15 ms·

When thinking about these kind of problems I really wished we had some good -and publicly available- tools for simulating the world. While the predictions made

When thinking about these kind of problems I really wished we had some good -and publicly available- tools for simulating the world. While the predictions made by such a model would be most certainly wrong in most cases (the system is just too complex to be simulated reliably), it would allow us to narrow down the number of possible outcomes of a given political situation. This would make an interesting business case for a startup, and I can imagine that hedge funds and investors are already using such tools to their advantage, it would be hugely beneficial to have something like that for the general public to use though.

coldtea 10y ago
>When thinking about these kind of problems I really wished we had some good -and publicly available- tools for simulating the world. While the predictions made by such a model would be most certainly wrong in most cases (the system is just too complex to be simulated reliably), it would allow us to narrow down the number of possible outcomes of a given political situation. Actually it wouldn't. The world is more chaos theory than determinism in political scales.
- sago 10y ago
  Such simulations don't assume determinism. A complex dynamic system will have evolutions that follow a probability distribution. The simulation should approximate the p.d., and that is useful information. We do this with weather, market forecasting, and physical modelling.
  - coldtea 10y ago
    >Such simulations don't assume determinism. Chaotic systems (as in chaos theory) don't make predictions either, whatever the probability distribution. Too many factors and butterflies in Beijing. >We do this with weather, market forecasting, and physical modelling Don't know about physical modelling, but in weather and market forecasting we do it with dire results.
    - sago 10y ago
      Unfortunately you're wrong on both counts. The butterfly-in-Beijing metaphor is normally a good sign of someone who misunderstood the dynamics of complex systems. IIRC there's a lecture series from Santa Fe on intro to complex dynamics that might help. And 'dire results': the tyranny of the perfect! We get very good results from both, compared to the intrinsic information of the system. Again you're thinking about single results rather than probability distributions. It's a hard habit to crack, again, an intro to the field would help.
      - coldtea 10y ago
        >The butterfly-in-Beijing metaphor is normally a good sign of someone who misunderstood the dynamics of complex systems. IIRC there's a lecture series from Santa Fe on intro to complex dynamics that might help. I didn't speak of "complex systems" (heck, even a three-body gravity system is complex), I spoke of chaotic systems. >And 'dire results': the tyranny of the perfect! We get very good results from both, compared to the intrinsic information of the system. Again you're thinking about single results rather than probability distributions. No, I'm thinking about the inability to e.g. make any weather forecast for more than 7-10 days at most. >It's a hard habit to crack, again, an intro to the field would help. Again with the condescending study suggestions. It's a bit rich from someone who asks for a pie in the sky "social oracle".
        sago 10y ago
        > Again with the condescending study suggestions. I'm sorry it came over that way. But there should be nothing condescending about pointing out that you are misunderstanding the field. Knowledge and science are a real thing, they don't exist to inconvenience you. Complex systems (including chaos theory) is a fascinating area of math, for which there are a lot of learning resources online. It is not an insult to encourage you to go down the rabbit hole and move beyond the superficial (and wrong) stereotypes. > It's a bit rich from someone who asks for a pie in the sky "social oracle". Different person. Check the GP. Also probably straw-man, but you'd have to ask the GP.
jasode 10y ago
>, it would allow us to narrow down the number of possible outcomes of a given political situation. ... , it would be hugely beneficial to have something like that for the general public to use though. I know it sounds cynical but I don't think such a simulation tool would make any difference to convince the general public. As a thought experiment, imagine if we had a computer simulation in the spirit of Conway's Game of Life[1] or Schelling's Segregation Model[2] to simulate the spread of viruses and breakdown of herd immunity when you have anti-vaccination parents. Showing that "evidence" to anti-vaxxers isn't going to change their minds because they distrust the "government" and "scientists". (E.g. "And who created that simulation?!? The government scientists who want to make our kids autistic!") Same situation with Brexit. The big multinational companies favor the EU but the "little guy" sees his job getting threatened by immigrants. Asking everyday people to be scientific and weigh the "big picture" instead of being emotional about their immediate livelihoods isn't realistic. [1]https://en.wikipedia.org/wiki/Conway%27s_Game_of_Life https://en.wikipedia.org/wiki/Conway%27s_Game_of_Life [2]http://www.natureincode.com/code/various/schelling.html http://www.natureincode.com/code/various/schelling.html
- back_beyond 10y ago
  Questioning the accuracy and objectivity of such a simulation seems perfectly valid.
  - nitrogen 10y ago
    Sure, but doing so without a rational justification is not.
jessriedel 10y ago
We have one something even better and, in particular, more universal, trustworthy, and verifiable: prediction markets. https://en.wikipedia.org/wiki/Prediction_market https://en.wikipedia.org/wiki/Prediction_market Unfortunately, they are generally illegal in the US.
- tobias3 10y ago
  In case of Brexit they didn't work at all. See e.g. http://www.economist.com/blogs/graphicdetail/2016/06/polls-versus-prediction-markets http://www.economist.com/blogs/graphicdetail/2016/06/polls-v...
  - jessriedel 10y ago
    That's silly. Prediction markets still gave a ~15% change to "Leave" at their most pessimistic, and no one says the roulette wheel "didn't work" when black comes up three times in a row. The vote outcome was clearly a surprise to the prediction markets, to newspaper, to the pundits, and to the real markets where people lost billions of dollars. Prediction markets aren't magic, they aggregate views. If they say there's a 15% chance of something happening, by golly, sometimes it will happen. The point is (1) they are self-correcting and (2) they have a much better track record that polls. Not even the "market can stay irrational longer than you can stay solvent" rule applies. With prediction markets, you know exactly when and how the outcome will be settled.
    - TheOtherHobbes 10y ago
      That makes prediction markets non-falsifiable. There's a simple fallacy here - you can't use a market mechanism to define the probability of the outcome of a one-off event like a major referendum. All you have are hopes and fears about the probability, which are always - by definition - limited and subjective. In fact you're simply quantifying opinions about opinions. That's one step removed from what polls do. Polls fail because of practical sampling errors (and sometimes because of political bias) - not because no one understands the theory of polling. There's no way a prediction market mechanism can be self-correcting, because by the time you have real data the question is settled. Some people may find this kind of thing useful. I'm not convinced it's any more than entertainment.
Pamar 10y ago
In case you haven't read it already, I suggest https://www.amazon.com/gp/aw/d/B00KTPM1H6 https://www.amazon.com/gp/aw/d/B00KTPM1H6 For a survey of (human) tools for this kind of analysis.
mathattack 10y ago
This is what economists do. The problem is non-linear relationships and large error margins in variables make it hard to predict. Think the Butterfly Effect on steroids.
askmike 10y ago
A lot of organizations use techniques like Scenario Planning[1], that forces them to already think about a wide scale of possible future scenarios. But in all honesty, it sounds like you are talking about a black swan[2] detector: The current immigration crisis plays a big role in the Brexit. Who could have predicted how western policy about everything in the middle east led to the growth of (the fear of) ISIS? If anyone wear to reliably predict these really big abstract _changes_, they would probably keep it to themselves and use it to make money on the markets. [1]: https://en.wikipedia.org/wiki/Scenario_planning https://en.wikipedia.org/wiki/Scenario_planning [2]: https://en.wikipedia.org/wiki/Black_swan_theory https://en.wikipedia.org/wiki/Black_swan_theory
- iofj 10y ago
  The thing about Black Swans is that they're easy to see, yet very, very hard to time. There is an asset bubble, for instance, at the moment (houses and a lot of other things are more expensive than justified). Gold price seems suppressed. And a government debt bubble. And a China bubble (China is exporting more than is justified by the economics, probably due to state subsidies). There was an oil price bubble, that could reflate (it would take a 5% difference in oil supply, maybe less, for the price to go back to $100 at least temporarily). Renewables could destroy the energy economy (this is happening, could accelerate. E.g. we could find a fantastic new battery principle, easy cheap large scale energy storage. The world would change to accommodate that faster than anyone can imagine). EU disintegration, and related shocks : e.g. a Deutsche Bank bankruptcy. Central banks switching tactics (either to helikopter money or the other way, interest rates back to 4%). More US states/territories cities going bankrupt. Large scale terror attacks. All of these would be large to massive financial shocks. They're also very likely all going to happen. Most of them are going to be caused by another one in that very list, like domino blocks all lined up, but currently standing up. Only one will really be set off in truly unpredictable fashion, and cause the other dominoes to topple. The question is, how to find (and time) the drop that overflows the bucket. That is not a trivial question and the answer could be that said drop won't fall for another decade (or two, why not). That's unlikely, but certainly not impossible. Whilst not many people were expecting the Brexit scenario to pass, ever since the Greek crisis (5-7 years ago now, depending on what event you think set it off) people have been worried about EU disintegration. That the EU, for various reasons, from languages, over border protections to finance habits does not have what it takes to stay together. Apparently they were right. The real fear is that it will cause bank bankruptcies. That may be underway. That may already have happened (a lot of EU banks aren't far from bankruptcy)
rwmj 10y ago
These tools exist. I was part of a team that wrote one [in OCaml!] for the British Home Office, in that case simulating the whole police / courts / prisons / probation system. It's kind of fun to see the effects on prisons of increasing police (more police find more crimes putting pressure on the rest of the system). Whether the predictions are accurate was unclear to me (I was a mere coder), but they are surely better than not doing any simulations at all.
- fulafel 10y ago
  Is this publically available?
  - rwmj 10y ago
    No.

coldtea 10y ago

>When thinking about these kind of problems I really wished we had some good -and publicly available- tools for simulating the world. While the predictions made by such a model would be most certainly wrong in most cases (the system is just too complex to be simulated reliably), it would allow us to narrow down the number of possible outcomes of a given political situation. Actually it wouldn't. The world is more chaos theory than determinism in political scales.

sago 10y ago
Such simulations don't assume determinism. A complex dynamic system will have evolutions that follow a probability distribution. The simulation should approximate the p.d., and that is useful information. We do this with weather, market forecasting, and physical modelling.
- coldtea 10y ago
  >Such simulations don't assume determinism. Chaotic systems (as in chaos theory) don't make predictions either, whatever the probability distribution. Too many factors and butterflies in Beijing. >We do this with weather, market forecasting, and physical modelling Don't know about physical modelling, but in weather and market forecasting we do it with dire results.
  - sago 10y ago
    Unfortunately you're wrong on both counts. The butterfly-in-Beijing metaphor is normally a good sign of someone who misunderstood the dynamics of complex systems. IIRC there's a lecture series from Santa Fe on intro to complex dynamics that might help. And 'dire results': the tyranny of the perfect! We get very good results from both, compared to the intrinsic information of the system. Again you're thinking about single results rather than probability distributions. It's a hard habit to crack, again, an intro to the field would help.
    - coldtea 10y ago
      >The butterfly-in-Beijing metaphor is normally a good sign of someone who misunderstood the dynamics of complex systems. IIRC there's a lecture series from Santa Fe on intro to complex dynamics that might help. I didn't speak of "complex systems" (heck, even a three-body gravity system is complex), I spoke of chaotic systems. >And 'dire results': the tyranny of the perfect! We get very good results from both, compared to the intrinsic information of the system. Again you're thinking about single results rather than probability distributions. No, I'm thinking about the inability to e.g. make any weather forecast for more than 7-10 days at most. >It's a hard habit to crack, again, an intro to the field would help. Again with the condescending study suggestions. It's a bit rich from someone who asks for a pie in the sky "social oracle".
      - sago 10y ago
        > Again with the condescending study suggestions. I'm sorry it came over that way. But there should be nothing condescending about pointing out that you are misunderstanding the field. Knowledge and science are a real thing, they don't exist to inconvenience you. Complex systems (including chaos theory) is a fascinating area of math, for which there are a lot of learning resources online. It is not an insult to encourage you to go down the rabbit hole and move beyond the superficial (and wrong) stereotypes. > It's a bit rich from someone who asks for a pie in the sky "social oracle". Different person. Check the GP. Also probably straw-man, but you'd have to ask the GP.

jasode 10y ago

>, it would allow us to narrow down the number of possible outcomes of a given political situation. ... , it would be hugely beneficial to have something like that for the general public to use though. I know it sounds cynical but I don't think such a simulation tool would make any difference to convince the general public. As a thought experiment, imagine if we had a computer simulation in the spirit of Conway's Game of Life[1] or Schelling's Segregation Model[2] to simulate the spread of viruses and breakdown of herd immunity when you have anti-vaccination parents. Showing that "evidence" to anti-vaxxers isn't going to change their minds because they distrust the "government" and "scientists". (E.g. "And who created that simulation?!? The government scientists who want to make our kids autistic!") Same situation with Brexit. The big multinational companies favor the EU but the "little guy" sees his job getting threatened by immigrants. Asking everyday people to be scientific and weigh the "big picture" instead of being emotional about their immediate livelihoods isn't realistic. [1]https://en.wikipedia.org/wiki/Conway%27s_Game_of_Life https://en.wikipedia.org/wiki/Conway%27s_Game_of_Life [2]http://www.natureincode.com/code/various/schelling.html http://www.natureincode.com/code/various/schelling.html

back_beyond 10y ago
Questioning the accuracy and objectivity of such a simulation seems perfectly valid.
- nitrogen 10y ago
  Sure, but doing so without a rational justification is not.

jessriedel 10y ago

We have one something even better and, in particular, more universal, trustworthy, and verifiable: prediction markets. https://en.wikipedia.org/wiki/Prediction_market https://en.wikipedia.org/wiki/Prediction_market Unfortunately, they are generally illegal in the US.

tobias3 10y ago
In case of Brexit they didn't work at all. See e.g. http://www.economist.com/blogs/graphicdetail/2016/06/polls-versus-prediction-markets http://www.economist.com/blogs/graphicdetail/2016/06/polls-v...
- jessriedel 10y ago
  That's silly. Prediction markets still gave a ~15% change to "Leave" at their most pessimistic, and no one says the roulette wheel "didn't work" when black comes up three times in a row. The vote outcome was clearly a surprise to the prediction markets, to newspaper, to the pundits, and to the real markets where people lost billions of dollars. Prediction markets aren't magic, they aggregate views. If they say there's a 15% chance of something happening, by golly, sometimes it will happen. The point is (1) they are self-correcting and (2) they have a much better track record that polls. Not even the "market can stay irrational longer than you can stay solvent" rule applies. With prediction markets, you know exactly when and how the outcome will be settled.
  - TheOtherHobbes 10y ago
    That makes prediction markets non-falsifiable. There's a simple fallacy here - you can't use a market mechanism to define the probability of the outcome of a one-off event like a major referendum. All you have are hopes and fears about the probability, which are always - by definition - limited and subjective. In fact you're simply quantifying opinions about opinions. That's one step removed from what polls do. Polls fail because of practical sampling errors (and sometimes because of political bias) - not because no one understands the theory of polling. There's no way a prediction market mechanism can be self-correcting, because by the time you have real data the question is settled. Some people may find this kind of thing useful. I'm not convinced it's any more than entertainment.

Pamar 10y ago

In case you haven't read it already, I suggest https://www.amazon.com/gp/aw/d/B00KTPM1H6 https://www.amazon.com/gp/aw/d/B00KTPM1H6 For a survey of (human) tools for this kind of analysis.

mathattack 10y ago

This is what economists do. The problem is non-linear relationships and large error margins in variables make it hard to predict. Think the Butterfly Effect on steroids.

askmike 10y ago

A lot of organizations use techniques like Scenario Planning[1], that forces them to already think about a wide scale of possible future scenarios. But in all honesty, it sounds like you are talking about a black swan[2] detector: The current immigration crisis plays a big role in the Brexit. Who could have predicted how western policy about everything in the middle east led to the growth of (the fear of) ISIS? If anyone wear to reliably predict these really big abstract _changes_, they would probably keep it to themselves and use it to make money on the markets. [1]: https://en.wikipedia.org/wiki/Scenario_planning https://en.wikipedia.org/wiki/Scenario_planning [2]: https://en.wikipedia.org/wiki/Black_swan_theory https://en.wikipedia.org/wiki/Black_swan_theory

iofj 10y ago
The thing about Black Swans is that they're easy to see, yet very, very hard to time. There is an asset bubble, for instance, at the moment (houses and a lot of other things are more expensive than justified). Gold price seems suppressed. And a government debt bubble. And a China bubble (China is exporting more than is justified by the economics, probably due to state subsidies). There was an oil price bubble, that could reflate (it would take a 5% difference in oil supply, maybe less, for the price to go back to $100 at least temporarily). Renewables could destroy the energy economy (this is happening, could accelerate. E.g. we could find a fantastic new battery principle, easy cheap large scale energy storage. The world would change to accommodate that faster than anyone can imagine). EU disintegration, and related shocks : e.g. a Deutsche Bank bankruptcy. Central banks switching tactics (either to helikopter money or the other way, interest rates back to 4%). More US states/territories cities going bankrupt. Large scale terror attacks. All of these would be large to massive financial shocks. They're also very likely all going to happen. Most of them are going to be caused by another one in that very list, like domino blocks all lined up, but currently standing up. Only one will really be set off in truly unpredictable fashion, and cause the other dominoes to topple. The question is, how to find (and time) the drop that overflows the bucket. That is not a trivial question and the answer could be that said drop won't fall for another decade (or two, why not). That's unlikely, but certainly not impossible. Whilst not many people were expecting the Brexit scenario to pass, ever since the Greek crisis (5-7 years ago now, depending on what event you think set it off) people have been worried about EU disintegration. That the EU, for various reasons, from languages, over border protections to finance habits does not have what it takes to stay together. Apparently they were right. The real fear is that it will cause bank bankruptcies. That may be underway. That may already have happened (a lot of EU banks aren't far from bankruptcy)

rwmj 10y ago

These tools exist. I was part of a team that wrote one [in OCaml!] for the British Home Office, in that case simulating the whole police / courts / prisons / probation system. It's kind of fun to see the effects on prisons of increasing police (more police find more crimes putting pressure on the rest of the system). Whether the predictions are accurate was unclear to me (I was a mere coder), but they are surely better than not doing any simulations at all.

fulafel 10y ago
Is this publically available?
- rwmj 10y ago
  No.