The Earth Mark II
June 7, 2010 9:36 AM Subscribe
It's been described as one of the most profound scientific initiatives of the 21st century - building an Earth Simulator. Not to figure out icepack melt, calculate global climate patterns, predict earthquakes or determine once and for all 'What do you get when you multiply six by nine'. No, this one's to prevent another kind of meltdown.
FuturIcT, one of the European Union's flagship research initiatives, has been tasked with building the earth simulator. The project will apply complex systems theory, statistical physics and evolutionary game theory to collect and organise data on social, economic and environmental processes at never before seen levels. To make this 'economic Manhattan project' work and still allow for the randomness that is human behavior, they are using Agent-based models.
Agent-based computational models can capture irrational behaviour, represent complex social networks and incorporate emergent patterns of complex systems. They have been used in the past to analyze epidemic spread, biowarfare threats, the growth and decline of ancient civilizations, and the human immune system. Now the EU wants to spend a billion Euros to see how Agent-based models could prevent another financial crisis. A big motivation for this is the feeling among some economists that politicians are 'flying the economy by the seat of their pants.'
FuturIcT, one of the European Union's flagship research initiatives, has been tasked with building the earth simulator. The project will apply complex systems theory, statistical physics and evolutionary game theory to collect and organise data on social, economic and environmental processes at never before seen levels. To make this 'economic Manhattan project' work and still allow for the randomness that is human behavior, they are using Agent-based models.
Agent-based computational models can capture irrational behaviour, represent complex social networks and incorporate emergent patterns of complex systems. They have been used in the past to analyze epidemic spread, biowarfare threats, the growth and decline of ancient civilizations, and the human immune system. Now the EU wants to spend a billion Euros to see how Agent-based models could prevent another financial crisis. A big motivation for this is the feeling among some economists that politicians are 'flying the economy by the seat of their pants.'
[Existing financial models] such as those used by the US Federal Reserve, by necessity stripped away most of the structure of a real economy. There are no banks or derivatives, much less sub-prime mortgages or credit default swaps — these introduce too much nonlinearity and complexity for equilibrium methods to handle. When it comes to setting policy, the predictions of these models aren’t even wrong, they are simply non-existent.There is also a sense that the complex nature of problems faced today "dwarfs the capacity of any individual's comprehension" and that "humankind needs to integrate all its knowledge from all domains into new science able to address the new kinds of problem that emerge in our ever more connected world."
Agent-based models potentially present a way to model the financial economy as a complex system, as Keynes attempted to do, while taking human adaptation and learning into account...
I have an old floppy from Maxis I'd be willing to share...
posted by RobotVoodooPower at 9:46 AM on June 7, 2010 [2 favorites]
posted by RobotVoodooPower at 9:46 AM on June 7, 2010 [2 favorites]
thank you for the wonderful post, Hardcore Poser, there is much to read and ponder here.
I imagine synthesis will be delicious.
posted by infini at 9:47 AM on June 7, 2010
I imagine synthesis will be delicious.
posted by infini at 9:47 AM on June 7, 2010
It's important to note that this is still just an unfunded proposal; there are many other candidate Flagship projects. However, this one seems to have the best publicist.
posted by gene_machine at 9:50 AM on June 7, 2010
posted by gene_machine at 9:50 AM on June 7, 2010
I really wonder about the reliability of such a device. How on earth could it have predicted the financial meltdown when all the ratings agencies were complicit? Garbage in garbage out. At least with climate simulators they have hard scientific data unaffected by a drive for profit.
posted by infinitefloatingbrains at 9:54 AM on June 7, 2010 [2 favorites]
posted by infinitefloatingbrains at 9:54 AM on June 7, 2010 [2 favorites]
And like another Asimov story .. will the computer be able to predict and correct for when the people feeding it data are cooking the books to bias the simulation in their direction?
posted by RobotVoodooPower at 9:54 AM on June 7, 2010
posted by RobotVoodooPower at 9:54 AM on June 7, 2010
Ever run a cybernetwork dedicated to modelling economic systems and managing it? You will, and the EU will bring it to you.
Cybersyn 2.0
posted by symbioid at 10:00 AM on June 7, 2010
Cybersyn 2.0
posted by symbioid at 10:00 AM on June 7, 2010
I approve of this, though I can't help but notice they forgot to carry the two.
Never believe a model, when the truth is staring at you down the barrel of a blaster.
posted by Salvor Hardin at 10:01 AM on June 7, 2010 [1 favorite]
Never believe a model, when the truth is staring at you down the barrel of a blaster.
posted by Salvor Hardin at 10:01 AM on June 7, 2010 [1 favorite]
that's a slippery slope from syn ...cyber or naut
posted by infini at 10:01 AM on June 7, 2010 [1 favorite]
posted by infini at 10:01 AM on June 7, 2010 [1 favorite]
I take a few issues with the FPP and the underlying article.
First of all, FuturIcT is the proposed research center, so it hasn't been 'tasked' with anything. Instead, this is the group of scientists asking the European Union for a billion dollars to design and build this new supercomputer.
To be fair to history, let's not forget that this isn't the first time the term "Earth Simulator" has been used to describe a supercomputer that can, gee, golly, whiz, do everything!
Let's not kid anybody, there are no large-scale agent-based models for modeling financial markets ready to deploy on the next big supercomputer, much less any of our running supercomputers. So this is a proposal not to just build a new supercomputer, but to greatly enhance our current data and software infrastructure for agent-based models. This is very ambitious, as the development of comparatively simpler high performance computational libraries frequently consume dozens of man-years of effort.
The clear winners here are the hardware designers and the power companies (it is expected that the next generation of supercomputers will require as much as 15-20 MW power, or on the order of 20 million dollars a year just to turn on). That doesn't mean that the citizens of the European Union are necessarily going to lose, but the PR fluff that gets tossed around with the unveiling of these machines seems to pile up in direct proportion to the price tag on the machine.
posted by onalark at 10:23 AM on June 7, 2010 [2 favorites]
First of all, FuturIcT is the proposed research center, so it hasn't been 'tasked' with anything. Instead, this is the group of scientists asking the European Union for a billion dollars to design and build this new supercomputer.
To be fair to history, let's not forget that this isn't the first time the term "Earth Simulator" has been used to describe a supercomputer that can, gee, golly, whiz, do everything!
Let's not kid anybody, there are no large-scale agent-based models for modeling financial markets ready to deploy on the next big supercomputer, much less any of our running supercomputers. So this is a proposal not to just build a new supercomputer, but to greatly enhance our current data and software infrastructure for agent-based models. This is very ambitious, as the development of comparatively simpler high performance computational libraries frequently consume dozens of man-years of effort.
The clear winners here are the hardware designers and the power companies (it is expected that the next generation of supercomputers will require as much as 15-20 MW power, or on the order of 20 million dollars a year just to turn on). That doesn't mean that the citizens of the European Union are necessarily going to lose, but the PR fluff that gets tossed around with the unveiling of these machines seems to pile up in direct proportion to the price tag on the machine.
posted by onalark at 10:23 AM on June 7, 2010 [2 favorites]
It's covered -- Civ5 will be out later this year (one hopes.)
posted by Zed at 10:40 AM on June 7, 2010 [1 favorite]
posted by Zed at 10:40 AM on June 7, 2010 [1 favorite]
Wouldn't it be easier to pay EVE Online a bit of grant money to tweak the game to even more successfully model market behavior?
posted by BrotherCaine at 10:46 AM on June 7, 2010 [1 favorite]
posted by BrotherCaine at 10:46 AM on June 7, 2010 [1 favorite]
Ride me, "Blue Seldon"
yes, a When Harry Met Sally reference. suck it up.
posted by davejay at 11:01 AM on June 7, 2010
yes, a When Harry Met Sally reference. suck it up.
posted by davejay at 11:01 AM on June 7, 2010
Not quite the case onalark - they have received funding.
This is indeed very ambitious, and no doubt there is some pie-in-the-sky goal setting going on (see the original proposal; some of the impact goals such as electricity micro-generation seem tangential at best). However, the concept of a multidisciplinary approach to complex systems is not new - the research has been very active for most of this decade (see the Science of Science) and one of the biggest customers is the US military. It's been well-funded, and a number of previous EU flagship initiatives have been in similar veins, so there's existing work to draw from right now.
posted by Hardcore Poser at 11:22 AM on June 7, 2010
This is indeed very ambitious, and no doubt there is some pie-in-the-sky goal setting going on (see the original proposal; some of the impact goals such as electricity micro-generation seem tangential at best). However, the concept of a multidisciplinary approach to complex systems is not new - the research has been very active for most of this decade (see the Science of Science) and one of the biggest customers is the US military. It's been well-funded, and a number of previous EU flagship initiatives have been in similar veins, so there's existing work to draw from right now.
posted by Hardcore Poser at 11:22 AM on June 7, 2010
Harcore Poser, perhaps you know something I don't? Wasn't the proposal submitted in April? I understand that Flagship is funded but that doesn't mean that FuturIcT will be approved.
I stand by my original statement on the lack of software infrastructure. I'll believe there are efficient scalable agent-based modeling simulations when I see them.
posted by onalark at 11:56 AM on June 7, 2010
I stand by my original statement on the lack of software infrastructure. I'll believe there are efficient scalable agent-based modeling simulations when I see them.
posted by onalark at 11:56 AM on June 7, 2010
And when their simulator shows that the Earth is an experiment run by mice, we can all put on our peril sensitive sunglasses and pretend it never happened.
posted by doctor_negative at 12:31 PM on June 7, 2010
posted by doctor_negative at 12:31 PM on June 7, 2010
HardCore Poser, I have been personally involved in the process of validating the Flagship concept. None of the proposals have yet received funding.
posted by gene_machine at 12:40 PM on June 7, 2010
posted by gene_machine at 12:40 PM on June 7, 2010
I have an old floppy from Maxis I'd be willing to share...
SimEarth was the second game I ever bought for my first computer. I still remember that I had a habit of dropping asteroids on India--not because I harbored any hatred for the subcontinent, but mainly because I found the resulting crater aesthetically pleasing.
posted by Mr. Bad Example at 1:03 PM on June 7, 2010
SimEarth was the second game I ever bought for my first computer. I still remember that I had a habit of dropping asteroids on India--not because I harbored any hatred for the subcontinent, but mainly because I found the resulting crater aesthetically pleasing.
posted by Mr. Bad Example at 1:03 PM on June 7, 2010
All I can go on is what the stories tell me (no connection to the project, just interested in it.) This link says it 'has received EUR 1 billion in funding,' as did another link; if that's not the case you'd think at least one of these news sites would retract their statement (naive though that thought may be.)
Setting that aside (and moving on to actual discussion) - how likely is it for them NOT to want something like this to happen, and to succeed? Europe was, and is, on the brink of disaster; any efforts to study our complex modern economy would be welcome, no?
posted by Hardcore Poser at 5:06 PM on June 7, 2010
Setting that aside (and moving on to actual discussion) - how likely is it for them NOT to want something like this to happen, and to succeed? Europe was, and is, on the brink of disaster; any efforts to study our complex modern economy would be welcome, no?
posted by Hardcore Poser at 5:06 PM on June 7, 2010
Unfortunately, I haven't seen any actual news articles on this, just the PR post, rehashes, and the proposal itself. No interviews, no explanation of the program and the funding, etc... If this were truly a 1 billion Euro funded project don't you think it would be getting a definite statement on that from sources like Nature or Wired? Right now, it just looks like the proposal's spin machine has gone a little viral around the Internet. The FPP doesn't do anything besides essentially repeat the spin.
I believe the goals of the project are admirable, but the real challenge in designing useful large-scale simulations comes at the modeling and software implementation, not at the hardware level. These guys don't appear to have much background in deploying these large-scale models (or anything related to high performance computing really), which is why I am casting my doubts.
posted by onalark at 1:17 AM on June 8, 2010 [1 favorite]
I believe the goals of the project are admirable, but the real challenge in designing useful large-scale simulations comes at the modeling and software implementation, not at the hardware level. These guys don't appear to have much background in deploying these large-scale models (or anything related to high performance computing really), which is why I am casting my doubts.
posted by onalark at 1:17 AM on June 8, 2010 [1 favorite]
This kind of thinking has been applied to Ecosystems for decades; glad to see economists waking up to the possibilities.
But remember that the model will always fail in some critical way. in a manner of speaking, the model is nothing, but the process of modelling is everything.
posted by eustatic at 12:02 PM on June 13, 2010
But remember that the model will always fail in some critical way. in a manner of speaking, the model is nothing, but the process of modelling is everything.
posted by eustatic at 12:02 PM on June 13, 2010
« Older And what if Chris Redfield was replaced by some... | “People are not going to befriend you unless there... Newer »
This thread has been archived and is closed to new comments
posted by the painkiller at 9:46 AM on June 7, 2010 [18 favorites]