Collapse - rigour, science and logic

[ccgwebmaster]

It strikes me it should be possible in principle to model/forecast the human world to some extent. I'm not sure about attempting that - probably beyond my abilities - but I think it would be interesting to at least explore how one might go about it.

In this light it would be useful to identify a general approach and any actual scientific papers (as opposed to personal opinions) that could be used to construct a model. It would also be relevant to note which areas information is missing in that would need to be clarified, or fundamental assumptions that must be made to attempt this. I would argue it is necessary to exclude black swan events where single things have a large impact and where we have no means of assessing probability (for instance, nuclear war - unless one can assess the probability of a particular country in particular circumstances going down this route).

As a starting point I think the following especially relevant:

• agriculture
• the likely range of effects of climate change upon it (including but not limited to the Arctic impacts)
• correlation between food supply and demand annually and by implications the prices in relation to a particular level of supply
• the relationship between food prices and conflict
• the identification of the countries likely to be affected and to what extent
• identifying special impacts for specific regions (eg that contain critical resources or overlook major trade routes)

One could include many factors - but my thinking is if one starts simple it isn't necessary to be entirely accurate as long as one treats it like weather forecasting and only attempts short range scenarios.

The goal would be to explore the dynamics of conflict and collapse for a range of agricultural scenarios, in an attempt to gain insight into how rapidly collapse could occur and what levels of stress would be required for it to become a significant risk. Given inability to predict specific events one would necessarily have to treat anything more complex as a statistical picture, by repeatedly running a model (if you know 25% of countries meeting certain criteria will have a civil war, you randomly pick each time the ones that do, for instance).

[ccgwebmaster]

With particular relevance to 4:

The general link between food prices and conflict:
http://necsi.edu/research/social/food_crises.pdf

And with thanks to Anne in https://forum.arctic-sea-ice.net/index.php/topic,192.msg4117.html#msg4117
An apparent start on identifying which countries are likely to experience effects
http://www.imf.org/external/pubs/ft/wp/2011/wp1162.pdf

[ccgwebmaster]

The sort of things I think would be especially relevant would be projections of global demand for food based upon changing population and affluence, versus yield performance in the context of climate change to date - and some means of attempting to predict the price of food (for a range of agricultural scenarios) in that light to feed into the conflict linkage (if anyone has those offhand).

But I'm also curious to see what else people think important to consider in such an endeavour, and what other evidence/research can be brought to bear on it.

[Lucas Durand]

CCG,
What you are proposing sounds a lot like what has already been done with the original "Limits to Growth" model.
The underlying approach used by the LTG model is firmly grounded in systems theory.
Ugo Bardi has plenty of material worth reading on this subject at his blog - I have linked two of his posts below:
http://cassandralegacy.blogspot.ca/2011/05/peak-oil-thermodynamics-and-stoic.html
http://cassandralegacy.blogspot.ca/2011/08/seneca-effect-origins-of-collapse.html

I would argue it is necessary to exclude black swan events where single things have a large impact and where we have no means of assessing probability...

This is exactly the flaw with all models that attempt to predict outcomes from complex systems.
Not such a big deal if the model is just being built "for fun" but a very big deal if risk taking behaviour is going to be justified based on model results.

Complex systems are "opaque" in the sense that non-linear relationships between components of the system create uncertainty which makes the prediction of outcomes impossible (no matter how fancy the math is).
This has profound implications within risk management fields - even if most "risk managers" don't know this, ignore it or pretend it isn't so.
Remember the bankers who not so long ago justified their risk taking with very advanced and detailed risk evaluation models only to blow up the banking system in the end... Ooops, who saw that coming?
When one operates in Black Swan territory, they are ignored at one's peril.

[Laurent]

I would like to see a place like the daily graph on the Neven's blog but with some nice graphs showing different things like the gold price, the debt of USA, Germany, Japon... not all of them but some your choice ! I am sure there is a lot of nice things to display and have a better understanding of what is going on !

There is a site in French (sorry) very interesting : http://lescrises.fr/
That may give you some ideas.

You have got the info on gold on the bloomberg site !
http://www.bloomberg.com/video/boj-is-affecting-gold-prices-blanch-IHgxf1B8TLWp5Gke9HlWUg.html

[ccgwebmaster]

CCG,
What you are proposing sounds a lot like what has already been done with the original "Limits to Growth" model.
The underlying approach used by the LTG model is firmly grounded in systems theory.
Ugo Bardi has plenty of material worth reading on this subject at his blog - I have linked two of his posts below:
http://cassandralegacy.blogspot.ca/2011/05/peak-oil-thermodynamics-and-stoic.html
http://cassandralegacy.blogspot.ca/2011/08/seneca-effect-origins-of-collapse.html

I would argue it is necessary to exclude black swan events where single things have a large impact and where we have no means of assessing probability...

This is exactly the flaw with all models that attempt to predict outcomes from complex systems.
Not such a big deal if the model is just being built "for fun" but a very big deal if risk taking behaviour is going to be justified based on model results.

Complex systems are "opaque" in the sense that non-linear relationships between components of the system create uncertainty which makes the prediction of outcomes impossible (no matter how fancy the math is).
This has profound implications within risk management fields - even if most "risk managers" don't know this, ignore it or pretend it isn't so.
Remember the bankers who not so long ago justified their risk taking with very advanced and detailed risk evaluation models only to blow up the banking system in the end... Ooops, who saw that coming?
When one operates in Black Swan territory, they are ignored at one's peril.

Thanks for the links. I must admit I hadn't previously appreciated that Limits to Growth considered things other than the finite nature of resources. I suppose I would argue that a version where pollution is a limiting factor rather than resources would be more representative of the real world - in terms of those links I'd be arguing pollution both in terms of increasingly extreme weather and in terms of escalating conflict.

Is the model they used "open source" so to speak? Or is there at least a detailed breakdown of the factors they considered to build it?

As for black swans - I think one is forced to ignore them, at least initially, as by definition if they're a black swan you have no framework to predict them. However, I question that they mean a model is useless - mostly I would argue they mean a model may tend to be optimistic and should be taken to represent a more optimistic scenario.

With respect to making such a model for fun or to justify risk taking I would argue it must surely be better to have some model at all to justify what one does, rather than operating from a basis of individual analysis and (to some extent) gut feel?

[ccgwebmaster]

Found this with respect to Limits to Growth:
http://web.ics.purdue.edu/~wggray/Teaching/His300/Illustrations/Limits-to-Growth.pdf\
Didn't notice any reference to including conflict in the model.

Although as far as I see so far the model only makes the most general predictions about the behaviour of the system - doesn't specifically attempt to narrow down the precise behaviour, the most likely decade of collapse, or the precise rate of collapse - as it depends on selections for the input variables?

For a short term model it seems to me one could largely ignore resource limitations, except inasmuch as they may be sourced in some cases only from a specific geographic region. The current availability of resources is therefore of interest but depletion becomes irrelevant unless it's a critical resource likely to run out within a decade.

[Lucas Durand]

Hi CCG,

Is the model they used "open source" so to speak? Or is there at least a detailed breakdown of the factors they considered to build it?

I don't know if the software is open source...
But here are some links to a couple of books that may contain some of the details you're looking for (I haven't read either of them but I know that the updated version of LTG contains information on what "components" of "the world" are built into the model).

http://www.amazon.com/Limits-Growth-Revisited-SpringerBriefs-Analysis/dp/1441994157
http://www.chelseagreen.com/bookstore/item/limitspaper

As far as I understand the model, it is not limited to any particular set of components and can be adjusted.

As for black swans - I think one is forced to ignore them, at least initially, as by definition if they're a black swan you have no framework to predict them. However, I question that they mean a model is useless - mostly I would argue they mean a model may tend to be optimistic and should be taken to represent a more optimistic scenario.

If results are important (ie, "safety" is a concern), I would argue that any model of a complex system should never be used alone as justification for intervention within that system - they can provide a false sense of security and can be a recipe for unintended consequences.
Modelling something complex (ie, something with generally non-linear dynamics) will create predictions that will almost certainly be "out of touch" with empirical reality thanks to the uncertainty involved.
Interventions within complex systems are "naive interventions" in the sense that the actual risks involved with intervention are impossible to know, making "risk management" an inherently incomplete defense against unwanted consequences.

If a model of a complex system is only being used as "food for thought" (sorry, "for fun" was not the best way to put it) then there is really no harm in its use.
However, using a model of a complex system as, say, the basis for a national economic policy is a recipe for disaster - and a failure to realize this is a set-up for Black Swan event at some point (possibly "good" but usualy "bad").

If this sounds like it should be obvious, it apparently isn't as history is littered with examples of people underestimating the complexity of the system that they intend to intervene in and end up causing all manner of unintended consequences as a result.

A good "rule of thumb" ought to be that anyone proposing some kind of intervention in any complex system (whether that be a human body, a banking system, an economy or an ecosystem) should have to bear the burden of proof that the proposed intervention will do no harm - and model results don't count as actual proof ;-)

With respect to making such a model for fun or to justify risk taking I would argue it must surely be better to have some model at all to justify what one does, rather than operating from a basis of individual analysis and (to some extent) gut feel?

Again, I think it depends on what's at stake.

Imagine the following situation:
You are on a ship with other people in a very thick fog.
The people who are piloting the ship want to accelerate into the fog along their proposed route and they are confident despite the fog because they have a "map".
But what about floating logs, other ships and maybe icebergs - hidden risks that do not show up on a "map"?
Would you be ok with this operation or would you like to get off the ship before it heads out?

[Lucas Durand]

Although as far as I see so far the model only makes the most general predictions about the behaviour of the system - doesn't specifically attempt to narrow down the precise behaviour, the most likely decade of collapse, or the precise rate of collapse - as it depends on selections for the input variables?

As I understand it, yes - the LTG model describes how a system (as constructed) responds to any combination of input variables.

For a short term model it seems to me one could largely ignore resource limitations, except inasmuch as they may be sourced in some cases only from a specific geographic region. The current availability of resources is therefore of interest but depletion becomes irrelevant unless it's a critical resource likely to run out within a decade.

I'm not sure I understand...
Could you clarify this?

[ccgwebmaster]

Although as far as I see so far the model only makes the most general predictions about the behaviour of the system - doesn't specifically attempt to narrow down the precise behaviour, the most likely decade of collapse, or the precise rate of collapse - as it depends on selections for the input variables?

As I understand it, yes - the LTG model describes how a system (as constructed) responds to any combination of input variables.

For a short term model it seems to me one could largely ignore resource limitations, except inasmuch as they may be sourced in some cases only from a specific geographic region. The current availability of resources is therefore of interest but depletion becomes irrelevant unless it's a critical resource likely to run out within a decade.

I'm not sure I understand...
Could you clarify this?

The main issue I have with the LTG approach is that they're not trying to make quantitative predictions - only qualitative ones, and even those with so many inputs up for varying produce very highly diverse outcomes (notwithstanding that we seem to have essentially committed to the collapse outcomes, vs the steady state ones theoretically available decades ago with the right choices).

The reason I say about resources - is that I think if you are examining collapse as a near future question - the question of depletion of resources is in most cases a little more gradual (decades), although it remains an important stressor. However - I wonder if resources won't cease to be available in terms of geography and politics rather than depletion, if collapse is triggered principally by other factors (climate change).

I think your point about interfering in a system is highly valid (and that of course is a problem with the financial market models...) - as soon as you model it, even if the model is accurate and valuable - starting to act on what the model can introduce a new element into the system that is necessarily not included in the model (especially if there are multiple actors as in the financial markets). That's true even if you kill all the black swans off.

As for the ship - clearly the precautionary principle applies. I would cut speed, switch on radar, increase look out - ideally look for shallow water to wait it out in (too shallow for big ships, which I am not).

However, the human race appears to have decided to follow an incomplete and out of date map, and to assume that the immense ice bergs ahead are too small or distant to worry about and will magically melt away as we steam towards them at increasing speed...

I think for now my conclusion is I've thought of too many things that I suspect nobody has researched enough to model them well - but I'll use this thread to note any papers etc that might advance things. For example - I'm not aware offhand of research into the dynamics of migration away from conflict regions? One would need some basis to predict the numbers of people that would tend to move around and ideally any instability it introduces into their destination region (rising unemployment and increasing competition for food/etc come to mind).

[Lucas Durand]

As for the ship - clearly the precautionary principle applies. I would cut speed, switch on radar, increase look out - ideally look for shallow water to wait it out in (too shallow for big ships, which I am not).

CCG,
I agree that the precautionary principle applies.

With respect to making such a model for fun or to justify risk taking I would argue it must surely be better to have some model at all to justify what one does, rather than operating from a basis of individual analysis and (to some extent) gut feel?

The precautionary principle is a matter of intuition - you don't need to know exactly what the risks are with accelerating that ship into the fog, only that you feel something nervous in your stomach that is telling you that nothing good can come from what is happening.

Intuition works very well at the human scale where experience is visceral.
A big part of our present predicament is that the unknown risk is generally far outside the human scale where it is not visceral and people feel no need to apply the precautionary principle.
It won't stay that way forever though - eventually that hidden risk will accumulate to the point that it will show up at people's doorsteps.