I’m a second-year mathematician, and I’ve been looking at this for my one non-math module. Here are some comments I can make:
Jeremy England in a 2015 paper (“Dissipative adaptation in driven self-assembly”): > At this point, dissipative adaptation should seem like too simple an idea to be true, not least because the reality is more complicated (even in principle) than the account of things we have managed to give within the confines of this Perspective. For one thing, our dis- cussion has trod rather lightly over the (addressable) issue of rare trajectories that dissipate much less than average, which are known to come to the fore in a confounding way when we try to compute exponential averages like in equation (4). Even more significantly, we must acknowledge that it is not always the case that a history of absorbing and dissipating work from a drive corresponds to a con- tinuing ability to do so in the present moment; sometimes the likeli- est outcome of a non-equilibrium process is a pile of broken shards rather than a resonantly vibrating goblet, which means we cannot generally assume that dissipation rates must always increase over time — indeed, in the case of the near-equilibrium linear-response regime, dissipative adaptation reduces to the requirement that there be extra entropy production during relaxation to the steady state, so that Prigogine’s condition of minimum entropy production holds at the end of the process.
I think he is aware of your “the devil is in the details” analysis which you gave in your “Earth” subsection. In all the papers of his I skimmed, he is aware that the reversal probability is very difficult to calculate.
What I think happened (epistemic status: low; my current best explanation based of some skimmed papers and some media response) is that he was very excited to have potentially explained life in 2014, and went on to make some “controversial” and dubious claims to journalists. While he did not push that narrative any further (the last journal article to reference England’s “theory of life” was in about 2017), there are a lot of people out there who have only engaged with those 2014-2017 articles and the attached hype of explaining life.
(In fact, I only conviced myself that this theory has very limited predictive power for the emergence of life today. What convinced me is that the “reversal probability” term can easily dominate the “dissipative adaptation” term; moreover, it is infeasible to estimate the reversal probability in a complicated system, due to the difficulty of averaging exponentials compounding the difficulty of modelling a high-entropy system, and hence it is infeasible to estimate whether the reversal probability term dominates the dissipative adaptation term in all but the simplest toy models. I can’t remember if you said this in this article, or England said this somewhere, but we’re back to calculating the probability of life emerging by calculating all the chemistry by hand.)
Interestingly, England again returns to dubious theories on life in his 2020 book, Every Life is On Fire. My guess (again, epistemic status: low) is that he actually started writing that book in 2014, and went through development hell until the publishing date; eventually, the book was published, but it never got to it’s central thesis, that dissipative adaptation explains life, because England had already rejected that thesis. He achieved his purpose on talking about the relation of the bible to his work, which I won’t say anything on because I find it irrelevant, but it does explain his motivation for publishing the book. The only chapter that really got into talking about dissipative adaptation was chapter 7 (out of eight); there were some very interesting things likening some results in non-equilibrium statistical physics to machine learning. Among them, [solving non-linear differential equations by letting a synthetic octopus arm flop about]. (https://www.frontiersin.org/journals/computational-neuroscience/articles/10.3389/fncom.2013.00091/full).
The conclusions about life I did get from the book:
When life does develop (as opposed to non-life), it will be an efficient dissipator of the local source of low entropy. Coming back to the plant example, plants absorb low-entropy light more efficiently than barren ground, which reflects a lot of it back into space in it’s low-entropy form (as opposed to the high-entropy black body radiation). In a sense, this is how dissipative adaptation can explain darwinian evolution? I’m not versed in evolutionary mathematics, so I’m not comfortable doing the math to verify this. Equally, I’m not entirely certain that my plant example isn’t (in some sense) a coincidence, as I can’t verify the maths.
Dissipative adaptation turns up in smaller examples of life utilising it over others. For example, as alluded to in the linked paper above, octopus propulsion is not fully controlled from the brain, but also relies on the interaction with the dynamics of water. I also think some good examples can be found in plants that respond to stimuli. How does a sunflower turn towards the sun? How does a wind-sculpted tree grow into that shape? The answer is likely probabilistic, and are examples of living things using their “environment for computation”.
One last thing: I misunderstood the point you were making when you were talking about blackholes. The point you were making was ‘”What maximizes entropy” is a bad morality’; what I thought I was reading was ‘dissipative adaptation does not work because it predicts that we will into a black hole and Earth developed complex life because the complex life did some nuclear fission after it was developed’. So, it seems that the only point I disagree on is that I think that a tree is in fact a more efficient dissipator than no tree; also, using a seed growing as an example of life to showcase dissipative adaptation is not a good example on your end, because you appeal to the intuition that a seed dying is more likely to invert than a seed growing — “tree-space” is large, and a seed is very robust against dying spontaneously; hence I disagree that your claim is intuitive, and evidently wrong, since a good seed is overwhelmingly likely to grow in any good soil. Plants CAN be good examples of dissipative adaptation in another way, as I already put into the bullet point above.
One last thing: I misunderstood the point you were making when you were talking about blackholes. The point you were making was ‘”What maximizes entropy” is a bad morality’; what I thought I was reading was ‘dissipative adaptation does not work because it predicts that we will into a black hole and Earth developed complex life because the complex life did some nuclear fission after it was developed’.
My point was a bit more complex. Yes, there’s absolutely the morality argument—obviously something that prescribes “thou shalt make black holes” is a dumb morality and I will not follow it. But there’s also a predictive power argument. At a planetary scale, putting aside all the complexity issues you rightly bring up, it may be possible that life truly maximises entropy production given certain constraints. The Earth would have more entropy as a black hole, but the potential barrier to reaching that state is enormous, and so we’re stuck in the local maximum of a planet teeming with life instead. But Beff and e-acc carry the argument all the way to the universal scale, and that’s where it breaks down, because at the universal scale, black holes absolutely do dominate entropy production, and everything else is a rounding error, so life becomes inconsequential for the ledger.
To make a practical example: suppose future humanity becomes a Kardashev 3 civilization, using up all the energy output of the Milky way and dissipating it at cosmic background temperature via radiation. That makes for an entropy production of approximately 3⋅1035W/K. Now suppose that this powerful civilization at some point predicts that two stellar black holes, each of 3 solar masses, will at some point in the future merge near an inhabited system, and this will cause trouble. With their immense power, this civilization finds a way to change the trajectory of one of those black holes, avoiding the merger, and save the system. Well, with that single change this civilization has averted the creation of roughly 3⋅1054J/K of entropy, that is, over 3 trillion years’ worth of their current entropy production! The civilization that does this will forever be a net negative in entropy creation for its whole existence, regardless of how much it splurges on using energy otherwise.
So, entropy production itself does not predict life at universal scales. It can’t. Life is just a tiny rounding error several digits down on that balance sheet. And even if on some local scales it may be possible that life is an avenue to maximizing entropy, overall those goals don’t stay aligned all the way to life taking over the universe.
the only point I disagree on is that I think that a tree is in fact a more efficient dissipator than no tree
I think that genuinely depends on details like the precise colour of the soil and efficiency of the plant. We know photosynthesis is not very efficient at energy conversion (IIRC the top efficiency belongs to the sugar cane and is a meager 8%). Also, you could probably make a more dissipative surface by putting up a very dark, very efficient solar panel and then using it to power a heater. I suppose there’s an argument that solar panels are created by life but that seems like a very tortuous way for thermodynamics to work.
I’m a second-year mathematician, and I’ve been looking at this for my one non-math module. Here are some comments I can make:
Jeremy England in a 2015 paper (“Dissipative adaptation in driven self-assembly”):
> At this point, dissipative adaptation should seem like too simple an
idea to be true, not least because the reality is more complicated
(even in principle) than the account of things we have managed to
give within the confines of this Perspective. For one thing, our dis-
cussion has trod rather lightly over the (addressable) issue of rare
trajectories that dissipate much less than average, which are known
to come to the fore in a confounding way when we try to compute
exponential averages like in equation (4). Even more significantly,
we must acknowledge that it is not always the case that a history of
absorbing and dissipating work from a drive corresponds to a con-
tinuing ability to do so in the present moment; sometimes the likeli-
est outcome of a non-equilibrium process is a pile of broken shards
rather than a resonantly vibrating goblet, which means we cannot
generally assume that dissipation rates must always increase over
time — indeed, in the case of the near-equilibrium linear-response
regime, dissipative adaptation reduces to the requirement that there
be extra entropy production during relaxation to the steady state, so
that Prigogine’s condition of minimum entropy production holds at
the end of the process.
I think he is aware of your “the devil is in the details” analysis which you gave in your “Earth” subsection. In all the papers of his I skimmed, he is aware that the reversal probability is very difficult to calculate.
What I think happened (epistemic status: low; my current best explanation based of some skimmed papers and some media response) is that he was very excited to have potentially explained life in 2014, and went on to make some “controversial” and dubious claims to journalists. While he did not push that narrative any further (the last journal article to reference England’s “theory of life” was in about 2017), there are a lot of people out there who have only engaged with those 2014-2017 articles and the attached hype of explaining life.
(In fact, I only conviced myself that this theory has very limited predictive power for the emergence of life today. What convinced me is that the “reversal probability” term can easily dominate the “dissipative adaptation” term; moreover, it is infeasible to estimate the reversal probability in a complicated system, due to the difficulty of averaging exponentials compounding the difficulty of modelling a high-entropy system, and hence it is infeasible to estimate whether the reversal probability term dominates the dissipative adaptation term in all but the simplest toy models. I can’t remember if you said this in this article, or England said this somewhere, but we’re back to calculating the probability of life emerging by calculating all the chemistry by hand.)
Interestingly, England again returns to dubious theories on life in his 2020 book, Every Life is On Fire. My guess (again, epistemic status: low) is that he actually started writing that book in 2014, and went through development hell until the publishing date; eventually, the book was published, but it never got to it’s central thesis, that dissipative adaptation explains life, because England had already rejected that thesis. He achieved his purpose on talking about the relation of the bible to his work, which I won’t say anything on because I find it irrelevant, but it does explain his motivation for publishing the book. The only chapter that really got into talking about dissipative adaptation was chapter 7 (out of eight); there were some very interesting things likening some results in non-equilibrium statistical physics to machine learning. Among them, [solving non-linear differential equations by letting a synthetic octopus arm flop about]. (https://www.frontiersin.org/journals/computational-neuroscience/articles/10.3389/fncom.2013.00091/full).
The conclusions about life I did get from the book:
When life does develop (as opposed to non-life), it will be an efficient dissipator of the local source of low entropy. Coming back to the plant example, plants absorb low-entropy light more efficiently than barren ground, which reflects a lot of it back into space in it’s low-entropy form (as opposed to the high-entropy black body radiation). In a sense, this is how dissipative adaptation can explain darwinian evolution? I’m not versed in evolutionary mathematics, so I’m not comfortable doing the math to verify this. Equally, I’m not entirely certain that my plant example isn’t (in some sense) a coincidence, as I can’t verify the maths.
Dissipative adaptation turns up in smaller examples of life utilising it over others. For example, as alluded to in the linked paper above, octopus propulsion is not fully controlled from the brain, but also relies on the interaction with the dynamics of water. I also think some good examples can be found in plants that respond to stimuli. How does a sunflower turn towards the sun? How does a wind-sculpted tree grow into that shape? The answer is likely probabilistic, and are examples of living things using their “environment for computation”.
One last thing: I misunderstood the point you were making when you were talking about blackholes. The point you were making was ‘”What maximizes entropy” is a bad morality’; what I thought I was reading was ‘dissipative adaptation does not work because it predicts that we will into a black hole and Earth developed complex life because the complex life did some nuclear fission after it was developed’. So, it seems that the only point I disagree on is that I think that a tree is in fact a more efficient dissipator than no tree; also, using a seed growing as an example of life to showcase dissipative adaptation is not a good example on your end, because you appeal to the intuition that a seed dying is more likely to invert than a seed growing — “tree-space” is large, and a seed is very robust against dying spontaneously; hence I disagree that your claim is intuitive, and evidently wrong, since a good seed is overwhelmingly likely to grow in any good soil. Plants CAN be good examples of dissipative adaptation in another way, as I already put into the bullet point above.
My point was a bit more complex. Yes, there’s absolutely the morality argument—obviously something that prescribes “thou shalt make black holes” is a dumb morality and I will not follow it. But there’s also a predictive power argument. At a planetary scale, putting aside all the complexity issues you rightly bring up, it may be possible that life truly maximises entropy production given certain constraints. The Earth would have more entropy as a black hole, but the potential barrier to reaching that state is enormous, and so we’re stuck in the local maximum of a planet teeming with life instead. But Beff and e-acc carry the argument all the way to the universal scale, and that’s where it breaks down, because at the universal scale, black holes absolutely do dominate entropy production, and everything else is a rounding error, so life becomes inconsequential for the ledger.
To make a practical example: suppose future humanity becomes a Kardashev 3 civilization, using up all the energy output of the Milky way and dissipating it at cosmic background temperature via radiation. That makes for an entropy production of approximately 3⋅1035W/K. Now suppose that this powerful civilization at some point predicts that two stellar black holes, each of 3 solar masses, will at some point in the future merge near an inhabited system, and this will cause trouble. With their immense power, this civilization finds a way to change the trajectory of one of those black holes, avoiding the merger, and save the system. Well, with that single change this civilization has averted the creation of roughly 3⋅1054J/K of entropy, that is, over 3 trillion years’ worth of their current entropy production! The civilization that does this will forever be a net negative in entropy creation for its whole existence, regardless of how much it splurges on using energy otherwise.
So, entropy production itself does not predict life at universal scales. It can’t. Life is just a tiny rounding error several digits down on that balance sheet. And even if on some local scales it may be possible that life is an avenue to maximizing entropy, overall those goals don’t stay aligned all the way to life taking over the universe.
I think that genuinely depends on details like the precise colour of the soil and efficiency of the plant. We know photosynthesis is not very efficient at energy conversion (IIRC the top efficiency belongs to the sugar cane and is a meager 8%). Also, you could probably make a more dissipative surface by putting up a very dark, very efficient solar panel and then using it to power a heater. I suppose there’s an argument that solar panels are created by life but that seems like a very tortuous way for thermodynamics to work.