The Oil Drum

We will develop the world energy model as an “open source” project – anyone with Internet access will be able to run the model and view the results of scenario runs.

That's not what "open source" means. Open source would mean that saavy users could get the source and hack it - add in whatever other parameters they wanted. Want to blow this wide open? Do that.

LTG3 kept talking about their model, sounding like it would be available somewhere. Maybe it is now, but it was not when I first got the book. Lost opportunity - not because I'd have done much with it - but someone much smarter than I would have.

Estimates of the linkage strengths between GDP and energy also become increasingly problematic as we move into uncharted energy-supply territory.

GDP won't be helpful. You will have to develop a different benchmark. Maybe a "footprint" or a composite of resources, sinks and throughputs. GDP masks that - turning sinks into benefits (where prices go up) and ignoring resources (clean air and water). If you use GDP and dollars, you might find there is enough in all those hedge funds to pave the Sahara with solar cells. The EROEI of a hedge fund and derivative, global finance - that's going to be interesting.

cfm in Gray, ME

Dear All,

My name is Andrea Bassi and I am one of the modelers working at this project. I will be constantly checking the discussion on this thread and will try to answer all questions. I may not be able to answer every single question or comment on every sentence, my apologies for that. Please fell free to write me an email if needed, though it would be good to keep the discussion on this forum to allow everyone to follow the discussion.

The tool chosen by ASPO-USA consist in an integrated model linking Society, Economy and Environment into one framework. The Millennium Institute has developed the T21 model over the past 24 years, in collaboration in the early phases with various professors and practitioners (e.g. MIT, Dartmouth College, Albany NY, etc.) More info at www.millennium-institute.org

T21 differs from optimization models in many ways, in fact it does not simply aim at optimizing the energy flow to minimize costs but is adds behavioral components (it looks both at the demand and supply side). The structure of the system analyzed is represented within a broad framework and causal relationship and feedback loops are the foundation of the methodology used. For this reason the model does not heavily rely on data and can capture transitions. A preliminary report on T21-USA is available at www.millennium-institute.org with an extensive explanation of the behavior of a few simulation runs.

The model is based on differential equations and it is characterized by nonlinearity, delays and a dynamic behavior created through the accumulation of flows into stocks (e.g. identified reserve -stock- oil recovery -flow- and cumulative production -stock-).
The structure of the model is built on causal relationships. Causality is the foundation of the model and correlation is one of its outputs.
Thanks to this approach, the model aims at representing reality by accounting for feedback loops (both reinforcing and balancing) observable in nature. Simulating new scenarios is also very easy and it takes about 10 seconds to simulate a 70 years period on a regular laptop (the USA model is available on MI’s website, no need to buy ay software to install it and run it). The model is transparent and we can always track changes and justify the behavior of the model.

Just to give you an idea of how the model works, population dynamics are generated by the accumulation of births, deaths, and migration into the stock of population (this stock is disaggregated into 82 age cohorts to calculate fertile women, labor force, social security and medicare beneficiaries, etc). Fertility (and births) is influenced by, among others, income, which together with emissions influences also mortality. Oil price, for instance, is influenced by the availability or oil reserves (both in the short term and in the longer term) in the US and the rest of the world (ROW), by demand/supply balance. Oil price has an impact on the economy, through total factor productivity, and on energy demand for electricity, oil and its substitutes.

T21 start simulating in 1980 and ends in 2050. The structure of the models aims at reproducing historical data until 2006 and then projecting future trends until 2050. Short-term oscillations are not the focus on this modeling exercise, we are looking at medium and longer term implications and project 5 to 10 years trends.

One of the goals of this project is to produce a policy tool able to inform policy making. For this reason a number of policy variables are included in the model and users can test their own assumptions and policies. The outcome of these changes will then be compared to the baseline scenario for the four main spheres of the model: Energy, Society, Economy and the Environment.
Since some of you mentioned games, we have been thinking of creating a game, SimCity-like, where users have to create their own future at the national and global level. The difference between the ASPO-USA model and a game, is basically the interface. SimCity and similar games are based on causality, the same foundation of the methodology used here. Users of T21-North America can simulate different assumptions and policies, we are working at a fully interactive interface (thanks to the ESF team for all the good work done so far!) and hopefully we will be able to work at a real videogame (most probably we will be working at a serious game).
Agent based is certainly an interesting methodology to do so, but it does not allow to understand where the behavior of the model comes from (please let me know if you would like to have a paper explaining the differences and similarities of system dynamics and agent based modeling). In fact, ABM builds on emergence, such an intriguing phenomena. On the other hand, our approach consist in transparency and full replicability, therefore we always want to know what facts generated a change with respect to the base scenario.

As Dick mentioned, we cannot forecast breakthrough in technology improvement of political crises. What we can do is to allow you to simulate such events in every given year of the simulation (2007 to 2050 –past events cannot be changes, but initial conditions such as total oil in place can be modified). Major improvements in oil recovery or energy efficiency (on top of what is simulated endogenously by the model) can be tested. In addition, in the T21-USA report you can find the simulation of the loss of 15% of world oil recoverable reserves (let’s say due a crisis in the Middle East). The consequences shown by the model are many: jump on oil price, higher exploration and recovery (due to the higher profitability of such investments), decline in GDP growth and energy demand, peak oil taking place earlier in time (with respect to the base case), faster transition towards renewable resources (a 5 years delay is considered to account for capacity and infrastructure building) and long term negative economic implications.

As any other model, T21 is a simplified representation of reality and it is not perfect (there is no such thing as a perfect model). This is still a work in progress and a preliminary North America version will be presented in Houston in October. Both the Millennium Institute and the ESF team are working hard to refine and improve the structure of the model and also collecting good data series.

To clarify a point on “open source”: anyone can have access to the source code of the model. In order to change it, you need to have Vensim (which has a very intuitive and user friendly interface) or be a good programmer in C. Dick was erroneously referring to the user version, which allows users to simulate various assumptions and policies without buying any software.

The model calculates carbon footprint and biocapacity. A large range of indicators can be used to “speak the language of different groups”.

John Freise and all, the World 3 model is also available for free for those who have Vensim for Ventana Systems. Let me know if you need more info on this.

Any reference or data that you think may be useful would be very helpful. We want to build an consistent model, able to identify unintended consequences and future paths of today’s choices. In order to do so we need to bring together very different schools of though and make them communicate with a single framework. We believe that T21 is such a tool, and will do our best to reach our goals.

We greatly appreciate your contribution and feedback, please keep posting and please asking questions. This is very helpful and will help us come up with a good model!

Andrea Bassi,

Millennium Institute
www.millennium-institute.org

Mr. Lawrence wrote:

...the results of these modeling efforts are NOT PREDICTIONS, but rather project the consequences of a specific combination of starting conditions and certain events that take place, or decisions that are made, in the course of a scenario run.

Let me paraphrase, and then give my objection. What you (and Dennis Meadows et al) mean to say is roughly:

"We aren't trying to make exact predictions of the future. We only want to show how everything is interrelated, and you can't go on forever at the current rate."

There are a couple problems here. I get it, that you don't want to be tied to specific predictions about specific events at specific dates. You aren't even trying to do that.

But if there is no predictive power, why should anyone pay attention at all?

Mr. Lawrence, you need to face it: these models DO make predictions! They are just very different sorts of predictions from what people are used to.

If, in the 1970s, LTG (Limits To Growth) did not present it's "standard run" showing collapse happening around 2050, then nobody would have ever heard of LTG.

I understand that there were all kinds of distortion in the press of the LTG results. Obviously you want to avoid that. But the best way to do so is to tackle this head-on.

The prediction that the "standard run" makes is this: if we continue with BAU (business as usual) then our future is collapse.

Meadows et al hurry to provide disclaimers. "We did not consider it (the standard run) to be the most probable future, and we certainly didn't present it as a prediction. It was just a place to start, a base for comparison." (p. 168 of LTG- 30 yr update)

I think this is a mistake. I understand the difficulty in communicating what these models do through the blunt instrument of the press. But please, don't water down your results in this way.

Of course, BAU is unlikely. It is likely that some of the measures that LTG describes will take place to some degree. And of course exact prediction is not possible.

I encourage you to find some other ways to say this. Maybe something like this: "The default collapse scenario is in no way a specific prediction about timing. But it is a general prediction, that if we continue BAU then we run up against limits and the consequences are severe. Most importantly, it is a tool for people to use to imagine the likely effects of policy changes."

Meadows et al boil it down to four bullet points following this sentence (p. 177 of LTG-30)

In summary, here are the central assumptions in the World3 model that give it a tendency to overshoot and collapse. If you wish to disagree with our model, our thesis, our book, or our conclusions, these are the points to contest:

(emphasis is theirs). They go on to list the essential features of the model: growth is desired & tends to be exponential; there are limits to sources and sinks; signals are distorted, delayed or denied; limits are erodable when overstressed, and there are strong non-linearities.

If these points could become widely known it would be a good start to having a future.

Best wishes for your project. I eagerly look forward to seeing the results.

Don't try to predict the future. Get ready for it.