Are there successors to the “Limits To Growth” World3 model?
The computer model used in the famous 1972 Limits To Growth study was groundbreaking at the time, but very simplistic. For example, it used global averages for everything, such as industrialization or fertility.
I've been looking for similar, more recent, computable models that have lots more detail, but I have come up empty.
Of course, there are very detailed climate models, for example, but none of the detailed models that I have found simulate the entire earth system in a closed loop; as a result, I think they have very limited predictive value. It seems such detailed closed-loop models should be tremendously useful.
Any pointers, anybody?
It turns out that the computer model was pretty accurate. In my mind, it can be compared to the wrong-headed thinking that technology will solve all our ills. Most of the technology that improved our lives dramatically was introduced 100+ years ago. Much of what has been introduced since then has simply been a distraction…
Yes, I saw that. I’m not surprised that reality corresponds to the model, as it was high-quality work, but the model itself — as the authors go through quite some pains explaining in Limits to Growth — makes tons of simplifying assumptions and it was never intended to make any kind of numerical predictions.
I think, however, now we have the data and the modeling experience and the computation that such a numerically accurate model could be made — just as it is being made for climate-only models. And I’m so surprised that it does not seem to exist.
One of the problems is how to curve-fit, without increasing swings away from true.
To some extent, that is solved by doing your modeling with the Parker-Sochacki solution to the Picard iteration, instead of standard numerical modeling.
Irony of ironies, the Parkes-Sochacki solution to the picard iteration is simple enough to do with pencil and paper, and standard math operations (add , subtract, divide, multiply). Despite that, it is sufficient to handle systems of differential equations using those methods.
My father provided a how-to for those interested, with a focus on astronomy; but the PS was specifically developed for studying population dynamics.
Just FYI, the numerical models we got were never good enough to allow us to predict other planets. All that required the PS. And last I knew, when people would ask the US Naval Observatory how to predict planetary positions, they point people to my father’s article.
Before we can curve fit, we need a suitable model IMHO 🙂
Frankly more detailed models are not needed in order to determine that this planet is royaly screwed. For example, I could create an incredibally complex nonlinear finite element model of the entire space shuttle where every single bolt is included in order to determine if one would break. One little parameter tweek could throw the results way off since it’s still an approximation and the cumulative effect of all of these tiny errors adds up pretty quickly. Would it predict one of those heat shield tiles failing off like it did during one of those tragic decents over TX? Apparently not. But will it reach orbit if everything holds together? That’s a simple lumped mass f=ma model where those tiny errors in the complex model are averaged out. The World3 model is a more complex “lumped mass” model. It’s “business as usual” model predicted the global average of what really happened during the last 30 years quite accurately. It doesn’t need to model every single butterfly out there to do so.
I agree with the general picture, but I’d love to have more details than an aggregate model like World3 can provide. Obviously not at the level of butterflies, but there is a long distance between integrating something in the order of 80 variables (like world3) and hundreds of billions of variables at least (in case of the butterfly). Specifically, the kind of model I have in mind could, conceivably, tell inform us about which calamity is likely going to befall whom at which time. World3 specifically disclaims anything related to timing, which for practical purposes (like what people discuss on this forums) is quite important.
For example, it could “predict” which world regions are at risk to become like Syria (drought, migrations, civil war…) and in which timeframe. Or, depending what they model, the quantitative effect of farming runoff on ocean algae, from there on the survival of certain sea creatures, the fishing and tourism industry in Florida, from there to …
I understand modeling of this kind is a difficult business, and also the famous saying: “all models are wrong. some models are useful.”. And I’m flabberghasted that somehow I can’t find any few-thousand to a million-or-so variables that usefully improve on world3.
As calamities might befall from many directions (financial day of reckoning; sudden disruption of oil/food distribution for one political/ecological reason or another ; pandemic outbreaks; civil/global wars…), I doubt anyone would be able to provide accuracy on such a precise level (unless being a psychic or a charlatan); in the meantime, there has been a 30 year update that is worth the read, even though it’s already outdated: http://donellameadows.org/archives/a-synopsis-limits-to-growth-the-30-year-update/
There are just too many variables at play in our complex world.
@NatalieO: The value of a model in my experience is exactly that it allows you to ask questions like the ones you bring up. It’s not that the model produces “one result” which is the definitive prediction of “the future”. As you say, that would be impossible. It is that we can play with what-if scenarios.
A good example would be a model for a bridge: the simulatino model hopefully predicts that the bridge stays up under normal circumstances. But unlike with the real bridge, we can play with our assumptions and see what happens. What if 100 overloaded monster trucks all enter at the same time in the right-most lane? How much will it bend? Does it start swinging? It is still safe? What if it happens during a period of record low air temperature and maintenance overlooked that 10% of the cables are lose? Such a model will, among other things, inform the bridge designer where the design needs to be improved, and the maintainer which things to repair first.
We have the same problem with “the world” as the bridge people with a bridge, except that without a model we are flying blind I’m afraid. Chris has done a marvelous job collecting data; if we had a model we could numerically connect the dots that we otherwise are left to guess. And just as it wasn’t obvious to most people that the “world3” model would predict “collapse” under most circumstances, it will not be obvious to any of us just how exactly “world failure” will occur, and how to prepare, unless it is modeled out.
P.S. Yes, I read the 30 year update.