This is the latest instance of the WORLD3 model, as in Limits to Growth – the 30 year update, from the standard Vensim distribution. It’s not much changed from the 1972 original used in Limits to Growth, which is documented in great detail in Dynamics of Growth in a Finite World (half off at Pegasus as of this moment).

There have been many critiques of this model, including the fairly famous Models of Doom. Many are ideological screeds that miss the point, and many modern critics do not appear to have read the book. The only good, comprehensive technical critique of World3 that I’m aware of is Wil Thissen’s thesis, Investigations into the Club of Rome’s WORLD3 model: lessons for understanding complicated models (Eindhoven, 1978). Portions appeared in IEEE Transactions.

My take on the more sensible critiques is that they show two things:

  • WORLD3 is an imperfect expression of the underlying ideas in Limits to Growth.
  • WORLD3 doesn’t have the policy space to capture competing viewpoints about the global situation; in particular it does not represent markets and technology as many see them.

It doesn’t necessarily follow from those facts that the underlying ideas of Limits are wrong. We still have to grapple with the consequences of exponential growth confronting finite planetary boundaries with long perception and action delays.

I’ve written some other material on limits here.

Files: WORLD3-03 (zipped archive of Vensim models and constant changes)

8 thoughts on “WORLD3-03”

  1. Thanks for posting the World3 model, it is not included in teh 30th Anniversary CD and I’ve been looking for it everywhere.

  2. Thank you for putting the model onto your page – the original book has it but print is so small and I needed a version that I could use within a dissertation and like the other comment the new book does not have it in.

    I agree that the model is simplified and does not take into account a number of factors, but like all models it is indicative rather than utterly accurate. I agree that the warning and the call for radical change still stands today.

  3. Hello Tom,

    Many thanks for your superb blog. I’m going to tell all my friends about it.

    I learned of you via your excellent [vimeo video](
    Vensim Model Analysis – World Dynamics

    A few words about myself. Science is my hobby. Stephen Jay Gould’s books got me hooked. I read Science and Nature magazines every week. I’ve also read Foreign Affairs magazine for over 40 years. None of which makes me a scientist or foreign policy expert. I am a fan of [Peter Sinclair’s Climate Denial Crock of the Week]( and have contributed financially to his work.
    I have a bachelors in math and masters degree in computer science, but know very little about modeling except that it isn’t a hoax 🙂

    Here, I’d like to ask a question about “model denial”. [This page](
    contains the following from karlmagnus:

    Don’t know about this version, but the original model included exploding exponential error terms. IF you have a model like Forrester’s with exponentials all over the place, and run a 40 year projection, the error terms explode off the screen, so “catastrophe” of some kind is unavoidable whatever assumptions you put in. I told Forrester this in a public meeting in October 1971; didn’t stop the charlatan propagating his nonsense.

    The last sentence marks this guy as a crank, but the critique is “mathy” enough so that I can’t clearly refute it. It misleads with a half truth, namely that roundoff and other numeric problems must be taken seriously in all computer calculations. So my question is, how would you respond?

    Your work is fascinating. I have lots of other questions, but I’ll leave them for later.


    1. Hi Edward –

      This strikes me as a completely vacuous statement. “Error terms” in particular has no obvious meaning. Is karlmagnus asserting that the economy is not growing exponentially? Or that positive feedback doesn’t exist?

      Ironically, you could paraphrase km’s comment to something like, “exponential growth always comes to a catastrophic end,” in which case he’s essentially asserting the Limits message.

      I’d put this in the “crank” bucket. There are legitimate critiques of World3 formulation issues (see for example Will Thissen’s thesis). This is not one.


  4. Yeah, we both agree about km. Let’s forget about him and see if we can improve my intuition. Unless I am mistaken, the farther out a weather forecast goes, the less reliable it is, presumably because the initial conditions aren’t known perfect. Or maybe for other reasons. Otoh, climate models don’t seem bothered by such things. So two questions:

    1. Why aren’t multi-year simulations like weather forecasts?
    2. Why don’t uncertainties “build up” in climate models and world3?

    I may be missing something really basic, but I don’t know what it is. Are there any better questions to ask?


    1. It`s the difference between things you can calculate (known/good estimates of ressources and withdrawals from them, calculating their interactions is then deterministic. Variation of input variables will then show, how much this influences outcomes). And things you cannot calculate, because they are chaotic. Read Gerd Gigerenzer about the fundamental differences between both. Gigerenzer kills most of big data efforts to predict the future this way. Michael (dentist, no expert at all (besides dentistry)).

      1. I’m also not a big fan of attempts to predict the global future with any precision. I think the problem has less to do with chaos, and more to do with the huge uncertainties involved due to the necessity of aggregating many concepts together (in World3, an example would be the NR stock, which contains all nonrenewable resources – an obvious but somewhat necessary simplification). Also, many social feedbacks (including things that would amplify catastrophe, like war) are simply excluded.

        However, none of that lets you escape the basic insight: if the aggregate human disturbance of the system is growing exponentially, the system’s state is erodable, and there are long delays in perceiving and responding to the problems this creates, overshoot and collapse is a likely behavior mode. The fact that it might be impossible in principle to say exactly when is not reassuring.

    2. Climate is the long-term statistics of weather. Weather is chaotic, but that doesn’t necessarily mean that climate is chaotic on the time scale of interest to us (a century or so).

      Consider a simple chaotic model like the Lorenz model (I think a Vensim version is here in my library). Small differences in initial conditions diverge exponentially, making point prediction impossible after a short time. But one can predict the statistics of the behavior, i.e. the envelope within which it operates, over long time periods.

      For climate, the question then is whether the aggregate energy balance of the earth is sensitive to the location of particular eddies or other turbulent phenomena. Almost certainly not. On time scales that are even longer (ice age cycling), there may be sensitive nonlinear dynamics, though it would appear that we’re pushing the system into a regime where that no longer happens.

      In any case, an argument for chaos in climate is an argument for high gains and nonlinearity, which is the opposite of the usual climate skeptic argument, that climate sensitivity is low. Can’t have it both ways.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.