8 ms·

I think the main problem I have with this is that it's not internally consistent. He talks about the evolution of a complex system by incremental improvements,

by nknight 15y ago

I think the main problem I have with this is that it's not internally consistent. He talks about the evolution of a complex system by incremental improvements, but apparently maintains the libertarian instinct of "This is broken, throw it out." instead of "This is broken, how do we improve it?". There are an enormous number of governance models that have never been seriously tried on a large scale. There is no fundamental reason to believe that the libertarian value of "shut it all down" is going to produce better results than any of the other possible models.

Jach 15y ago
I don't believe Eliezer has ever said that evolution works by incremental improvements, only by incremental changes that become increasingly complex as a system when there are dependencies. Therefore even if you have a super-complex system it can still suck, the complexity may add nothing to its goodness, incremental changes can be directly harmful, and the right solution may indeed be to throw it out and start over. This should be clear to any programmer. I'll profess historical ignorance here, though from my knowledge I believe that most instances of government change to a better government (and there have of course been changes to a worse government) have come through sudden revolutions that upset much of the previously existing power hierarchy. Additionally, to me it seems like government is naturally against any change at all, which precludes it from being improved easily. Eliezer has written numerous times elsewhere: "Not every change is an improvement, but every improvement is necessarily a change." So I don't think a "scrap the current system" is necessarily a bad idea, though I don't want bloody revolution.
- nknight 15y ago
  But start over with what? I don't interpret libertarianism as "throw it out and start over", but as "throw it out and do nothing". They seek to strip government down and put it in a sealed box of predetermined size that can never change. It should be equally clear to any programmer that boxing yourself in eventually leads to disaster.
  - Jach 15y ago
    I don't seek to argue for/against libertarianism here (neocameralism intrigues me more), I just aimed to show that your initial statements about a complex system improving automatically with changes were false. I also wanted to show that "throw it out" by itself isn't unreasonable. I'm not sure that the programming metaphor here fits, considering languages like Haskell et al. that do box you in have had tremendous success. There are also programs that have been 'running things' just fine for 10+ years unchanged. I'm not aware of any government that could claim that, though with the former I think one of the motivators for having a constitutional government in the first place is to box-in the government. The merits of that vs. having a "Lispy" government that does what it pleases can be argued some other time. (Separately, there is an argument of "premature boxing in" that I have used to argue against full-fledged unit tests early on, but that's getting even further off-topic.)
    - nknight 15y ago
      Actually, the metaphor works fine. Haskell doesn't fundamentally try to restrict what is done, but how it is done. Some things may be harder to do because of those restrictions, but they don't become entirely impossible, even if eventually you might find a corner case where you have to modify the language itself. Constitutions can (usually) be amended. The US Constitution in particular has essentially no restrictions on what can be done by amendment (and it's pretty obvious that the one operative restriction can be dealt with if necessary). This is done out of knowledge of our own ignorance -- we cannot possibly conceive of every situation society will face, nor how our governance models will interact with those situations. This is why libertarianism irritates me. It comes across as such an absolute. Few things are absolutes, and when you're talking about something that will affect millions of people, you can't realistically expect to arrive at an absolutely right solution that will work for everybody every time.
- loup-vaillant 15y ago
  Eliezer may not have been explicit about this but if your genes suck, you likely die before you spawn. Regardless of the complexity of your genome. So, any harmful incremental change would be eventually selected out of the gene pool (or at least won't spread). Neutral changes are unlikely to become universal. Therefore, only improvements can be universal enough to be build upon.
DevX101 15y ago
The inertia associated with implementing a new governmental model on a large scale is VERY high. If your experiment does not work, you may need a war or a revolution to abandon the experiment.
- nknight 15y ago
  Virtually every functional democracy (or reasonable facsimile thereof) has built-in mechanisms for changes, and most provide for quite radical/unbounded changes. Unless you're starting with the premise that governance shouldn't be subject to voting, revolution/war should not happen.
jerf 15y ago
I think you are putting words into his mouth with the "this is broken, throw it out". The reason why unconventional governance models are treated with general skepticism is one of the major points of the essay, which is that if the unconventional governance models are being designed and implemented by the same crew of people with the same basic beliefs that are dominant today, there's no particular reason to believe the outcome is going to be any different in the end. "What makes me a libertarian is that the prospect of having that reconfiguration done by the same system that managed to ban marijuana while allowing tobacco, subsidize ethanol made from corn, and turn the patent system into a form of legalized bludgeoning, makes me want to run screaming into the night until I fall over from lack of oxygen." Also, characterizing the libertarian answer as "shut it all down" is a strawman in its own right. I daresay one of the major ideas that prevents people from understanding libertarianism is understanding that "let the market deal with it" is not anything remotely resembling "shut it all down". The market provides very sharp constraints on what actions one may profitably take, and one of the advantages of the market system is that said constraints are much more closely connected with reality than any other commonly proposed system is. As an also-little-l libertarian I don't deny that some further regulation may be necessary, but defying the market is less like defying people and more like defying physics. You will pay the price. (In fact, we are, right now.) In particular I strongly believe in the need for regulations that ensures that the market stays free, so I'm very in favor of regulations to prevent monopolies from forming, and very hostile to government-granted market monopolies. If you understand the market as "it's all, like, greed man, and greedy people abusing other greedy people, and, like, evil, man", you really ought to spend some time understanding the theory of the market as an actual existing entity, rather than a politically-oriented strawman designed cognitively innoculate you against ideas that may actually change your mind if you were to truly learn about them. Perhaps it won't change your mind, but I'll say that most people online's efforts to pry me away from my libertarianism fails on the grounds that almost nobody actually understands it in the first place, which puts them at a bit of a disadvantage. (Remember, almost nobody is raised libertarian. Few libertarians hold their beliefs unexamined, the way that those who were simply raised liberal or conservative can, though of course not all do.)
- nknight 15y ago
  Well, first, if you want to complain that I or anyone else don't understand libertarianism, I think you should first have a long chat with how your fellow professed libertarians portray it. Amongst other things, you're one of only two or three of the dozens of purported libertarians I've encountered who claim to be in favor of any commercial regulation beyond fraud and contract enforcement issues. The near-universal stance has been that monopolies are a natural byproduct of the free market, and not a "problem". As for the rest... Well, I've never had a problem with the idea that a properly competitive market would be far superior to where we are now, but libertarianism, at least as portrayed by professed libertarians, doesn't restrict itself to that. It seeks to completely tear down social safety nets in favor of that market and ditch environmental and safety regulations. My own experiences and those of my family are all that is necessary to convince me of the disaster that would result by going that far. I'm the last person who's going to defend the ridiculous things cited (marijuana/tobacco, corn ethanol, patents) and many others, but if you're being intellectually honest, you also have to look at the bad things that have resulted from unbounded market forces (serious pollution/poisoning, worker deaths, building collapses, ...) until government stepped in and imposed regulations on the "free market". There are tradeoffs here. If you don't think these things are part of libertarianism, you need a new name for your beliefs, because the existing one is too corrupted by people who either don't actually share your beliefs, or don't understand that they've done a horribly negligent job of explaining them.
_delirium 15y ago
I mostly agree, though I don't have a strong position on the correct answer. But as someone who does study complex systems sometimes in my day job, I find that argument for libertarianism extremely unconvincing. Most complex systems you find do not, left alone, function in any sort of reasonable way. They are full of crazy oscillating attractors, pathological feedback loops, more-stable and less-stable states that you (or an actor in the system) can deliberately manipulate it into, etc., etc. "Well-behaved" dynamical systems are comparatively hard to design even when you're deliberately designing, and ones you just find somewhere are usually not well behaved. Which makes me particularly skeptical of "the market just works" type analysis, which claims that this particular complex dynamical system does not have all those pathologies, and in fact can be analyzed with fairly high-level, aggregate laws like the classical economics analysis of supply/demand adjustment. Heck, even fluid dynamics (another complex system) doesn't work that way--- all our general laws about how pressure/flow/velocity/etc. relate only apply to specific flow regimes, and can have different or even opposite-of-normal relationships in different flow regimes (especially anything turbulent). One argument for why there should be some intervention would be: the classical-economics analysis of how economies should work is an appealing one, but is not in fact how they always work, so the government's job should be to dampen out the pathologies and turbulence that would take us out of the "normal" regime where the classical laws do apply. That's sort of the line taken by ordoliberalism. The question would be whether it can, on the whole, do more good than harm. But I wouldn't start from the assumption that what you get without government intervention operates "well" or "ideally".
- bermanoid 15y ago
  I think your concerns are absolutely spot on, and they match mine almost exactly - I've spent practically my whole working life modeling complex systems, and pathologies are the rule, not the exception, so it strikes me as unbelievably naive to just assume that there's something magical about free markets when in reality they are made up of millions of different entities with different objectives, costs, and constraints. Even when you're simulating a system that's provably stable, when simulation errors and noise creep in, you're batting about 50/50 as to whether your particular simulation method will go off the rails or not (as anyone that's ever tried to simulate a damped harmonic oscillator starting from Newton's laws and the obvious method of approximating a derivative knows quite well). In simulation we take stability arguments and proofs very seriously, the presumption being that if you can't prove that a system is stable, you can't assume anything about it (the exception being physical processes, where stability of the underlying can usually be taken as a given because of anthropic principal arguments and we're merely tasked with finding the proof, as well as a simulation scheme that doesn't screw up - that's a lot more difficult than it sounds, especially for complex systems like fully relativistic gravity). These concerns are of the utmost priority in most fields that involve complex systems, especially processes that we have a hand in influencing, except for economics and politics, where everyone's pet arguments are balls to the wall, and tend to either imply that we should be living with full blown communism or complete unregulated anarchy (okay, I suppose contract law and the military are typically carved out as the exceptions). It's one of the few areas I've ever seen where none of the popular arguments seem to have any sort of balance - and I'm not talking "balance" as in fairness, I'm talking about "balance" as in trivial idea that if both extremes suck (which they usually do, whatever particular idea or disagreement you're arguing over), and something in between is better, then there exists some optimum between the extremes. Step one involves admitting that either extreme has a problem, but instead everyone seems to keep arguing that the whole problem is just that we're not actually at that extreme, 100%.

by nknight 15y ago

Jach 15y ago

I don't believe Eliezer has ever said that evolution works by incremental improvements, only by incremental changes that become increasingly complex as a system when there are dependencies. Therefore even if you have a super-complex system it can still suck, the complexity may add nothing to its goodness, incremental changes can be directly harmful, and the right solution may indeed be to throw it out and start over. This should be clear to any programmer. I'll profess historical ignorance here, though from my knowledge I believe that most instances of government change to a better government (and there have of course been changes to a worse government) have come through sudden revolutions that upset much of the previously existing power hierarchy. Additionally, to me it seems like government is naturally against any change at all, which precludes it from being improved easily. Eliezer has written numerous times elsewhere: "Not every change is an improvement, but every improvement is necessarily a change." So I don't think a "scrap the current system" is necessarily a bad idea, though I don't want bloody revolution.

nknight 15y ago
But start over with what? I don't interpret libertarianism as "throw it out and start over", but as "throw it out and do nothing". They seek to strip government down and put it in a sealed box of predetermined size that can never change. It should be equally clear to any programmer that boxing yourself in eventually leads to disaster.
- Jach 15y ago
  I don't seek to argue for/against libertarianism here (neocameralism intrigues me more), I just aimed to show that your initial statements about a complex system improving automatically with changes were false. I also wanted to show that "throw it out" by itself isn't unreasonable. I'm not sure that the programming metaphor here fits, considering languages like Haskell et al. that do box you in have had tremendous success. There are also programs that have been 'running things' just fine for 10+ years unchanged. I'm not aware of any government that could claim that, though with the former I think one of the motivators for having a constitutional government in the first place is to box-in the government. The merits of that vs. having a "Lispy" government that does what it pleases can be argued some other time. (Separately, there is an argument of "premature boxing in" that I have used to argue against full-fledged unit tests early on, but that's getting even further off-topic.)
  - nknight 15y ago
    Actually, the metaphor works fine. Haskell doesn't fundamentally try to restrict what is done, but how it is done. Some things may be harder to do because of those restrictions, but they don't become entirely impossible, even if eventually you might find a corner case where you have to modify the language itself. Constitutions can (usually) be amended. The US Constitution in particular has essentially no restrictions on what can be done by amendment (and it's pretty obvious that the one operative restriction can be dealt with if necessary). This is done out of knowledge of our own ignorance -- we cannot possibly conceive of every situation society will face, nor how our governance models will interact with those situations. This is why libertarianism irritates me. It comes across as such an absolute. Few things are absolutes, and when you're talking about something that will affect millions of people, you can't realistically expect to arrive at an absolutely right solution that will work for everybody every time.
loup-vaillant 15y ago
Eliezer may not have been explicit about this but if your genes suck, you likely die before you spawn. Regardless of the complexity of your genome. So, any harmful incremental change would be eventually selected out of the gene pool (or at least won't spread). Neutral changes are unlikely to become universal. Therefore, only improvements can be universal enough to be build upon.

DevX101 15y ago

The inertia associated with implementing a new governmental model on a large scale is VERY high. If your experiment does not work, you may need a war or a revolution to abandon the experiment.

nknight 15y ago
Virtually every functional democracy (or reasonable facsimile thereof) has built-in mechanisms for changes, and most provide for quite radical/unbounded changes. Unless you're starting with the premise that governance shouldn't be subject to voting, revolution/war should not happen.

jerf 15y ago

I think you are putting words into his mouth with the "this is broken, throw it out". The reason why unconventional governance models are treated with general skepticism is one of the major points of the essay, which is that if the unconventional governance models are being designed and implemented by the same crew of people with the same basic beliefs that are dominant today, there's no particular reason to believe the outcome is going to be any different in the end. "What makes me a libertarian is that the prospect of having that reconfiguration done by the same system that managed to ban marijuana while allowing tobacco, subsidize ethanol made from corn, and turn the patent system into a form of legalized bludgeoning, makes me want to run screaming into the night until I fall over from lack of oxygen." Also, characterizing the libertarian answer as "shut it all down" is a strawman in its own right. I daresay one of the major ideas that prevents people from understanding libertarianism is understanding that "let the market deal with it" is not anything remotely resembling "shut it all down". The market provides very sharp constraints on what actions one may profitably take, and one of the advantages of the market system is that said constraints are much more closely connected with reality than any other commonly proposed system is. As an also-little-l libertarian I don't deny that some further regulation may be necessary, but defying the market is less like defying people and more like defying physics. You will pay the price. (In fact, we are, right now.) In particular I strongly believe in the need for regulations that ensures that the market stays free, so I'm very in favor of regulations to prevent monopolies from forming, and very hostile to government-granted market monopolies. If you understand the market as "it's all, like, greed man, and greedy people abusing other greedy people, and, like, evil, man", you really ought to spend some time understanding the theory of the market as an actual existing entity, rather than a politically-oriented strawman designed cognitively innoculate you against ideas that may actually change your mind if you were to truly learn about them. Perhaps it won't change your mind, but I'll say that most people online's efforts to pry me away from my libertarianism fails on the grounds that almost nobody actually understands it in the first place, which puts them at a bit of a disadvantage. (Remember, almost nobody is raised libertarian. Few libertarians hold their beliefs unexamined, the way that those who were simply raised liberal or conservative can, though of course not all do.)

nknight 15y ago
Well, first, if you want to complain that I or anyone else don't understand libertarianism, I think you should first have a long chat with how your fellow professed libertarians portray it. Amongst other things, you're one of only two or three of the dozens of purported libertarians I've encountered who claim to be in favor of any commercial regulation beyond fraud and contract enforcement issues. The near-universal stance has been that monopolies are a natural byproduct of the free market, and not a "problem". As for the rest... Well, I've never had a problem with the idea that a properly competitive market would be far superior to where we are now, but libertarianism, at least as portrayed by professed libertarians, doesn't restrict itself to that. It seeks to completely tear down social safety nets in favor of that market and ditch environmental and safety regulations. My own experiences and those of my family are all that is necessary to convince me of the disaster that would result by going that far. I'm the last person who's going to defend the ridiculous things cited (marijuana/tobacco, corn ethanol, patents) and many others, but if you're being intellectually honest, you also have to look at the bad things that have resulted from unbounded market forces (serious pollution/poisoning, worker deaths, building collapses, ...) until government stepped in and imposed regulations on the "free market". There are tradeoffs here. If you don't think these things are part of libertarianism, you need a new name for your beliefs, because the existing one is too corrupted by people who either don't actually share your beliefs, or don't understand that they've done a horribly negligent job of explaining them.

_delirium 15y ago

I mostly agree, though I don't have a strong position on the correct answer. But as someone who does study complex systems sometimes in my day job, I find that argument for libertarianism extremely unconvincing. Most complex systems you find do not, left alone, function in any sort of reasonable way. They are full of crazy oscillating attractors, pathological feedback loops, more-stable and less-stable states that you (or an actor in the system) can deliberately manipulate it into, etc., etc. "Well-behaved" dynamical systems are comparatively hard to design even when you're deliberately designing, and ones you just find somewhere are usually not well behaved. Which makes me particularly skeptical of "the market just works" type analysis, which claims that this particular complex dynamical system does not have all those pathologies, and in fact can be analyzed with fairly high-level, aggregate laws like the classical economics analysis of supply/demand adjustment. Heck, even fluid dynamics (another complex system) doesn't work that way--- all our general laws about how pressure/flow/velocity/etc. relate only apply to specific flow regimes, and can have different or even opposite-of-normal relationships in different flow regimes (especially anything turbulent). One argument for why there should be some intervention would be: the classical-economics analysis of how economies should work is an appealing one, but is not in fact how they always work, so the government's job should be to dampen out the pathologies and turbulence that would take us out of the "normal" regime where the classical laws do apply. That's sort of the line taken by ordoliberalism. The question would be whether it can, on the whole, do more good than harm. But I wouldn't start from the assumption that what you get without government intervention operates "well" or "ideally".

bermanoid 15y ago
I think your concerns are absolutely spot on, and they match mine almost exactly - I've spent practically my whole working life modeling complex systems, and pathologies are the rule, not the exception, so it strikes me as unbelievably naive to just assume that there's something magical about free markets when in reality they are made up of millions of different entities with different objectives, costs, and constraints. Even when you're simulating a system that's provably stable, when simulation errors and noise creep in, you're batting about 50/50 as to whether your particular simulation method will go off the rails or not (as anyone that's ever tried to simulate a damped harmonic oscillator starting from Newton's laws and the obvious method of approximating a derivative knows quite well). In simulation we take stability arguments and proofs very seriously, the presumption being that if you can't prove that a system is stable, you can't assume anything about it (the exception being physical processes, where stability of the underlying can usually be taken as a given because of anthropic principal arguments and we're merely tasked with finding the proof, as well as a simulation scheme that doesn't screw up - that's a lot more difficult than it sounds, especially for complex systems like fully relativistic gravity). These concerns are of the utmost priority in most fields that involve complex systems, especially processes that we have a hand in influencing, except for economics and politics, where everyone's pet arguments are balls to the wall, and tend to either imply that we should be living with full blown communism or complete unregulated anarchy (okay, I suppose contract law and the military are typically carved out as the exceptions). It's one of the few areas I've ever seen where none of the popular arguments seem to have any sort of balance - and I'm not talking "balance" as in fairness, I'm talking about "balance" as in trivial idea that if both extremes suck (which they usually do, whatever particular idea or disagreement you're arguing over), and something in between is better, then there exists some optimum between the extremes. Step one involves admitting that either extreme has a problem, but instead everyone seems to keep arguing that the whole problem is just that we're not actually at that extreme, 100%.