In 1972, Donella Meadows, Dennis Meadows, Jørgen Randers, and William Behrens III published The Limits to Growth, a book about the consequences of unchecked population growth and economic growth. The book was very popular at the time, selling 12 million copies. As a part of my work on the project "Can we know what to do about AI?" I did a preliminary investigation of the claims in the book, whether they've been born out, and what the book's impact has been.

The book uses the framework of "systems dynamics," which was pioneered by Jay ForresterPaul Krugman criticized Forrester as having been unaware of prior overlapping work by economists. So it's possible that the ideas in The Limits to Growth are less novel than it appears on the face of things. I haven't investigated the extent to which this is the case, but may do so later. This blog post focuses on The Limits to Growth.

My initial impression of the book based on what people have written about it was very different from the impression that I formed upon reading the book. The book has been misrepresented (whether denotatively or connotatively) both by critics and by sympathizers. In the first section below, I discuss how the book has been misrepresented, and in the second section I discuss what the book says.

Misrepresentations of The Limits to Growth

In his 2008 paper A comparison of limits to growth with 30 years of reality, Graham Turner describes how the book has been misrepresented by critics:

This is perhaps partly a result of sustained false statements that discredit the LtG. From the time of its publication to contemporary times, the LtG has provoked many criticisms which falsely claim that the LtG predicted resources would be depleted and the world system would collapse by the end of the 20th century. Such claims occur across a range of publication and media types, including scientific peer reviewed journals, books, educational material, national newspaper and magazine articles, and websites (Turner, unpublished).

An article in The Nation seems to suggest that the book's reputation has been tarnished by association with others who made outlandish claims:

[Bjorn] Lomborg and others have accused Limits of hysteria. The book is not hysterical. It was, however, taken up by a social milieu heavily populated by hysterics like the peak oilers and the (sometimes crypto-racist) population bugs. Paul Ehrlich embodies this most clearly: for almost fifty years, he’s been pushing a simple-minded Malthusian condemnation of population growth. In the late 1960s, with world population at 3.5 billion, Ehrlich said in a TV interview: “Sometime in the next fifteen years, the end will come. And by ‘the end’ I mean an utter breakdown of the capacity of the planet to support humanity.” Limits, meanwhile, never made such an outlandish claim. 

What The Limits to Growth actually says

Based on my reading of the book:

  • The authors make very few, if any, strong claims. Their core claim is that if exponential growth of resource usage continues, then there will likely be a societal collapse by 2100. They're very explicit about not making specific predictions, and their remarks carry a strong connotation that the scenarios that they work out are intended as thought experiments.
  • Overall, I'm very impressed by the book's epistemic standard. The authors are generally careful to qualify their claims as appropriate. The book doesn't look naive even in retrospect, which is impressive given that it was written 40 years ago.
  • Some people have the impression that book didn't address substitutability (the fact that when a resource becomes scarce, its price goes up, and this incentivizes people to come up with lower cost substitutes for the resource, lowering the need for the resource). This is not the case — the authors address substitutability at length in Chapter Four. 

    They argue that

    (a) Even if we had unlimited resources, exponential growth of resource use would still result in exponential growth of pollution, unless pollution is curbed.

    (b) The marginal cost of reducing pollution grows astronomically as pollution tends to zero, so that it might not be possible to quell the increase in pollution coming from an exponential increase in resource usage.

    (c) A sufficiently large increase in pollution could lead to societal collapse.

    so that there's a need to limit resource usage.

Policy recommendations

The authors discuss potential ways to mitigate the problems that they identify, but only at a theoretical level. They recognize the complexity of the implementation details, and don't make policy recommendations.

There may have been examples of people trying to implement policies based on the book and surrounding literature. I plan on investigating this further. I'd appreciate any references from readers.

Subsequent work by the authors

According to a synopsis of the authors' 30 year update, 

Now in a new study, Limits to Growth: The 30-Year Update,the authors have produced a comprehensive update to the original Limits, in which they conclude that humanity is dangerously in a state of overshoot. While the past 30 years has shown some progress, including new technologies, new institutions, and a new awareness of environmental problems, the authors are far more pessimistic than they were in 1972. Humanity has squandered the opportunity to correct our current course over the last 30 years, they conclude, and much must change if the world is to avoid the serious consequences of overshoot in the 21st century.

I have not reviewed this work. It may not have high relevance to the project at hand, because it could be that not enough time has passed for it to be possible to check the veracity of the substantive predictions. I may investigate further.

Graham Turner's 2008 paper A comparison of limits to growth with 30 years of reality argues that the trajectory of civilization since the publication of The Limits to Growth is in line with a scenario in the book that leads to societal collapse. The significance of this is unclear, because past historical trends need not predict future historical trends.

Other relevance to MIRI

The book has relevance to addressing AI risk that extends beyond the project at hand.

Like MIRI, the authors were interested in coping with a threat that the world had never faced before (on such a large scale), and one that could arrive with little notice. The authors make the point that people tend to think in terms of linear growth and decay rather than exponential growth and decay, and that unchecked exponential growth of resource usage could result in a very sudden problem (by human standards) of great scarcity and/or pollution.

As such, it may be fruitful to take a closer look at the authors' remarks on these points.

New Comment
16 comments, sorted by Click to highlight new comments since:

Like MIRI, the authors were interested in coping with a threat that the world had never faced before (on such a large scale), and one that could arrive with little notice. The authors make the point that people tend to think in terms of linear growth and decay rather than exponential growth and decay

I sometimes think of "Limits to Growth" and "Unfriendly AI" as rivals for the trophy of worst existential risk, with the folks who are most concerned about one of the risks viewing the folks concerned about the other with deep suspicion. ("Huh... your pet disaster will never happen; we should be worrying about mine instead!").

It's certainly useful to identify the common ground between the camps. Both scenarios involving taking existing exponential trends very seriously. In both cases, society as a whole has not been taking the trends seriously (or is even actively engaged in dismissal or ridicule of the future concerns). In both cases the lack of preparedness means that there are unlikely to be good results.

Can you elaborate on what type of societal collapse was (conditionally) predicted? What kinds of things were they saying would happen in this scenario?

Is this post broken? This seems to be cut off halfway.

Possibly a draft that was posted by accident before it was ready?

Yeah, that was the situation, I just finished and posted. Thanks.

There may not have been any misrepresentations of The Limits to Growth.

Yes, thanks

The marginal cost of reducing pollution grows astronomically as the fraction that you want to reduce it by tends toward zero,

Minor point, I had to re-read this a few times to get what you meant. The use of the adverb "by" made me think that you meant the marginal cost of reducing pollution a small amount would be higher than reducing it a large amount.

Thanks, I modified this accordingly.


The marginal cost of reducing pollution grows astronomically as pollution tends to zero

In the book, does this cost refer to the cost of a technological solution to pollution, or does it also/instead refer to the cost of society coordinating to reduce pollution?

[This comment is no longer endorsed by its author]Reply

There's not a clear divide between the two things (e.g. it could be that what's needed is a technology that facilitates coordination). I don't remember the exact wording in the book.


I ask because as the costs of pollution rise, so might the benefits.

[This comment is no longer endorsed by its author]Reply

What are the benefits of pollution?


People don't pollute randomly. They pollute as a necessary by-product of other activities. There aren't, as far as I know, any benefits to pollution per se, but there are benefits to the sorts of activities that produce pollution. The costs of pollution may rise, but that doesn't imply that at any point the cost of pollution won't be worth it (except possibly on the margin due to the externality).

[This comment is no longer endorsed by its author]Reply

Ok, I understand what you're saying now. The matter under discussion is the negative externalities of pollution (perhaps to future generations). I don't know if there's enough Uranium or Plutonium for this to be a realistic hypothetical, but one could imagine a world in which nuclear reactor waste accumulated to such a degree so as to substantially reduce the amount of inhabitable land.

Thermodynamic inefficiency will always produce at least some heat pollution and I think it's safe to predict that the cost of achieving 99.99% efficiency is at least an order of magnitude more expensive than 99.9% efficiency.