3 ms·

> It’s governed by laws beyond our control as surely as the behavior of materials are governed by thermodynamics. It's funny you make this comparison. Because

> It’s governed by laws beyond our control as surely as the behavior of materials are governed by thermodynamics. It's funny you make this comparison. Because so many of these supposed laws about "the market" violate an effective equivalent to the second law of thermodynamics for information systems: P=/=NP. The claims often made about the transcendent nature of the laws governing "the market" violate the computational properties of reality as we understand it [1]. If "the market" behaved as so many claim then we would have proof of P=NP. Evolution is also an emergent behavior, of biochemical systems. But we also understand how it functions as an algorithm. So much so we now use it to aid in the design the wing shapes of our planes, the design of new high efficiency radios, and to find new algorithms. The entire universe is governed by laws beyond our control. That doesn't prevent us from understanding them and bending them to our will. As we have done with ropes and levers, atoms and proteins, so too can we understand and engineer markets. As it happens markets are likely in a similar class of algorithm as evolution: efficient approximations of NP problems. This is why people use markets as the basis of things like prediction markets, scheduling algorithms, and bin packing. "The market" is a claim to the "god is in the gaps" no different than intelligent design is. It even does the whole song and dance of elevating an approximation algorithm to some unknowable divine will. An approximation algorithm we already understand and use in engineering everyday. [1] Make no mistake, bits of information are physically connected to reality, a non-trivial NP algorithm running on human timescales would boil the oceans. See: https://en.wikipedia.org/wiki/Landauer%27s_principle https://en.wikipedia.org/wiki/Landauer%27s_principle

QuesnayJr 6y ago
This isn't known to be true. In terms of computational complexity, market clearing is PPAD (https://en.wikipedia.org/wiki/PPAD_(complexity) https://en.wikipedia.org/wiki/PPAD_(complexity)). It may turn out that PPAD is computationally tractable even if P != NP, though it is thought to be unlikely.
- SolarNet 6y ago
  I would argue that reinforces my point rather than weaken it. My point being that the market is an understandable thing becomes more reasonable if it turns out it's not even an intractable problem that we are discussing. But I also think you are misunderstanding the argument I was replying to in the first place, because a lot of the claims about "the market" are about the second order allocation of resources that markets cause and not (just) the clearing of them. For example market clearing is how we arrive at the price given the buy and sell orders, but the original goal was the efficient allocation of resources. For which markets are obviously not a perfect solution (it's relatively easy to create order books which end up with unfulfilled orders despite money/weight being available to argue for that allocation), even if they are a good approximate solution (via market clearing). The claims about "the market" are market fundamentalism, that the market is the ideal allocation structure, that we can't find a better or even equivalent solution, because that's just how good markets are. That was what I was rejecting. Not market clearing.
  - samatman 6y ago
    > The claims about "the market" are market fundamentalism, that the market is the ideal allocation structure, that we can't find a better or even equivalent solution, because that's just how good markets are. It's more usual to accept a weaker claim: that markets are the most efficient approximation to the perfect algorithm for price discovery, that is, they are the best way we know of to approximate perfect pricing given the constraint of time. Insisting that no better approach is possible is a strong claim; you'll find it in the wild, but it's always dicey to make confident predictions about the future of human ingenuity. But at the same time, there's no reason to waste much breath on technology which doesn't exist: markets are what we have, and they're what we should use, with the burden of proof firmly on the inventor of anything new to demonstrate that it has greater efficiency.
    - SolarNet 6y ago
      > It's more usual to accept a weaker claim: that markets are the most efficient approximation to the perfect algorithm for price discovery This claim is just as problematic. You may view it as weaker, but algorithmicly speaking it's the same nonsense. Algorithms in similar complexity classes are usually translatable to each other, which is to say they can be framed in terms of each other. You appear to be claiming that markets are the most efficient variant of an algorithmic class. Which is, again, nonsense. It might have some very nice qualities compared to other algorithms within it's class, but to claim it's the most efficient is, again, problematic. A claim like that would require proof, not just the absence of contradicting evidence. > Insisting that no better approach is possible is a strong claim; you'll find it in the wild Ok, but I do believe the poster I was originally disagreeing with is arguing that: > the market is an emergent behavior. It’s governed by laws beyond our control as surely as the behavior of materials are governed by thermodynamics. > If the behavior of your markets fails to match the reality of the market, the best you can hope for is to only introduce a small amount of inefficiency. Of course, utopians who ignore the real nature of the market usually find themselves lethally encumbered by deadweight loss This appears to be a strong claim that nothing can be more efficient than "the market" without introducing inefficiency or "lethal[.]" dead-weight loss. > markets are what we have, and they're what we should use, with the burden of proof firmly on the inventor of anything new to demonstrate that it has greater efficiency. This is a fallacious argument against my point. To begin with, a new algorithm in the same class as markets is likely to have the same or similar efficiency. And it will be the other properties of it that are more interesting. But the real issue is that the double standard people argue with the efficient market hypothesis (and especially the economic calculation problem) is problematic because it misunderstands computation. You simultaneously claim (in this post) to have the best approximation due to a lack of contradicting evidence, but then also want alternatives to provide proof of beating that physically impossible standard. My main response though is: just because markets are the best thing we have does not mean we should accept absurd claims about them. Just because solar energy is effectively unlimited, does not make it a perpetual motion machine.
ChrisLomont 6y ago
>an effective equivalent to the second law of thermodynamics for information systems: P=/=NP. This is not true. For example, the second law is only a statistical claim - the larger the system, the more likely it holds. But the second law is not a law against entropy ever going the other way. It's entirely possible, but not likely that another universe pops out of a quantum fluctuation at any moment. See the concept of a Boltzmann Brain for where physicists think this leads. And P != NP is not a law at all - it's a guess, since no one knows if it's true. Complexity classes certainly changed with the introduction of quantum computer, allowing (very) few problems with known exponential classical Turing time to be done in polynomial quantum Turing time. It is also known that if closed timelike loops are allowed in computation then P=NP, but it's not known if we can engineer such physical items. >The claims often made about the transcendent nature of the laws governing "the market" violate the computational properties of reality as we understand it As evidence of the falsity of the above claims, TQFTs routinely compute NP hard problems, which is why for some time Friedman has tried to leverage them to solve NP hard computational problems. Here's [1] but one paper showing this to be true. "Non-Abelian topological quantum field theories exhibit the mathematical features necessary to support a model capable of solving all ⧣P problems, a computationally intractable class, in polynomial time. " Roughly, TQFTs perform certain "computations" on knots in polynomial time that are known to be NP hard in classical or current quantum computing models. [2] is another nice take in Annals of Mathematics, the most prestigious journal in math. Thus reality routinely solves NP hard problems. Thus your claims are not true. [1] https://www.pnas.org/content/95/1/98 https://www.pnas.org/content/95/1/98 [2] https://annals.math.princeton.edu/wp-content/uploads/annals-v170-n2-p18-p.pdf https://annals.math.princeton.edu/wp-content/uploads/annals-...

QuesnayJr 6y ago

This isn't known to be true. In terms of computational complexity, market clearing is PPAD (https://en.wikipedia.org/wiki/PPAD_(complexity) https://en.wikipedia.org/wiki/PPAD_(complexity)). It may turn out that PPAD is computationally tractable even if P != NP, though it is thought to be unlikely.

SolarNet 6y ago
I would argue that reinforces my point rather than weaken it. My point being that the market is an understandable thing becomes more reasonable if it turns out it's not even an intractable problem that we are discussing. But I also think you are misunderstanding the argument I was replying to in the first place, because a lot of the claims about "the market" are about the second order allocation of resources that markets cause and not (just) the clearing of them. For example market clearing is how we arrive at the price given the buy and sell orders, but the original goal was the efficient allocation of resources. For which markets are obviously not a perfect solution (it's relatively easy to create order books which end up with unfulfilled orders despite money/weight being available to argue for that allocation), even if they are a good approximate solution (via market clearing). The claims about "the market" are market fundamentalism, that the market is the ideal allocation structure, that we can't find a better or even equivalent solution, because that's just how good markets are. That was what I was rejecting. Not market clearing.
- samatman 6y ago
  > The claims about "the market" are market fundamentalism, that the market is the ideal allocation structure, that we can't find a better or even equivalent solution, because that's just how good markets are. It's more usual to accept a weaker claim: that markets are the most efficient approximation to the perfect algorithm for price discovery, that is, they are the best way we know of to approximate perfect pricing given the constraint of time. Insisting that no better approach is possible is a strong claim; you'll find it in the wild, but it's always dicey to make confident predictions about the future of human ingenuity. But at the same time, there's no reason to waste much breath on technology which doesn't exist: markets are what we have, and they're what we should use, with the burden of proof firmly on the inventor of anything new to demonstrate that it has greater efficiency.
  - SolarNet 6y ago
    > It's more usual to accept a weaker claim: that markets are the most efficient approximation to the perfect algorithm for price discovery This claim is just as problematic. You may view it as weaker, but algorithmicly speaking it's the same nonsense. Algorithms in similar complexity classes are usually translatable to each other, which is to say they can be framed in terms of each other. You appear to be claiming that markets are the most efficient variant of an algorithmic class. Which is, again, nonsense. It might have some very nice qualities compared to other algorithms within it's class, but to claim it's the most efficient is, again, problematic. A claim like that would require proof, not just the absence of contradicting evidence. > Insisting that no better approach is possible is a strong claim; you'll find it in the wild Ok, but I do believe the poster I was originally disagreeing with is arguing that: > the market is an emergent behavior. It’s governed by laws beyond our control as surely as the behavior of materials are governed by thermodynamics. > If the behavior of your markets fails to match the reality of the market, the best you can hope for is to only introduce a small amount of inefficiency. Of course, utopians who ignore the real nature of the market usually find themselves lethally encumbered by deadweight loss This appears to be a strong claim that nothing can be more efficient than "the market" without introducing inefficiency or "lethal[.]" dead-weight loss. > markets are what we have, and they're what we should use, with the burden of proof firmly on the inventor of anything new to demonstrate that it has greater efficiency. This is a fallacious argument against my point. To begin with, a new algorithm in the same class as markets is likely to have the same or similar efficiency. And it will be the other properties of it that are more interesting. But the real issue is that the double standard people argue with the efficient market hypothesis (and especially the economic calculation problem) is problematic because it misunderstands computation. You simultaneously claim (in this post) to have the best approximation due to a lack of contradicting evidence, but then also want alternatives to provide proof of beating that physically impossible standard. My main response though is: just because markets are the best thing we have does not mean we should accept absurd claims about them. Just because solar energy is effectively unlimited, does not make it a perpetual motion machine.

ChrisLomont 6y ago

>an effective equivalent to the second law of thermodynamics for information systems: P=/=NP. This is not true. For example, the second law is only a statistical claim - the larger the system, the more likely it holds. But the second law is not a law against entropy ever going the other way. It's entirely possible, but not likely that another universe pops out of a quantum fluctuation at any moment. See the concept of a Boltzmann Brain for where physicists think this leads. And P != NP is not a law at all - it's a guess, since no one knows if it's true. Complexity classes certainly changed with the introduction of quantum computer, allowing (very) few problems with known exponential classical Turing time to be done in polynomial quantum Turing time. It is also known that if closed timelike loops are allowed in computation then P=NP, but it's not known if we can engineer such physical items. >The claims often made about the transcendent nature of the laws governing "the market" violate the computational properties of reality as we understand it As evidence of the falsity of the above claims, TQFTs routinely compute NP hard problems, which is why for some time Friedman has tried to leverage them to solve NP hard computational problems. Here's [1] but one paper showing this to be true. "Non-Abelian topological quantum field theories exhibit the mathematical features necessary to support a model capable of solving all ⧣P problems, a computationally intractable class, in polynomial time. " Roughly, TQFTs perform certain "computations" on knots in polynomial time that are known to be NP hard in classical or current quantum computing models. [2] is another nice take in Annals of Mathematics, the most prestigious journal in math. Thus reality routinely solves NP hard problems. Thus your claims are not true. [1] https://www.pnas.org/content/95/1/98 https://www.pnas.org/content/95/1/98 [2] https://annals.math.princeton.edu/wp-content/uploads/annals-v170-n2-p18-p.pdf https://annals.math.princeton.edu/wp-content/uploads/annals-...