Forgot your password?
typodupeerror
Earth Supercomputing Science

Living Earth Simulator Aims To Simulate Everything 241

Posted by timothy
from the why-didn't-the-mice-think-of-that? dept.
H3xx writes "An international group of scientists is aiming to create a simulator — nicknamed The Living Earth Simulator — that will collect data from billions of sources and use it to replicate everything happening on Earth, from global weather patterns and the spread of diseases to international financial transactions or congestion on highways. The project aims to advance the scientific understanding of what is taking place on the planet, encapsulating the human actions that shape societies and the environmental forces that define the physical world. Perhaps this is Asimov's concept of Psychohistory come to fruition."
This discussion has been archived. No new comments can be posted.

Living Earth Simulator Aims To Simulate Everything

Comments Filter:
  • Everything? (Score:2, Interesting)

    by ferongr (1929434) on Wednesday December 29, 2010 @06:21AM (#34696676)
    Somehow I doubt that all the computing machines in the word combined have the necessary processing power to computationally simulate *everything* that happens on the planet, even when if we try to limit the variables. So I'll just go ahead and assume the science team will compromise on a flawed model which produces equally flawed results.
    • by Chrisq (894406) on Wednesday December 29, 2010 @06:46AM (#34696788)

      Somehow I doubt that all the computing machines in the word combined have the necessary processing power to computationally simulate *everything* that happens on the planet, even when if we try to limit the variables. So I'll just go ahead and assume the science team will compromise on a flawed model which produces equally flawed results.

      The interesting bit comes when the simulation reaches the point that the computer simulation started. The computers then have to simulate themselves running the simulation.

      • by noidentity (188756) on Wednesday December 29, 2010 @06:56AM (#34696834)

        The interesting bit comes when the simulation reaches the point that the computer simulation started. The computers then have to simulate themselves running the simulation.

        And this isn't just an academic point. If their predictions are of any value, they will be incorporated into major decisions made, and thus will be critical for the simulator to predict.

        • by DoofusOfDeath (636671) on Wednesday December 29, 2010 @07:22AM (#34696944)

          But if the system converges to a fixpoint, the infinite recursion can be avoided, right?

          Not that it seems likely to actually convege to a fixpoint. I'm just thinking aloud.

        • by Crypto Gnome (651401) on Wednesday December 29, 2010 @07:50AM (#34697040) Homepage Journal
          Cue Heat-Death of the planet in

          Five
          Four
          Three
          Two
          .
        • by ultranova (717540) on Wednesday December 29, 2010 @08:17AM (#34697164)

          And this isn't just an academic point. If their predictions are of any value, they will be incorporated into major decisions made, and thus will be critical for the simulator to predict.

          Fortunately, the solution is (relatively) simple: when using simulations in decision making, they run what-if -scenarios. The Earth Simulator can simply save the current state in a checkpoint, then run these what-if -scenarios.

          The real problem is predicting the likely parameters used in these simulations. That, and all other parts of the simulator modeling human behaviour, pretty much require either the simulator to be able to simulate human mind(s), or to be interactive.

          • by Chrisq (894406) on Wednesday December 29, 2010 @09:37AM (#34697624)

            And this isn't just an academic point. If their predictions are of any value, they will be incorporated into major decisions made, and thus will be critical for the simulator to predict.

            Fortunately, the solution is (relatively) simple: when using simulations in decision making, they run what-if -scenarios. The Earth Simulator can simply save the current state in a checkpoint, then run these what-if -scenarios.

            Good luck with that. So far attempts to model the effect of computerised trading on the stock market (much simpler than everything) have failed miserably. Also you have the chaos problem, where small changes can cause divergent and difficult to predict behaviour. What would have happened if we had a clear unambiguous warning of the sub-prime crisis a year before it happened? Would it have happened much earlier, been avoided, or just mitigated to an extent? What if the model included a prediction of what effect the warning would have - and so on.

    • by captainpanic (1173915) on Wednesday December 29, 2010 @08:13AM (#34697140)

      Somehow I doubt that all the computing machines in the word combined have the necessary processing power to computationally simulate *everything* that happens on the planet, even when if we try to limit the variables. So I'll just go ahead and assume the science team will compromise on a flawed model which produces equally flawed results.

      Every model is flawed according to that definition.

      They'll try to simplify the earth, and model it... and hopefully it can predict future events with a certain degree of certainty.

      I agree that the word "everything" is too strong... but it's just sad and silly that the entire Slashdot forum attacks these guys because they said this.

      There is some value in this exercise. Just like you can model an ant colony, you can probably model the world. We're all awfully predictable anyway.

    • Re:Everything? (Score:5, Insightful)

      by dbIII (701233) on Wednesday December 29, 2010 @09:34AM (#34697604)
      It entirely depends on what level of detail you want as to whether the model is valid or not. Consider the attached image of myself as an example - it's highly accurate for that level of detail :)
    • by T Murphy (1054674) on Wednesday December 29, 2010 @11:56AM (#34699140) Journal
      A "model of everything" is a very interesting problem. Although I agree their results will be flawed, I disagree with you implication that it's useless research. These guys have no choice but to develop a flawed simulation, but if these guys do good work someone else might come along later and realize they have the means to reduce or eliminate some of those flaws. Making a useful (although still flawed) simulation probably can't be done all at once, so if these guys make the first stepping stone it could prove invaluable for future researchers.
  • by pspahn (1175617) on Wednesday December 29, 2010 @06:21AM (#34696680)
    I'm guessing they end up simply computing Pi.
  • by lul_wat (1623489) on Wednesday December 29, 2010 @06:33AM (#34696714)
    or would it be Third Life? Maybe it'll name itself.
  • by beef623 (998368) * on Wednesday December 29, 2010 @06:36AM (#34696728)
    They could save themselves a lot of work and just get Dwarf Fortress...
  • by azgard (461476) on Wednesday December 29, 2010 @06:37AM (#34696736)

    I am not sure if having everything in a single model is a good approach. Certainly seems too computationally expensive.

    I would much prefer if people start looking into socio-economic modeling. We now have pretty good climatic models, and this started in the 50s, when there computers had almost no power. Now we have such power, and we could feasibly simulate social (i and economic behavior of whole country on a home PC.

    We could start like this: Each person in simulation would have possible actions (these would be predefined), and decide on which action to use based on habit, previous experience and copying from friends of its social network. Actions could range from economic (produce something of value) through social (meet someone) to political (enter contract with someone).

  • by santax (1541065) on Wednesday December 29, 2010 @06:38AM (#34696742)
    Just buy a copy of sims 3 and be done with it.
  • I hope... (Score:2, Funny)

    by Mr Z (6791) on Wednesday December 29, 2010 @06:38AM (#34696750) Homepage Journal
    I hope they've put some deep thought into this....
  • by Parker Lewis (999165) on Wednesday December 29, 2010 @06:45AM (#34696778)
    Name it Matrix.
  • by cstacy (534252) on Wednesday December 29, 2010 @07:06AM (#34696882)

    Yes but is the simulator written in BASIC?

  • by DoofusOfDeath (636671) on Wednesday December 29, 2010 @07:18AM (#34696924)

    Weather is a chaotic system, and weather affects living things in very significant ways. I'm sure there are plenty of other chaotic non-linearities in what they're trying to simulate as well.

    Doesn't such instability doom any world simulator to crappy fidelity?

    • by aminorex (141494) on Wednesday December 29, 2010 @12:41PM (#34699710) Homepage Journal

      Weather simulations are very useful. Doesn't that doom this effort to producing a very useful result?

      • by DoofusOfDeath (636671) on Wednesday December 29, 2010 @02:27PM (#34701294)

        Weather simulations are very useful. Doesn't that doom this effort to producing a very useful result?

        Not necessarily. My basic concern is that weather is chaotic, and probably something dependent upon the weather has its own intrinisic chaotic properties even if the speicied weather were accurately forecast. So I'm thinking that the overall simulated system would have a composite degree of chaos much greater than that of just the global weather system.

  • by Crypto Gnome (651401) on Wednesday December 29, 2010 @07:39AM (#34696990) Homepage Journal
    Clearly this is an attempt to invoke Laplace's Demon [wikipedia.org].
    • by DigiShaman (671371) on Wednesday December 29, 2010 @11:21AM (#34698606) Homepage

      I have personal (philosophical) reservations about this concept, for it negates the concept of "free will". But assuming such a simulator can provide a reliable rough estimate of our universe in both past, present, and future. Well, get ready for Precog justice.

      Imagine for a moment. You're life was already fucked the moment after the creation of the universe some 13.75 billion years later.

      • by owlstead (636356) on Wednesday December 29, 2010 @02:41PM (#34701484)

        "I have personal (philosophical) reservations about this concept, for it negates the concept of "free will".

        In other words, because I don't want it to be like that, it can't be like that.

        Personally, I've always thought that the idea of "free will" is in the ropes. Even if there is no such thing as predictability, how can we actually change the physical processes that happen in our brain, if we ourselves are a product of them?

        That does not actually mean that we should change because of that, since maybe the *idea* that there is free will may be enough. We have invented all these morals and laws around them, and our brain certainly takes that data into account.

        • by DigiShaman (671371) on Wednesday December 29, 2010 @03:13PM (#34701974) Homepage

          In other words, because I don't want it to be like that, it can't be like that.

          In other words, you didn't get my meaning. What I meant was that I find the concept not proof, to be disturbing. Being that Laplace's Demon is a thought experiment without absolute proof doesn't reflect what you thought I said.

          So no, I'm not sticking my head in the sand here pal. I'm quite well aware of that fact.

  • by Crypto Gnome (651401) on Wednesday December 29, 2010 @07:41AM (#34697002) Homepage Journal
    Any sufficiently accurate simulation is indistinguishable from reality. Corollary: Any simulation distinguishable from reality is insufficiently accurate. Phil P
    • by Burnhard (1031106) on Wednesday December 29, 2010 @09:42AM (#34697662)
      Indeed this is quite right. I'm reminded of something I read recently: "according to the laws of physics, a feather and a cannon ball should fall with the same velocity. In reality, my feather was blown into a tree",

      A certain section of scientists seem to lack the necessary insight into what they're actually doing when they develop a model on a computer. It's a conceptual representation, not a realistic simulation and it can only model those concepts that are well understood.

      If economic models fail (demonstrably) and weather forecasting models fail (again, demonstrably on the short, medium and long-term), then why do these people think that combining them isn't going to be a fail an order of magnitude greater? Joseph Wiezenbaum wrote a great book in the 1970's called Computer Power and Human Reason. I think the people behind this should read it.
      • by Paul Fernhout (109597) on Wednesday December 29, 2010 @11:52AM (#34699090) Homepage

        I agree with you in general about the limits of simulations and even intelligence itself.

        Still, simulations can be used to:
        * predict (you are right, they often fail for reasons of chaos theory and limited accuracy or missing aspects);
        * understand (where you play what ifs to see the consequences of your assumptions);
        * to gain insight (something other than understanding of details, where you gain a sense of the gestalt, a feeling, or some new summarizing key idea, like I say with my sig about the irony of the tools of abundance in the hands of those thinking in terms of scarcity -- maybe we need a simulation about creating simulations to have scientists gain the insight about simulations you suggest many lack? :-);
        * assess risk (to some extend, by Monte Carlo methods for well understood processes)
        * to consolidate knowledge in an organized explicit way (you can't hand wave as much when you have to implement ideas in code);
        * educate intellectually (as fun toys to play with and learn from);
        * educate practically (to learn skills by trial and error, basically failing faster and safer like in a flight simulator or nuclear power plant simulator or surgical simulator);
        * educate emotionally (to see consequences and possibilities and related narrative, often as games);
        * entertain (relates to the above, but is a different focus);
        * to serve as a focus for political policy debates about future scenarios (including as different simulators with different assumptions describe different implications of policy -- note weather forecasters use multiple weather models plus their intuition and experience to make forecasts);
        * as a form of self-justifying artwork;
        * as a way to create entirely new worlds to explore inspired by nature but (as you suggest) often very different;
        * probably many more -- in the sense of, what good is a blank sheet of paper?

        I learned some of this from thinking about what people like Steven C Bankes at RAND had to say in the 1980s and 1990s:
        http://www.rand.org/pubs/authors/b/bankes_steven_c.html [rand.org]
        As well as people like Seymour Papert (of Microworlds educational software fame).
        http://www.papert.org/ [papert.org]
        Or Alan Kay and Dan Ingalls and others with Smalltalk as a simulation environment. As well as what futurists (WFS) and risk modellers (RAMAS) have to say. And from making a simulation about gardening in the 1990s (with my wife, as a more than six person-year labor of love released with source under the GPL):
        http://www.gardenwithinsight.com/ [gardenwithinsight.com]

        One concern I have about simulations of living creatures (especially intelligent or self-aware ones that can feel some kind of virtual pleasure or virtual pain, like in agent-based simulations) is, what are the ethics? As in, do not do unto others that which you do not want done unto yourself (unless they like that kind of stuff)...
        http://www.simulation-argument.com/ [simulation-argument.com]
        http://www.rfreitas.com/Astro/LegalRightsOfRobots.htm [rfreitas.com]

        See also:
        http://en.wikipedia.org/wiki/Simulation [wikipedia.org]

        • by Burnhard (1031106) on Wednesday December 29, 2010 @03:29PM (#34702176)
          Simulations/models can be useful as an aid to understanding, but in reality what happens is model outputs are given the term "prediction" in press-releases. When economists do it, we all raise an eyebrow. When scientists do it, for some reason people think it's come from the Oracle at Delphi.

          In my view this story is a good example of the arrogance and stupidity embedded in our scientific establishment. Moreover, I have no doubt that the control climate models have given activist scientists over the political process, despite the fact they're unquestionable wrong, makes me wonder whether this project is being constructed for the same Machiavellian reasons. Once complete, expect to see shocking predictions that demand immediate real-world action to avert disaster.

          btw: your concern about simulations of living creatures are unfounded, given that qualities like pleasure or pain are more than likely non-computable. That is to say, to instantiate the given phenomenological property, it is not sufficient to instantiate the given functional property ("I have a variable called Pain, therefore my virtual creature feels pain"). Qualia require conscious experience, whereas a cognitive model describes only the functional correlates of consciousness, not the conscious experience itself.
  • by Idou (572394) on Wednesday December 29, 2010 @08:47AM (#34697326) Journal
    How do we know we are not already living in one of these things? If so, this seems a bit redundant . . .
  • by andersh (229403) on Wednesday December 29, 2010 @08:54AM (#34697358)

    The simulation is a European project, part of the FuturICT-programme, a part of the European Union research framework programme [europa.eu].

    It intends to unify hundreds of the best scientists in Europe in a 10 year 1 billion EUR program to explore social life on earth and everything it relates to. The FuturICT flagship will produce historic breakthroughs and provide powerful new ways to manage challenges that make the modern world so difficult to predict, including the financial crisis.

    The FuturICT Knowledge Accelerator is a previously unseen multidisciplinary international scientific endeavour with focus on techno-socio-economic-environmental systems. The three main achievements of the FuturICT flagship will be the establishment of
    - a Living Earth Simulator (global-scale simulation of techno-socio-economic systems),
    - Crisis Observatories (for financial instabilities, scarcity of resources, emerging risks and conflicts, epidemics, etc.), and
    - an Innovation Accelerator (identifying innovations early on, evaluating them across disciplines and supporting co-creation projects between different scientific disciplines, business, and governance).

    • by khallow (566160) on Wednesday December 29, 2010 @11:18AM (#34698562)

      The FuturICT flagship will produce historic breakthroughs and provide powerful new ways to manage challenges that make the modern world so difficult to predict, including the financial crisis.

      The only thing difficult about the financial crisis was predicting when it would occur. That it would occur was obvious from the levels of risk that the financial and real estate institutions were taking on.

      • by owlstead (636356) on Wednesday December 29, 2010 @02:47PM (#34701602)

        "The only thing difficult about the financial crisis was predicting when it would occur. That it would occur was obvious from the levels of risk that the financial and real estate institutions were taking on."

        The main problem with the financial crisis is getting things to change while so many people have short term, highly personal reasons to let it exists. This is such a big problem that most people seem to address this issue by putting their hands on their ears and shout "LALALALALA" on top of their voice.

  • by Nyder (754090) on Wednesday December 29, 2010 @09:11AM (#34697472) Journal

    Are our super computers even capable of doing the calculations for, well, everything on earth?

    I'm going to say, no.

    You'd have to do too many shortcuts to get accurate results, let alone we don't know how most the crap works in this world anyways. Sure, we know some, and learn more, but enough to simulate it?

    I'm going to put this up there with Duke Nuke'm coming out before 2010 is over.

  • by argStyopa (232550) on Wednesday December 29, 2010 @10:08AM (#34697868) Journal

    Starting from such farcical first assumptions, its hard to believe this stupid idea is going to get funding, much less even conceivably work.

    Look: every "sim" rationalizes inputs, creates a facade of activity that looks like real life, so as to produce reasonably realistic outputs. The point of modeling is to EXCLUDE as many variables as you can, to make what you're studying as simple as possible, to try to draw it down to its essentials.

    To build a true sim, you have to start at the "lowest protocols" that input your system. Given that they want to sim the entire world, you're talking about starting with an engine that comprehensively models the physics and chemical processes that create the earth. (And that's assuming that you can safely take solar input as a given, and assume that your 'isolated' environment is the Earth, not entirely a safe assumption when you start considering the impact on solar variability to Earth weather, lunar gravity on tides, etc.) As I understand, we *don't* even have a good understanding of those very basic systems such as what actually is at the core of our planet, or the intricacies of plate tectonics, etc. (Note: once you can do this, I'm sure all those scientists already working their whole lives to model the 'simple' basics of planetary physics and volcanoes would love a copy of the sim to that point.) So we'd have to rationalize the up-feed in the sim from that lowest protocol - already, at the very basic levels, we're essentially guessing at inputs, not simming them. Whups.

    'Above' geophysical processes, there's the physical processes of water, weather, etc. that produce the topology of the surface of the world (or do you think that geography, climate, weather, etc. don't play a part in individual or collective choices?); only once you have that simmed properly can you then even START to model the biosphere. Once you model the biosphere comprehensively, you can start to model people in it, and then you're faced with the comprehensive issues of psychology, free will, collective behaviors, human perversity. Finally, once you have this robust giant model, then you can finally start simming the meta systems atop human society - economics, social welfare, etc.

    This is not Asimov, and this goofball is no Hari Seldon. He's talking about the 'critical importance' of simulating (and getting useful results from) the omega step in this project, when we can't even manage the alpha. That's farce. One might as well start a project analyzing the economics of interstellar trade, because the results are probably going to be useful long before this project produces anything of value. (But I'm pretty certain it'll be spending piles of $$ well before that, right?)

    "Regardless of the challenges the project faces, the greater danger is not attempting to use the computer tools we have now - and will have in future - to improve our understanding of global socio-economic trends, says Dr Helbing." ...Should read instead as "Regardless of the challenges the project faces, the greater danger is not attempting to use the computer tools we have now - and will have in future - to try to gather some funding for an open ended and unachievable research project," says Dr Helbing, "This nonsense could keep me in BMW's, first-class airline seats, and symposia in cool locations for the rest of my life."

  • by BigSlowTarget (325940) on Wednesday December 29, 2010 @10:09AM (#34697870) Journal

    So does Civ V. So does my globe. Anything can be used to simulate anything else poorly. Nothing can be used to simulate everything else exactly. The quality of the output will be determined by the areas they focus, the choices made when they write the algorithms and the data they get to put into the system. 'We're simulating the whole world and how it interacts' provides little information about the quality or usefulness of the simulation.

  • by yotto (590067) on Wednesday December 29, 2010 @10:28AM (#34698048) Homepage

    Not even close.
    Psychohistory had nothing to do with simulating things on a computer. It was a mathematical construct to predict human behavior in groups, in the same way that you can predict what a large enough group of atoms will do even if you can't know what any one atom will be doing at any given time.

    This is more like a Matrix 0.001 alpha. I imagine their Neo being a little piece of CoreWars code.

  • by Joe The Dragon (967727) on Wednesday December 29, 2010 @10:29AM (#34698064)

    I want a simcity like game with a custom road system Simcity 4 + plugs ins is nice but old and buggy.

    Citys xl lacked the custom road systems and the highway interchanges are to big and just have roundabouts.

  • by gearloos (816828) on Wednesday December 29, 2010 @11:03AM (#34698368)
    Yes Mr. Smith, it started in 2010.
  • by kimvette (919543) on Wednesday December 29, 2010 @11:28AM (#34698712) Homepage Journal

    In a related news story, a group of scientists have formed a company named Magrathea with long term plans of building a massive planet which will be capable of fostering the design and manufacture of custom luxury planets.

  • Uncertainty Reigns (Score:4, Insightful)

    by thethibs (882667) on Wednesday December 29, 2010 @11:38AM (#34698858) Homepage

    The big problem with this is that most of the world's and humanity's interesting systems are chaotic. You may get lucky and find an attractor or two.

    In any case, simulation can show you plausible futures, but they'll have no predictive value. The outputs will be little more than cybernetic speculative fiction.

    On the other hand, there's no explaining chaos to a politician, or to a scientist who believes that more data and higher resolution are all that's needed to clear up the confusion; the grant money will keep flowing.

  • by PPH (736903) on Wednesday December 29, 2010 @11:46AM (#34698990)

    ... we are nothing more than a simulation created by some super alien CS student as a senior project this should be interesting. It also has implications as to whether our universe will die slowly due to entropy buildup (a memory leak in the simulation) or in some cataclysmic, world ending event (when the simulation is ported to VB.Net and crashes).

  • by houghi (78078) on Wednesday December 29, 2010 @11:47AM (#34699022)

    Does the Living Earth Simulator contain the Living Earth Simulator? And then does that one as well, like a Russian Mandelbrot doll.

  • by houbou (1097327) on Wednesday December 29, 2010 @03:03PM (#34701842) Journal
    if we are already living in a simulation, I hope the one being built from this collective of intelligence, will correct the flaws :)
  • by Nom du Keyboard (633989) on Wednesday December 29, 2010 @03:40PM (#34702332)
    Can it lead me to a major lottery win? If not, how useful is it to me?

No amount of careful planning will ever replace dumb luck.

Working...