5 ms·

Higher-level causation exists (but I wish it didn’t)

74 points by darwhy 9y ago

Mathnerd314 9y ago
> Reverse Hollywoodism: The forces of good have every possible advantage, from money to expertise to knowledge to overwhelming numerical superiority. Yet somehow good still fumbles. Somehow a string of improbable coincidences, or a black swan or an orange Hitler, show up at the last moment to let horribleness eke out a last-minute victory, as if the universe itself had been rooting for horribleness all along. That’s our world. This is kind of like Murphy's Law. Let's quote Richard Dawkins and Richard Hand from the Wikipedia article: (https://en.wikipedia.org/wiki/Murphy%27s_law https://en.wikipedia.org/wiki/Murphy%27s_law) > So-called laws like Murphy's law and Sod's law are nonsense because they require inanimate objects to have desires of their own, or else to react according to one's own desires. Dawkins points out that a certain class of events may occur all the time, but are only noticed when they become a nuisance. He gives as an example aircraft noise interfering with filming. Aircraft are in the sky all the time, but are only taken note of when they cause a problem. This is a form of confirmation bias whereby the investigator seeks out evidence to confirm his already formed ideas, but does not look for evidence that contradicts them. > The law of truly large numbers should lead one to expect the kind of events predicted by Murphy's law to occur occasionally. Selection bias will ensure that those ones are remembered and the many times Murphy's law was not true are forgotten. So reverse hollywoodism is really a theory on the limitations of humans in their ability to be scientific: when humans go from the data to the trends (stories), they ignore/distort some of the information to fit the narrative. (See e.g. Trump's election; really not that unpredictable, given data like https://www.washingtonpost.com/news/the-fix/wp/2016/09/23/trump-is-headed-for-a-win-says-professor-whos-predicted-30-years-of-presidential-outcomes-correctly/ https://www.washingtonpost.com/news/the-fix/wp/2016/09/23/tr..., but easy to miss when reading the headlines)
dkarapetyan 9y ago
Scott seems upset.
- ianai 9y ago
  Bad case of the Mondays.
ppod 9y ago
Leaving aside the funny but flippant conclusion, I think this is still a bit harsh on the causal channel capacity argument of the paper discussed. The under-emphasised part is that our brains and bodies do have an artificially imposed uniform distribution over microstates, and that is why macro-level causation is so useful (and often wrong) for us. Of course that might seem obvious, and doesn't mean reductionism is false, and it's not something a reductionist would disagree with, but I think it formalises and makes explicit an assumption that often conceals the root of the disagreement. The channel capacity thing provides an explanation (an excuse) for having to settle for macrostates. The 'normalization trick' is another important reminder that our brains are scale-invariant. https://xkcd.com/915/ https://xkcd.com/915/
- hcs 9y ago
  I don't see why "having to settle for macrostates" is anything more than the observation that we use heuristics. Maybe because I don't understand your statement "our brains and bodies do have an artificially imposed uniform distribution over microstates"? The issue seems more due to the number of states than anything about their initial distribution. Caveat that I have read the post but not the paper.
nemo1618 9y ago
>Maybe, as the physicist Yakir Aharonov has advocated, our universe has not only a special, low-entropy initial state at the Big Bang, but also a "postselected final state," toward which the outcomes of quantum measurements get mysteriously "pulled" This sounds a lot like Terence McKenna's "Transcendental Object at the End of Time". But on the whole... wow, you ever get so upset about a presidential election that you invent a new theory of metaphysics?
- mileszim 9y ago
  The remark on the Aharonov post-selection is weird coming from Aaronson's usual hard adherence to known physics - postselection is relevant in the quantum world only. For Aaronson to twist it into an argument of cosmology seems markedly out of character.
  - yakult 9y ago
    It is, of course, a joke. He's doing a reductio ad absurdem at the end of a tightly argued post about information theory. It's the advantage of doing a blog post rather than a paper, I suppose.
    - mileszim 9y ago
      Good point, I liked the work he's criticizing so I think I'm currently primed to take the post too seriously
albertTJames 9y ago
How can this reach the first page of HN. This is not science, nor it is sound logic.
- zepto 9y ago
  Then you should be trivially able to provide a simple explanation of why, or a reference to something that rebuts it.
  - sidlls 9y ago
    Not necessarily. Some articles of this sort are so skew to reason that they defy refutation by virtue of how far out of bounds they are.
    - Mathnerd314 9y ago
      They're "not even wrong" (http://rationalwiki.org/wiki/Not_even_wrong http://rationalwiki.org/wiki/Not_even_wrong), just a waste of time
      - zepto 9y ago
        Can you say why this article is not even wrong?
    - ue_ 9y ago
      Then you should be able to say in what way the reason was skewed, or why it's invalid, even if it's out of bounds.
- Mathnerd314 9y ago
  https://news.ycombinator.com/newsguidelines.html https://news.ycombinator.com/newsguidelines.html > Please don't submit comments complaining that a submission is inappropriate for the site. If you think a story is spam or off-topic, flag it by clicking on its 'flag' link. If you think a comment is egregious, click on its timestamp to go to its page, then click 'flag' at the top. (Not all users see flag links; there's a small karma threshold.)
ianai 9y ago
I feel like I understood 30% of what he said. Any chance at a quick refresher?
- ignoramous 9y ago
  From what I understand, the blogger says he agrees that scientific reductionism is false, or specifically that details abt microstates of a system (like say, states of quantum objects in countless atoms in a system) don't really have a say in the system's macrostates. The blogger also argues against an experimentation (to prove reductionism false) introduced by the paper he is discussing in the article quite convincingly, I must say. Here's a couple of extracts from the article that should clear the premise up for you: "For here is the argument from the Entropy paper, for the existence of macroscopic causality that’s not reducible to causality in the underlying components." <the argument follows> "...scientific reductionism is false. There is higher-level causation in our universe, and it’s 100% genuine, not just a verbal sleight-of-hand. In particular, there are causal forces that can only be understood in terms of human desires and goals, and not in terms of subatomic particles blindly bouncing around."
  - aoeuasdf1 9y ago
    I'm pretty sure the second part of the blog is tongue-in-cheek.
  - ianai 9y ago
    Thank you for that! In that context, I could maybe see an argument based on chaos theory. i.e. systems being essentially infinitely large compared to their component parts. There could be macro-states that are indistinguishable at micro-scale given any computational ability. But then...you would have to be arguing against the ability to experiment. One way to compute macro-states of a system would be to duplicate that system. Without getting much further into it...I see this becoming a serious rabbit hole in no time.
latently 9y ago
An interesting technicality from the post and paper is that the measure of causal information (mutual information between the initial and final state) bears some resemblence to the Lyapunov exponent as it is used to measure whether a system is on the edge of chaos. When the exponent is 1 (IIRC) the system does not diverge exponentially when the initial conditions are changed slightly and the system is said to be on the edge of chaos and to have good generalization ability. Anywhere else and the system is either damped or chaotic and you don't expect "interesting" stuff to happen there, such as higher-order "causal" effects. (seriously though, why are people so obsessed with causality when it's clear that there is almost never one "causal" description. let it go!)
- yakult 9y ago
  Because sometimes there is, and when you know you can control or at least predict the process. Sometimes this makes you mad dosh.
tominous 9y ago
That took a sudden turn! Normally I just go to Slate Star Codex when I want a mixture of sparkling insight and teleological fiction -- though at least there the fiction is clearly labelled. Back in 2010 I made the prediction that (regardless of the truth) climate change would look increasingly like religion by 2020. Looks like my prediction is right on track.
- yk 9y ago
  I strongly doubt that, even if we get a paper claiming the detection of god tomorrow, for some reasonable definition of god. Then it would take time to plan follow up experiments, include that into theory, work out the theological implications, nail down parameters, etc. You know, all the stuff that climate science did in the 80ies.
dwaltrip 9y ago
This is an atypical post from Scott Aaronson, no? I feel like things kind went off the deep end in the last section there.

74 points by darwhy 9y ago

> Reverse Hollywoodism: The forces of good have every possible advantage, from money to expertise to knowledge to overwhelming numerical superiority. Yet somehow good still fumbles. Somehow a string of improbable coincidences, or a black swan or an orange Hitler, show up at the last moment to let horribleness eke out a last-minute victory, as if the universe itself had been rooting for horribleness all along. That’s our world. This is kind of like Murphy's Law. Let's quote Richard Dawkins and Richard Hand from the Wikipedia article: (https://en.wikipedia.org/wiki/Murphy%27s_law https://en.wikipedia.org/wiki/Murphy%27s_law) > So-called laws like Murphy's law and Sod's law are nonsense because they require inanimate objects to have desires of their own, or else to react according to one's own desires. Dawkins points out that a certain class of events may occur all the time, but are only noticed when they become a nuisance. He gives as an example aircraft noise interfering with filming. Aircraft are in the sky all the time, but are only taken note of when they cause a problem. This is a form of confirmation bias whereby the investigator seeks out evidence to confirm his already formed ideas, but does not look for evidence that contradicts them. > The law of truly large numbers should lead one to expect the kind of events predicted by Murphy's law to occur occasionally. Selection bias will ensure that those ones are remembered and the many times Murphy's law was not true are forgotten. So reverse hollywoodism is really a theory on the limitations of humans in their ability to be scientific: when humans go from the data to the trends (stories), they ignore/distort some of the information to fit the narrative. (See e.g. Trump's election; really not that unpredictable, given data like https://www.washingtonpost.com/news/the-fix/wp/2016/09/23/trump-is-headed-for-a-win-says-professor-whos-predicted-30-years-of-presidential-outcomes-correctly/ https://www.washingtonpost.com/news/the-fix/wp/2016/09/23/tr..., but easy to miss when reading the headlines)

dkarapetyan 9y ago

Scott seems upset.

ianai 9y ago
Bad case of the Mondays.

ppod 9y ago

Leaving aside the funny but flippant conclusion, I think this is still a bit harsh on the causal channel capacity argument of the paper discussed. The under-emphasised part is that our brains and bodies do have an artificially imposed uniform distribution over microstates, and that is why macro-level causation is so useful (and often wrong) for us. Of course that might seem obvious, and doesn't mean reductionism is false, and it's not something a reductionist would disagree with, but I think it formalises and makes explicit an assumption that often conceals the root of the disagreement. The channel capacity thing provides an explanation (an excuse) for having to settle for macrostates. The 'normalization trick' is another important reminder that our brains are scale-invariant. https://xkcd.com/915/ https://xkcd.com/915/

hcs 9y ago
I don't see why "having to settle for macrostates" is anything more than the observation that we use heuristics. Maybe because I don't understand your statement "our brains and bodies do have an artificially imposed uniform distribution over microstates"? The issue seems more due to the number of states than anything about their initial distribution. Caveat that I have read the post but not the paper.

nemo1618 9y ago

>Maybe, as the physicist Yakir Aharonov has advocated, our universe has not only a special, low-entropy initial state at the Big Bang, but also a "postselected final state," toward which the outcomes of quantum measurements get mysteriously "pulled" This sounds a lot like Terence McKenna's "Transcendental Object at the End of Time". But on the whole... wow, you ever get so upset about a presidential election that you invent a new theory of metaphysics?

mileszim 9y ago
The remark on the Aharonov post-selection is weird coming from Aaronson's usual hard adherence to known physics - postselection is relevant in the quantum world only. For Aaronson to twist it into an argument of cosmology seems markedly out of character.
- yakult 9y ago
  It is, of course, a joke. He's doing a reductio ad absurdem at the end of a tightly argued post about information theory. It's the advantage of doing a blog post rather than a paper, I suppose.
  - mileszim 9y ago
    Good point, I liked the work he's criticizing so I think I'm currently primed to take the post too seriously

albertTJames 9y ago

How can this reach the first page of HN. This is not science, nor it is sound logic.

zepto 9y ago
Then you should be trivially able to provide a simple explanation of why, or a reference to something that rebuts it.
- sidlls 9y ago
  Not necessarily. Some articles of this sort are so skew to reason that they defy refutation by virtue of how far out of bounds they are.
  - Mathnerd314 9y ago
    They're "not even wrong" (http://rationalwiki.org/wiki/Not_even_wrong http://rationalwiki.org/wiki/Not_even_wrong), just a waste of time
    - zepto 9y ago
      Can you say why this article is not even wrong?
  - ue_ 9y ago
    Then you should be able to say in what way the reason was skewed, or why it's invalid, even if it's out of bounds.
Mathnerd314 9y ago
https://news.ycombinator.com/newsguidelines.html https://news.ycombinator.com/newsguidelines.html > Please don't submit comments complaining that a submission is inappropriate for the site. If you think a story is spam or off-topic, flag it by clicking on its 'flag' link. If you think a comment is egregious, click on its timestamp to go to its page, then click 'flag' at the top. (Not all users see flag links; there's a small karma threshold.)

ianai 9y ago

I feel like I understood 30% of what he said. Any chance at a quick refresher?

ignoramous 9y ago
From what I understand, the blogger says he agrees that scientific reductionism is false, or specifically that details abt microstates of a system (like say, states of quantum objects in countless atoms in a system) don't really have a say in the system's macrostates. The blogger also argues against an experimentation (to prove reductionism false) introduced by the paper he is discussing in the article quite convincingly, I must say. Here's a couple of extracts from the article that should clear the premise up for you: "For here is the argument from the Entropy paper, for the existence of macroscopic causality that’s not reducible to causality in the underlying components." <the argument follows> "...scientific reductionism is false. There is higher-level causation in our universe, and it’s 100% genuine, not just a verbal sleight-of-hand. In particular, there are causal forces that can only be understood in terms of human desires and goals, and not in terms of subatomic particles blindly bouncing around."
- aoeuasdf1 9y ago
  I'm pretty sure the second part of the blog is tongue-in-cheek.
- ianai 9y ago
  Thank you for that! In that context, I could maybe see an argument based on chaos theory. i.e. systems being essentially infinitely large compared to their component parts. There could be macro-states that are indistinguishable at micro-scale given any computational ability. But then...you would have to be arguing against the ability to experiment. One way to compute macro-states of a system would be to duplicate that system. Without getting much further into it...I see this becoming a serious rabbit hole in no time.

latently 9y ago

An interesting technicality from the post and paper is that the measure of causal information (mutual information between the initial and final state) bears some resemblence to the Lyapunov exponent as it is used to measure whether a system is on the edge of chaos. When the exponent is 1 (IIRC) the system does not diverge exponentially when the initial conditions are changed slightly and the system is said to be on the edge of chaos and to have good generalization ability. Anywhere else and the system is either damped or chaotic and you don't expect "interesting" stuff to happen there, such as higher-order "causal" effects. (seriously though, why are people so obsessed with causality when it's clear that there is almost never one "causal" description. let it go!)

yakult 9y ago
Because sometimes there is, and when you know you can control or at least predict the process. Sometimes this makes you mad dosh.

tominous 9y ago

That took a sudden turn! Normally I just go to Slate Star Codex when I want a mixture of sparkling insight and teleological fiction -- though at least there the fiction is clearly labelled. Back in 2010 I made the prediction that (regardless of the truth) climate change would look increasingly like religion by 2020. Looks like my prediction is right on track.

yk 9y ago
I strongly doubt that, even if we get a paper claiming the detection of god tomorrow, for some reasonable definition of god. Then it would take time to plan follow up experiments, include that into theory, work out the theological implications, nail down parameters, etc. You know, all the stuff that climate science did in the 80ies.

dwaltrip 9y ago

This is an atypical post from Scott Aaronson, no? I feel like things kind went off the deep end in the last section there.