“For a successful technology, reality must take precedence over public relations, for Nature cannot be fooled.”

Richard Feynman


Olduvai Theory: Sliding Towards a Post-Industrial Stone Age, by Richard Duncan

by Richard C. Duncan, Ph.D.
Institute on Energy and Man,
June 27, 1996

Indeed, the ability to control energy, whether it be making wood fires or building power plants, is a prerequisite for civilization.
— Isaac Asimov, 1991


In 1989, I concluded that the life-expectancy of Industrial Civilization is horridly short. This hypothesis was defined in terms of a measurable index, world energy-use per person, and named the “transient-pulse theory of Industrial Civilization.” I sketched its maximum point at 1990, followed by a persistent decline (see Note 1). Back then, however, I had no data to support this claim.

The ratio of world annual energy-use to world population gives a robust, testable profile of Industrial Civilization. Over the past six years, I devised a quantitative basis for the theory and gathered several sets of world energy and population data to test it (Note 2). In these pages, the name “Olduvai theory” means the same as “transient-pulse theory,” used in previous papers (Note 3).

[ Note 1: ‘Industrial Civilization’ includes all capital investments and international trade agreements such as GATT, EU, and NAFTA. ‘GATT’ means General Agreement on Tariffs and Trade; etc.
   Note 2: ‘World energy’ includes oil, natural gas, coal, nuclear- and hydroelectricity. Energy and population data are available from several sources, e.g., United Nations publications.
   Note 3: Since the 1950s, the Olduvai Gorge in Tanzania has been strongly associated with human origins and the Stone Age way-of-life. In this discussion, ‘Olduvai theory’ is a metaphor. It suggests our impending return to a Stone Age way-of-life.]

Please have a quick look at Figures 1 and 2. But before getting into the details, in the next section I’d like tell a detective story.


I would rather discover a single fact, even a small one, than debate the great issues at length without discovering anything at all.
— Galileo Galilei, c. 1640

My Odyssey with the Olduvai theory began thirty-two years ago during a lecture series titled, Of Men and Galaxies, given at the University of Washington by cosmologist Sir Fred Hoyle.

It has often been said that, if the human species fails to make a go of it here on Earth, some other species will take over the running. In the sense of developing high intelligence this is not correct. We have, or soon will have, exhausted the necessary physical prerequisites so far as this planet is concerned. With coal gone, oil gone, high-grade metallic ores gone, no species however competent can make the long climb from primitive conditions to high-level technology. This is a one-shot affair. If we fail, this planetary system fails so far as intelligence is concerned. The same will be true of other planetary systems. On each of them there will be one chance, and one chance only. (Hoyle, 1964; emphasis added)

I was fascinated—and stunned. His soft-spoken proposal seemed incredulous, bizarre, preposterous—and possibly inevitable. A return to the Stone Age? Deep cultural and material impoverishment? However nobody else in the audience seemed the least concerned. Perhaps Hoyle was just giving a lead-in to his next science fiction thriller. So for the next decade I went about my way: raising kids, building airplanes and teaching engineers. Haunted by Hoyle’s hypothesis.

Then in 1975 and 1976, conferences took me to Colorado and, by-the-by, to Mesa Verde National Park where the magnificent, long-deserted cliff dwellings of the Anasazi made it clear that all civilizations are ephemeral. But Fred Hoyle wasn’t reiterating the tired old-saw of historian-philosophers such as Spencer, Spengler, Sorokin and Toynbee: i.e., the endless rise-and-fall cycles of civilizations. He was talking about something quite different, more profound, more pervasive. Global Industrial Civilization has no cycles at all. It’s “a one-shot affair.” Exponential growth, exponential decline. That’s it.

Time passed, and the years from 1985 to 1992 found me working for power company in Saudi Arabia. While there, I traveled widely, including visits to Ethiopia, China, India, the (then) USSR, etc. Mainly I traveled to answer the question, “Is Fred Hoyle right?” The question, of course, was not about the durability of any one of these nations (most looked fragile, some non-existent), but about the life-expectancy of Global Industrial Civilization itself.

Based on what I’d seen around the world, industrialization isn’t evolving toward sustainability. Just the opposite. Hoyle was right. “This is a one shot affair…. there will be one chance, and one chance only.” So the real question was, “How long will it last?” A thousand years? A million years? Or what? So in 1989, just before leaving on a trip to East Africa (and incidentally, the Olduvai Gorge), I dug through some books. There was no lack of speculation. Estimates differed wildly; a summary appears in Table 1. [ For completeness, my 1991(a) estimate is included. ]

Table 1. Estimates of the Life-Expectancy of Industrial Civilization

Haldane1927“39 million years”
Russell1949“it cannot long continue”
Drake1961one million years
Watson1969potentially “millions of years”
Arrester1971natural response, about 200 years
Meadows, et al.1972natural response, 100-200 years
O’Neill1976“even our success becomes failure”
Leakey1977about 100 years
Harris1977“a bubble-like nature”
Crick1981short to 10,000 years or more
Laszlo1987“extremely short” to very long

Back in Saudi Arabia, I began a paper for presentation at the American Society of Engineering Educators Conference in New York, October, 1989. The title was, “Evolution, Technology, and the Natural Environment: A Unified Theory of Human History.” So to say, the “coming-out” of the Olduvai theory. I concluded,

The broad sweep of human history can be divided into three phases:

  • The first, or pre-industrial phase was a very long period of equilibrium when economic growth was limited by simple tools and weak machines.
  • The second, or industrial phase was very short period of non-equilibrium that ignited with explosive force when powerful new machines temporarily lifted all limits to growth.
  • The third, or de-industrial phase lies immediately ahead during which time industrial economies will decline toward a new period of equilibrium, limited by the exhaustion of non-renewable resources and continuing deterioration of the natural environment. (Duncan, 1989)

In that paper I used “world average energy-use per person” as a measurable indicator of Industrial Civilization. I sketched the peak at 1990. Only one problem— I had no hard data to test the theory. But I did have Hoyle’s hypothesis and my own round-the-world observations of global conditions and trends.

Then, in the nick-of-time, an article appeared in the September issue of Scientific American showing that world average energy-use per person had peaked in about 1973, and had since gone into steep decline (Gibbons, et al., 1989; their curve is included in Figure 2). At the end of my presentation, a handful of engineers from the audience gathered around and we discussed my theory and its consequences. Most agreed. However I needed more data.

Again I returned to Saudi Arabia. But there, with meager access to data, I struggled with several more papers (e.g., Duncan, 1990; 1991a; 1991b). Then somehow the editor of a small journal got wind of what I was doing and invited me to submit an article. It was published as, “The Life-Expectancy of Industrial Civilization: The Decline to Global Equilibrium” (Duncan, 1993c). In the main it concluded,

  • Industrial Civilization can be described by a single pulse waveform of duration X, as measured by average energy-use per person per year.
  • The life-expectancy of Industrial Civilization is less than one-hundred (100) years: i.e., X < 100 years.

But data was still lacking.

I moved back to the USA, and by early-1993 had tested the Olduvai theory against two new sets of data. The energy data was from the energy industry itself, and the population data was from the United Nations [i.e., BP (1992) and UN (1992)].

Eureka! The Olduvai signature appeared at last. The British Petroleum and United Nations data confirmed that world per capita energy-use had peaked in about 1978 and subsequently had declined (Duncan, 1993b). Only the Olduvai theory could explain the peak and decline. In contrast, both the “exponential growth theory” (of “mainstream” economics) and “steady-state” theory (of “utopian” economics) failed.

By late-1993, I had further tested the Olduvai theory against several more sets of data, e.g., Davis (1990) and the UN (1993). In both cases, the historic peak occurred in about 1978. Again, the Olduvai theory prevailed (Note 4). I reported these results in a paper, “Sustainability—Is There a Middle Road?: The Transient-Pulse Theory of Industrial Civilization” (Duncan, 1993a).

[ Note 4: In the foregoing tests, I purposely avoided any dull mathematics. Only “long-division” was used: i.e., the ratio of world energy-use to population. Any 6th grader could do it. That’s important. ]

While new energy and population data was coming in, I spent the next two and a half years developing a better method for predicting the energy production life-cycle (see Duncan, 1996). Although theoretically important, that work isn’t relevant here because the Olduvai theory is arbitrated by historic data only.

Next, we’ll take a closer look at the theory.


The moon landing may be our Great Pyramid, an accomplishment never to be equaled.
— Alan Cromer, 1993

Figure 1 is qualitative. Descriptive only. A visual aide. Right-brain stuff. It’s a sketch of the Olduvai theory. The “Olduvai signature,” I call it. So please don’t try to scale out the horizontal or vertical axes. (We’ll do that later.)

Figure 1. The Olduvai Theory of Industrial Civilization

1. Pre Industrial Phase [c. 3 000 000 BC to 1765]

  • A – Tool making (c. 3 000 000 BC)
  • B – Fire used (c. 1 000 000 BC)
  • C – Noelithic agricultural revolution (c. 8 000 BC)
  • D – Watts steam engine of 1765 Industrial Phase (1930-2025)

2. Industrial Phase [1930 to 2025, estimated ]

  • E – Per capita energy-use 37% of peak value
  • F – Peak energy-use
  • G – Present energy-use
  • H – Per capita energy-use 37% of peak value

3. Post Industrial Phase [c. 2100 and beyond ]

  • J, K, and L = Recurring future attempts at industrialization fail.
    Other scenarios are possible.

[Note 5: In Figure 1, it may be helpful to think of the curve as income per person per year in dollars. Or perhaps as material standard of living. Better yet, just remember the little cartoon folks.]

Figure 1 divides the very long span of human history into three phases: (1) PreIndustrial, (2) Industrial, and (3)Post-Industrial. Seven events are marked on the left part of the curve (i.e., points A through G). Likewise, five hypothetical events are marked on the future part of the curve (i.e., H through L).

Phase 1, the Pre-Industrial Phase, spans thousands of millennia of sustainable conditions when society was powered exclusively by (renewable) solar energy. It began some three million years ago when our hominid ancestors started making simple tools (point A, Figure 1). The tools, in turn, made possible greater energy-use in such forms as food, fiber and shelter. Epic milestones leisurely passed, including the use of fire at about one million BCE and the Neolithic Agricultural Revolution at about 8,000 BCE. The end of the Pre-Industrial Phase is marked at 1765, the year James Watt invented the condensing steam engine (point D, Figure 1).

Phase 1 was followed by a transition period—i.e., The Industrial Revolution— delimited by the years 1765 and 1930 (points D and E, Figure 1).

Phase 2, the Industrial Phase, comprises the shaded portion of Figure 1. The life expectancy of Industrial Civilization is defined as the duration in years (x) between the leading and lagging “37% points” (i.e., points E and H). It is a short, extravagant period when transportation, commerce and industry were powered predominantly by (nonrenewable) fossil-fuels. Historic data (presented later) quantifies the peak period of the curve: i.e., the years between points E and G. Using that data, I mark the beginning of the Industrial Phase at 1930 (point E), the year average energy-use per person reached 37% of its peak value.

Note that the peak of Industrial Civilization was reached in about 1977 (point F), less than fifty years after it began. More significant, Figure 1 identifies the global energy “watershed”. For the first time in the gaping millennia of human existence, average per capita energy-use peaked and began to decline!

As I read it, the descent into the Olduvai valley will be steep and swift. A scenario of Phase 3, the Post-Industrial Phase, is sketched in Figure 1 (i.e., from point I onward) wherein Industrial Civilization has disintegrated into farming villages, kinship tribes and rogue bands. The surviving population will have “achieved” permanent sustainability—at the subsistence level.

Of course, other scenarios are possible. For example, “The human species may follow the road to extinction rather than revert to the berry-picking stage” (Georgescu-Roegen, 1971). Or more recently, “The danger of extinction is real … It is time to face the facts” (Leslie, 1996). However, because the circumstances of human society beyond the end of the second phase (i.e., point H. Figure 1) don’t effect my thesis, the third phase is de-emphasized in the remainder of this discussion.

Hard to believe? Yes indeed. So let’s do the numbers.


Physicists learned to realize that whether they like a theory or they don’t like a theory is not the essential question. Rather, it’s whether or not the theory gives predictions that agree with experiment.
— Richard Feynman, 1985

Figure 2 is quantitative. Numeric. Left-brain stuff. It’s an accurately scaled graph showing the peak period of Industrial Civilization between 1950 and 1995. So please do scale out the horizontal and vertical axes. And if you’d like to go back to 3,000,000 BCE, Isaac Asimov (1991) gives all the numbers.

Figure 2. World Average Energy-Use Per Person
Comparison of Four Sets of Historic Data

“BOE” means Barrels of Oil Equivalent

As far as I know, credit goes to Robert H. Romer (1985) for first publishing the peak-period data for world per capita energy-use. He gives the peak at 1979, followed by a sharp decline through 1983, the last year of his data. However, this information was published as a relatively opaque worksheet. And curiously, no mention was made about the energy watershed. His data are graphed in Figure 2.

Credit likewise goes to Gibbons, et al. (1989; see Note 6) for an early publication of the peak-period of world per capita energy-use. The authors displayed the data as a viewer-friendly graph that peaked in 1973, followed by a steep downward slope through 1985. Here again, no mention was made about the significance of the peak or decline. Their curve is included in Figure 2.

[Note 6: Dr. Gibbons is Science Advisor to President Clinton.]

As previously mentioned, in 1993 I published two papers containing extensive world per capita energy-use data and presented that data as both worksheet values and plotted graphs. Moreover, I emphasized the importance of the peak and the implications of long-term decline. My first paper (1993b) shows the peak at 1978 and decline through 1991. My second paper (1993a) shows the peak at 1980 and decline through 1992. The two-year difference is due to the use of independent sets of data. In June 1996, I updated my tests of the Olduvai theory. This latest test shows the peak at 1978 followed by decline through 1995. The data are graphed in Figure 2.

A separate data test of the theory was made by F. M. Wright in June 1996. He showed the per capita energy-use peak at 1978 and overall decline continuing through 1994, as graphed in Figure 2. The slight difference between our curves results from the fact that Wright used US Census Bureau data for world population, while I used UN data. Note however that our curves are virtually superimposed.

Table 2 (below) summarizes the data tests. These tests show that, (1) on the average, world per capita energy-use peaked in 1977, and (2) the subsequent rate of decline has been about 0.90% per year. The Olduvai theory explains this data. In contrast, however, the exponential-growth theory and the steady-state theory both fail. While Table 2 isn’t (yet) the Rosetta Stone of Industrial Civilization, each new set of data takes it one year closer.

Table 2. Data Tests of the Oduvai Theory

Years of
Rate of
et al.

* “BOE” means Barrels of Oil Equivalent

Now let’s stand back and look at the big picture.


On the whole, however, it is only out of pride or gross ignorance, or cowardice, that we refuse to see in the present the lineaments of times to come.
— Marguerite Yourcenar, 1951

Mental blockbusters have exploded throughout the history of human inquiry. “Revolutions” they’re called. But typically they only pricked human egos, and ruffled vested interests and tired-old dogmas. Thus, the past discoveries (such as the solar-centric theory) were benign because such psycho-threats could simply be flouted or ignored. But the Olduvai theory is different because, willy-nilly, it will adversely impact the lives of almost everybody.

Back in 1989 I became deeply depressed when I concluded that our greatest scientific achievements will soon be forgotten and our most cherished monuments will crumble to dust. But more so, I knew that my children would feel the pressure, and will likely suffer. That really hurt.

In time however, my perspective changed. Now I just treat the Olduvai theory like any other scientific theory. Nothing personal. Each year, I gather the data. Update Figure 2. And watch the theory unfold. Let the chips fall. What else?

Still, the impending Post-Industrial Stone Age is a tragedy because it really isn’t inevitable. There’s no absolute reason why we couldn’t live in material sufficiency on this planet for millions of years. But prudence isn’t our forte. “Even our success becomes failure.” And, in a way, it’s not our fault. Long ago Natural Selection dealt us a bad hand—we’re sexually prolific, tribal, short-term and self-centered. And after thousands of years of trying, Culture hasn’t changed that. And there is no sign that She will.

Backward to the future. Forward to the past. Almost perfect symmetry.


Industrial Civilization doesn’t evolve. Rather, it rapidly consumes “the necessary physical prerequisites” for its own existence. It’s short-term, unsustainable. “This is a one shot affair…. there will be one chance, and one chance only.”

Energy-use per person is used as a measurable index of industrialization. In 1989, I proposed the Olduvai theory of Industrial Civilization, as illustrated in Figure 1.

  • Industrial Civilization can be described by a single pulse waveform of duration X, as measured by average energy-use per person per year.
  • The life-expectancy of Industrial Civilization is less than one-hundred (100) years: i.e., X < 100 years.

In 1989 I postulated that per capita energy-use had peaked and was already on the decline. But back then I lacked sufficient data to test the theory.

By 1996, however, I had successfully tested the Olduvai theory against numerous sets of data. Four of these tests are graphed in Figure 2. The following facts emerge.

  1. On the average, world per capita energy-use reached a maximum value (i.e., a peak) in 1977.
  2. The 1977-1995 rate of decline has averaged 0.90% per year.
  3. Per capita energy-use will continue to decline as long as the world population growth rate exceeds the energy growth rate.
  4. If the decline continues (and extinction is avoided), human societies will bottom out at the subsistence level of energy-use.

The Olduvai theory explains the Figure 2 data, but the exponential-growth theory (of mainstream economics) and the steady-state theory both fail.

The Olduvai theory cannot be overthrown (i.e., scientifically rejected) by outrage or indignation. However, it can be overthrown by either, (1) demonstrating that the four sets of data in Figure 2 are in error, or (2) by gathering additional data over the next few decades and demonstrating that the Olduvai theory cannot explain that data. In any case, the data will be the final arbiter.

Suggestion: If you’re a ‘lady or man from Missouri’ or a ‘doubting Teresa or Thomas’ (and you should be), then go to your library, get the data, and test the Olduvai theory yourself. I’ll gladly include your findings in the next update of Figure 2.


Asimov, I. & White, I. (1991) The March of The Millennia: A Key Look at History. New York: Walker.
BP (1996 & previous ed.). BP Statistical Review of World Energy. British Petroleum Company, London.
Cromer, A. (1993). How high, high-tech? Northeastern University Magazine, May. Boston.
Davis, G. R. (1990). Energy for planet earth. Scientific American 263, 21-27.
Duncan, R. C. (1996). The Mexican petroleum ‘play’ in two ‘acts’: Taking hold of oil production data. System Dynamics Conference Proceedings. System Dynamics Society, Cambridge, MA.
Duncan, R. C. (1993a). Sustainability—Is there a middle road? Moses Greeley Parker Lecture Series. Lowell, MA.
Duncan, R. C. (1993b). The realities of world energy production: A prediction based on historic data. Humanist Association of Massachusetts, Cambridge, MA.
Duncan, R. C. (1993c). The life-expectancy of Industrial Civilization: The decline to global equilibrium. Population and Environment 14, 325-357.
Duncan, R. C. (1991a). The life-expectancy of Industrial Civilization. System Dynamics Conference Proceedings (Bangkok). Systems Dynamics Society, Cambridge, MA.
Duncan, R. C. (199lb). The evolution of social control: Is a world society governable? Proceedings of the Preparing for a Sustainable Society Conference, Ryerson Polytechnical Institute, Toronto.
Duncan, R. C. (1990). A unified theory of human history: Summary presentation. AESR Newsletter, April, v. III, n. 1.
Duncan, R. C. (1989). Evolution, technology and the natural environment. Proceedings of the ASEE Conference, Binghamton, New York.
Feynman, R. P. QED, Princeton University Press, Princeton, NJ.
Georgescu-Roegen, N. (1971). The Entropy Law and The Economic Problem. The University of Alabama Distinguished Lecture Series #1.
Gibbons, J. H., Blair, P. D., & Gwin, H. L. (1989). Strategies for energy use. Scientific American, 3, 86-93.
Hoyle, F. (1964). Of Men and Galaxies. University of Washington Press, Seattle.
Leslie, J. (1996). The End of the World: The Science and Ethics of Human Extinction. Routledge, London.
Romer, R. H. (1985). Energy Facts and Figures. Spring Street Press, Amherst, MA.
UN (1996 & previous ed.). Statistical Yearbook. United Nations Organization, New York.
Wright, F. M. (1996). Personal communication. Seattle, WA.
Yourcenar, M. (1951). Memoirs of Hadrian. (1968 ea.) Penguin, Middlesex, UK.

Download Duncan & Youngquist’s new world oil production model [Dead Link]