “abiogenic” oil then percolates toward the surface where it accumulates in reservoirs that are accessible to drilling. This has major implications. First, oil is being continuously produced from energy stored deep within the earth’s core and has the potential to refill depleted reservoirs. If this process proceeds rapidly enough relative to depletion, then oil could be considered a renewable resource. Second, oil could potentially accumulate in a variety of reservoir types. Therefore, geological formations not previously thought to be worthy of exploration might yield significant reserves. For detailed descriptions of the geology and economic consequences of abiotic oil, see www.gasresources.net

Let’s put this into the context of the hypothesis of Peak Oil. First, “oil” is a class of molecules. If our definition of that class includes “hydrocarbons formed from previously living things” then abiotic oil either can’t be a member of that class or we need to modify our class boundaries. Moreover, the statement “all oil comes from living things” is clearly falsified by abiotic oil. No more, no less. Period. However, previously we established that Peak Oil is an event that has characteristics of an individual. Therefore, falsifying the statement that “all oil comes from living things” does not falsify the reality of Peak Oil. The Peak Oil hypothesis is tested by bringing all the available, relevant data to bear on the question of depletion. This will include all oil generated from abiotic sources, if such oil exists.

The crucial point, then, is that the discussion about abiotic oil is really a debate about some of the data that might be used to test the Peak Oil hypothesis. Documented abiotic fields, fields that are “refilling,” and new fields missed in previous oil exploration should be incorporated into our dynamic evaluation of oil depletion, but this will require actual estimates. Even if abiotic oil has been shown to be geochemically feasible, even it has been extracted from the ground in significant quantities, even if it can be shown to be replenishing some depleted reserves, and even if it is shown to be the source of all petroleum, the real issue is its ultimate effect on depletion/production models. Simply “proving” (we don’t generally prove things in science, only support or refute hypotheses) that abiotic oil exists does not refute Peak Oil. This is not to say that a discussion of abiotic oil is a waste of time, only that so far the significance of abiotic oil relative to Peak Oil has been completely overblown.

Let’s also make a clear distinction between the data and the debate. We are interested in the data that test Peak Oil. The Peak Oil discussion has been littered with blogs dripping with sarcasm, blustery thousand-dollar challenges, accusations of who failed to show up for which debate, the employment history of participating geologists, economists, and academics on and on ad nauseum. Some of this sparring may relate to issues of data quality, but for the most part these exchanges are tactical posturing that has nothing to do with evaluating the Peak Oil hypothesis. This is not to say that science operates outside the sociopolitical realm. The biologist Lysenko set Russian genetics back decades by using his political connections to focus their national research program on the inheritance of acquired characteristics in wheat [11]. The unfortunate reality is that, like Lysenko, there are those in the world of oil who are far more interested in furthering their own agendas than in pursuing the truth.

With regard to Peak Oil, the first and very urgent order of business should be to get the best available data tabulated on both sides of the hypothesis. I understand this is no small undertaking, but this is no small issue either. Secondly, the data must be made publicly available in an honest, accessible, and hopefully digestible format.


The depth and breadth of my bias should be transparent. I’m an academic with expertise in an arcane branch of evolutionary biology involving the reconstruction of evolutionary trees. I’ve been interested in hydrocarbon depletion for about two years but claim no expertise in that field beyond that of a scientist trying to evaluate the data. I subscribe to Mike Ruppert’s “From the Wilderness.” This is hardly a case of hero worship, but I find his geopolitical perspective a welcome breath of fresh air in the stultified air of corporatized, self-censoring mass media.

Do I believe in the logical structure of Peak Oil as an hypothesis? Absolutely. Do I also believe that what I have outlined as a course of data transparency and testing will happen soon enough to matter? Not really. The data up to this point strongly suggest that we are using the stuff up a lot faster than it is being found. It doesn’t take a “brainiac” to figure out that even if further exploration results in more oil, it’ll be gone sooner than later, that is, if we don’t fry ourselves first with an apocalypse of nuclear war and/or irreversible global warming. Moreover, the analyses of Duncan and Younquist [8] indicate that an imminent global peak in production can only be postponed for a short time even under scenarios that include very large new discoveries. Finally, human consciousness has not resulted in any self-regulation, whether it is reproduction in the third world or consumption in the first. The smart money is with the laws of physics and biology. You’re welcome to call that pessimistic. I call it reality.

I reserve (no pun intended) the right to change my mind. But based on what I’ve seen so far I’ve given in to the end of oil and am moving on to the acceptance stage in a grieving process over the imminent loss of Industrial Civilization. I continue to live in the dominant culture of technology, economic growth, and consumerism. I’m typing this essay on a laptop computer with a broadband Internet connection. Hell, I’m remodeling my bathroom! But psychologically I’ve ditched this scene and have created a certain amount of detachment from it. In my alter life I’m trying to develop coping strategies for future upheaval and for the continuation of humanity beyond hydrocarbons.
I’m also a husband, the father of two teenage kids, and the son of aging parents. Every single day I wonder what the immediate future holds for my family and what I can do now to best plan for it. I wonder what to tell my kids. My wife says we can’t “steal their hope” so for the time being I try to keep my mouth shut. I don’t sleep well. My close friends say “don’t worry, be happy.” My response is that I hope they are right and that we can celebrate in 2010 by going out to a nice dinner, clinking glasses, and having them say, “I told you so.” My ego will happily take the punishment.

Time will soon tell if this was all for nothing, whether the Peak Oil hypothesis will ultimately be falsified and pitched in an ashcan with a picture of Cassandra and the label “Doomsday Nonsense.” But right now it appears there is a tremendous amount of work to do, and it makes me extremely angry to see the Peak Oil discussion, perhaps the most important discussion that humanity should now be carrying on, trivialized by taunting, character assassination, irrelevant information, and poorly formed arguments. It’s not a perfect world, but certainly we can do better

1. Ghiselin, M.T., Metaphysics and the Origin of Species. 1997: State University of New York. 377

2. Campbell, C.J., Oil and Gas Liquids 2004 Scenario. 2004: Uppsala Hydrocarbon Depletion Study Group, www.peakoil.net/uhdsg/Default.htm.

3. Duncan, R.C., The Peak of World Oil Production and the Road to the Olduvai Gorge. 2000: Geological Society of American (GSA), www.hubbertpeak.com/duncan/olduvai2000.htm.

4. U.S.-Saudi Relations and Global Energy Security. 2004: Center for Strategic and International Studies, www.csis.org/energy/040427agenda.htm

5. Fleay, B.J., B Eng, and M. Eng, Climaxing Oil: How Will Transport Adapt? 1998: Chartered Institute of Transport in Australia, National Symposium, wwww.istp.murdoch.edu.au/teaching/N212/n212content/topics/topic5/02

6. Youngquist, W., Geodestinies: The Inevitable Control of Earth Resources over Nations and Individuals. 1997, Portland, Oregon: National Book Company. 500

7. Hanson, J. Energetic Limits to  Growth. 1999: Business Communications Co, Inc., www.dieoff.com/page175.htm

8. Duncan, R.C. and W. Youngquist, Encircling the Peak of World Oil Production. Natural 1999: Resources Research, 8: 219-232

9. Lyncy, M., The New Pessimism about Petroleum Resources: Debunking the Hubbert Model (and Hubbert Modelers). 2004: Gas Resources Corporation, Houston. www.gasresources.net/

10. Simmons, M., Matt Simmons on the State of the Energy Industry and His 13 Points of Light. 2004: Global Public Media, www.globalpublicmedia.com/INTERVIEWS/MATTSIMMONS/Matt

11. Joravsky, D., The Lysenko Affair. 1970: Cambridge, Harvard University Press. 459