Evolution for Everyone

Just as Brutus was a close companion to Caesar but proved to be his undoing, evolutionary theory seemed to provide a rock-solid foundation for individualism– until Lynn Margulis came along.

Lynn is famous so you might already know her story. In the 1970’s she proposed the radical theory that nucleated (eukaryotic) cells evolved not by small mutational steps from bacterial (prokaryotic) cells, but as symbiotic communities of bacteria that became so integrated that the group became a higher-level organism. She was fiercely opposed but carried the day, an accomplishment so great that she was admitted into the National Academy of Sciences in 1983.

The concept of organism as group was generalized in the 1990’s by John Maynard Smith and Eors Szathmary in two books titled The Major Transitions of Evolution and The Origins of Life: From the Birth of Life to the Origins of Language. Their theory was multilevel selection theory with a twist. The evolution of group-level adaptations requires a process of group-level selection and is undermined by selection within groups. Now for the twist: The balance between levels of selection is not static but can itself evolve. When between-group selection sufficiently dominates within-group selection, the group becomes a super-organism and the lower-level organisms acquire the status of organs. The evolution of nucleated cells was just one of many major transitions, preceded by the evolution of the first cells and possibly even the origin of life itself as groups of cooperating molecular interactions, and followed by the evolution of multicellular organisms, social insect colonies, and–as we shall see–human social groups.

John Maynard Smith is the same person who opposed group selection in the 1960’s, as I recount in T&R VII and IX. Somehow he managed to make group selection the lynch pin of major transitions without ever acknowledging an error in his earlier views. Eors Szathmary, his coauthor, fully appreciates the continuity of ideas. Joel Peck is another theoretical biologist who is fully comfortable with multilevel selection and was close to John during the last years of his life. I recently asked Joel how John could have been so schizophrenic about his early and later ideas. Joel shrugged and replied with a question of his own: How does anyone believe in ideas that are inconsistent with each other, such as scientists who also accept religious dogma? Perhaps Maynard Smith simply couldn’t bring himself to thoroughly reconsider his earlier ideas on the basis of his own later ideas.

It might seem that the evolution of multi-cellular organisms from single-celled organisms would be simple when they begin as a single cell, because then all the cells are genetically identical and there can be no selection within groups. Wrong. Mutations occur with every cell division and there are thousands of cell divisions in the lifetime of a multi-cellular organism such as you and I. Mutant cells that are selectively advantageous within the organism are certain to arise and spread, regardless of their effect on the welfare of the group. Cancer is just multilevel selection in which we are the groups. Over the eons, elaborate physiological mechanisms have evolved by between-individual selection to suppress within-individual selection as much as possible.

The concept of major transitions is one of the most important developments in modern evolutionary thought, but the implications for individualism and multilevel selection theory are seldom discussed. Individualism is the claim that individual organisms are a privileged level of the biological hierarchy, that all of nature and human nature can be explained in terms of individual-self interest, that groups can emphatically not be regarded as organisms writ large, that the idea of individuals evolving “for the good of the group” is deeply erroneous. If within-group selection invariably trumps between-group selection, then all of these claims would be true and evolution would indeed provide a rock-solid foundation for individualism. That’s why the issues at stake in the 1960’s seemed, and were, momentous.

The issues remain momentous when we decide that between-group selection is important after all and can even dominate within-group selection in the case of major transitions. If we noted the rock-solid foundation that evolutionary theory seemed to provide individualism before, we should equally note when the foundation crumbles. Individualism has nothing left to stand on. Individual organisms are highly integrated and tightly regulated societies. Organisms turn into mere groups when their organization is disrupted by natural selection from within. Mere groups turn into organisms when between-group selection trumps within-group selection. When selection operates at both levels, individuals become strange hybrids of solid citizens and self-seekers, while groups become strange hybrids of coordinated units and dysfunctional outcomes of conflict from within.

Some people will mourn and others will celebrate the death of individualism, but everyone needs to bury it and move on to explore the implications of multilevel selection theory, especially in the realm of human affairs.


  1. #1 NewEnglandBob
    November 9, 2009

    I gave these a shot since you came to scienceblog but I still do not see a case made for group selection. I will take this blog out of my RSS reader.

  2. #2 piker
    November 9, 2009

    Individualism has been defined as the doctrine or belief that all actions are determined by, or at least take place for, the benefit of the individual, not of society as a whole. As far as philosophers are concerned, that concept died with Ayn Rand. Even the economists have now acknowledged its passing. Biologists are just now getting around to burying it?

  3. #3 Sam C
    November 10, 2009

    Individualism has nothing left to stand on. Individual organisms are highly integrated and tightly regulated societies.

    I think this nonsense explains why you don’t “get it”. You don’t really have any understanding of units of selection, do you? As far as you are concerned, group selection is The One Truth, so everything must be group selection. To a man who only has a hammer…

    So this is just another post that is just a load of petty-minded snarking: nyaa, nyaa, Maynard Smith woz rong an’ I iz rite. If this is your idea of “truth and reconciliation”, well, group selection is a blue banana.

    I popped your QR paper into the recycling this morning. Couldn’t see anything in it, just as these posts are free of interest.

  4. #4 daedalus2u
    November 10, 2009

    Wow. I am a newcomer to the individual vs. group selection wars and it is interesting to see how the different sides are dealing with/approaching this. It reminds me a lot of the autism wars. Of the extreme need by many individuals who are neurologically typical (including senior autism researchers) to impute pathology into anything and everything that does not match their ways of thinking and relating.

    As I understand what DSW is saying, he is not saying that there is no such thing as individual selection, or that group selection explains everything, or that group selection is The One Truth.

    All he is saying is that individual selection does not explain all things that are observed in evolving populations. The example of the chickens clearly shows that. It shows completely different results when chickens are selected on an individual basis and on a group basis. The hypothesis that group selection is no different than individual selection is clearly false. That doesn’t prove that individual selection is not important, all it proves is that group selection can be important too, and so cannot be ignored.

    Pointing out where people have ignored group selection and have instead invoked The One Truth of individual selection is not to deny the reality that individual selection can and does happen. There is no data showing that group selection cannot happen. Simplistic models may claim to show that group selection can’t happen, but those models are so simplistic and flawed as to not be applicable in the real world.

  5. #5 don't be frightened
    November 10, 2009

    What is so scary to you people about considering selection acting at multiple levels? Hamilton, EO Wilson (Mr. Sociobiology), social insect biologists, cultural evolutionists, primate biologists, sexual conflict biologists and some guy you might have heard of named Charles Darwin, all recognized the simple fact that selection acts at multiple levels. Group selection is not the end all but it is clearly relevant in some scenarios and you only lose understanding of the system if you discount it. Wilson is showing that discounting group selection altogether is unwise. All of the previous models on kin altruism, reciprocity, tit for tat, indirect reciprocity all rely on multiple group populations or they fail, plain and simple. It is a group of reciprocators who outcompete a group of non-recips, not a single recip outcompeting a non-recip in a one vs. one. It is a cooperative kin group who outcompete a non-coop kin group, not a cooperative sibling outcompeting his/her selfish sibling. Tit-for-Tat loses to a selfish partner in a one vs. one. Punishment does not evolve if considering the population a single group. This is even recognized by the authors of the models. Yes, of course, it boils down to an individual’s fitness, but is the pathway to how this fitness is acheived. Is fitness this best acheived by gaining a relative fitness advantage over his neighbors eventhough it may cost you in the long run or by cooperating with his neighbors to obtain a greater fitness over others in the population. So I ask you Sam C, do you understand the units of selection? or would you rather explain how all of these models (i.e. tit for tat) work within a single group? If you could figure that out there is a seat in the National Academy of Sciences waiting for you.

  6. #6 Luke Vogel
    November 10, 2009


    Excellent series on T&R, enjoyed it a great deal. I’m sure we’ll be seeing much more on multilevel selection.

    And congratulations on the debut of *EvoS Journal*. I hope it is successful, it’s obviously very needed.

    ~ “The first issue of EvoS Journal: The Journal of the Evolutionary Studies Consortium — a new open-access on-line peer-reviewed journal designed to promote the education of evolutionary theory in colleges and universities — is now available. The journal is published by the Evolutionary Studies Consortium, of which NCSE is a member institution. The consortium seeks to “facilitate the development and implementation of Evolutionary Studies Programs at colleges and universities across the United States”; the original model for such programs is David Sloan Wilson’s Evolutionary Studies Program at Binghamton University.”

  7. #7 David Sloan Wilson
    November 11, 2009

    Thanks to daedalus2u and don’t be frightened for coming to my rescue. Multilevel selection is the reasonable middle ground, compared to categorical claims that all genetic evolution is at the gene level (true but no argument against group selection) or that within-group selection invariably trumps between-group selection (just plain wrong).

    Thanks to Luke Vogel for plugging the EvoS consortium and its journal. Group selection is not my only interest and I look forward to completing the T&R series so I can move on to other topics.

  8. #8 abb3w
    November 14, 2009

    don’t be frightened: What is so scary to you people about considering selection acting at multiple levels?

    Candidly, the math gets pretty terrifying.

    The root of evolutionary selection is the second law of thermodynamics, which is already enough math to make most people queasy. To comprehensively handle multi-level selection means that you have to consider the entropy flow for every definable boundary; that is, across the boundary between every subset of the universe and it’s complement set. (Can you say “power set”, children?)

    Thus, most humans prefer to simplify the math to working at the boundaries with most obvious impact. “Organisms” are the penultimate head of the list. (“Biosphere” is higher up, but until we have evidence of life outside Earth geosync, the associated math becomes trivial.) Like most linearizations, it’s sloppy— but useful… and non-linear equations get pathological very, very easily.

    Since I’m not having to do the math and get results useful enough to do engineering design (such as “experiment design”), but merely diddle around with mathematical philosophy, I’m not bothered.

  9. #9 daedalus2u
    November 14, 2009

    DSW, I am fearless when it comes to agreeing or disagreeing with ideas, even my own depending on what the data says. I have worked and continue to work extremely hard to conform my thinking to reality. It always does surprise me when other people don’t. My autism research is helping me to understand why some people do have such a hard time conforming their thinking to reality, and recognizing when they are not doing it.

    abb3w, the math does not become terrifying, it becomes impossible. All physiological systems are comprised of multiple coupled non-linear parameters. A system of coupled non-linear parameters becomes intractable when the number of parameters is more than 2 or 3. When the number is more than 5, the problem is essentially impossible. When the number is thousands or tens of thousands as it is in physiology, the math is impossible by many orders of magnitude even if every system was completely deterministic and had zero degrees of freedom (which isn’t what physiology has).

    The paper you linked to is quite problematic. The authors use terms in ways that are inconsistent with their normal meaning and make some assertions that are actually not correct. The last paragraph on page 3067 is simply not correct and how ever comforting it is to think of evolution in such simplistic terms it is not correct and will impede understanding until such simplistic and wrong conceptualizations are abandoned. It appears to me that the authors understand neither entropy nor evolution.

    They state that evolution favors greater effectiveness in dissipating energy, but that is clearly false. Photosynthesis only works by capturing some of the energy from the high energy photons and storing it as chemical energy rather than allowing it to immediately dissipate as heat. A plant is a less effective way of dissipating solar energy than is a black body. If evolution favored greater effectiveness at energy dissipation, then photosynthesis would never evolve. All evolved systems capture energy, which means they are less effective at dissipating that energy than a system that immediately dissipated that energy as heat.

    Entropy doesn’t work the way the authors suggest. Boltzmann’s great formulation, S=k*Ln(omega) is about partitioning energy into states. The more possible states the energy can be partitioned into, the higher the entropy. Two warm crystals have higher entropy than the sum of a cold crystal and a hot crystal because there are more possible ways to assign that energy. After mixing, there are more phonons but of lower energy and so there are more ways that energy can be distributed among the states that are in the two crystals.

    Everything that happens only occurs because entropy has increased. Entropy is a state function; it only depends on the final state, not on the path by which that state has been reached. Evolution depends on dissipation of free energy, but the degree of free energy dissipation has no relationship to the kind or degree of evolution that has gone on. Their arguments bear some resemblance to those of Creationists who claim that because the process of evolution generates more information that the second law of thermodynamics prevents it from happening. This is not at all the case. The continued metabolism of living organisms dissipates vastly more free energy and creates vastly more entropy than could possibly be represented by information changes in DNA. It is like comparing the energy of a lifetime of food consumption to the energy in a few pg of DNA in a fertilized egg. The metabolism of a hundred tons of food can support the rearranging of few pg of DNA.

  10. #10 abb3w
    November 14, 2009

    daedalus2u: When the number is more than 5, the problem is essentially impossible.

    There is no efficient solution; this, however, is different degree of “impossible” from having no effective solution. It’s the difference between an engineering impossibility and a mathematical one; a problem in complexity P, versus complexity R. Pure Mathematicians don’t get too bothered until you’re well into RE/Co-RE territory (if not further); solution of O(n)=A(Gn,Gn) steps is “merely tedious”.

    daedalus2u: They state that evolution favors greater effectiveness in dissipating energy, but that is clearly false.

    Hardly “clear” to me. You might care to look into some of Adrian Bejean’s work; in particular, the paper on forests (doi:10.1016/j.jtbi.2008.06.026) he has on his site looks related.

    daedalus2u: A plant is a less effective way of dissipating solar energy than is a black body.

    My instinct is that this may not be the particular mathematical sense of “effective” in question; but I’m afraid that you’d have better luck asking Annila about that. I can follow his math, but doing problem-solving at that level is beyond my skill.

  11. #11 daedalus2u
    November 15, 2009

    By “impossible” I didn’t mean non-computable, just that human resources and abilities are insufficient to do it. I am an engineer, so if something can’t be done with the resources available then it is “impossible” in the sense I was using. If a solution is known to be not achievable via a certain path, it doesn’t make sense to try and get a solution via that path.

    Systems of multiple coupled non-linear parameters are not predictable long term. The weather will never be predictable a year in advance. The reason is because the weather is chaotic and differential initial conditions have macroscopic effects. Physiology is chaotic too. There are many “strange attractors” in the weather. What the weather will be on a particular day is not predictable a year in advance, but we know that there will be weather on that day. Similarly we can’t predict the phenotype of an individual from their genotype, but any given genotype will produce a phenotype (one out of a very great many possible phenotypes). Each of those different phenotypes is a strange attractor too. In a chaotic system it can take differential change to get from one strange attractor to another. We know that can happen, we can’t calculate it because we don’t have the resolution to do so.

    If you did have the computational power to calculate the phenotype a genotype would produce, there is not a unique answer. If you took 1,000,000 identical genotypes and grew them into 1,000,000 different organisms you would get 1,000,000 different phenotypes. If you could model that process with 100% fidelity you would get multiple phenotypes too.

    In a mathematical sense, since the number of atoms in an organism is finite, the number of possible configurations of those atoms is finite also. The number of possible configurations is so large as to be not realizable.

    I looked a little bit at Bejean’s work. I don’t disagree with what he is saying, I see it as an optimization problem and if you optimize over the parameters he is using you will get solutions of the form that he is finding. The problem is that the parameters that evolution is optimizing over are not limited to the ones that he is using, and many (virtually all) of the parameters are not known. There is also the problem of hysteresis and local minima with an activation energy to get to a still lower minima. There are the boundary conditions, the other constraints and these are mostly not known either.

    Thermodynamically, the most stable material is nickel-62; everything lighter than niclel-62 can be fused to make nickel-62 and liberate energy, everything heavier than nickel-62 can be fissioned to nickel-62 and liberate energy. The kinetics for non-nickel-62 atoms to spontaneously convert to nickel-62 is exceedingly slow and not significant even though the energy liberated is very large. Thermodynamics is about energy and equilibrium, not about the time or the path it takes to reach equilibrium.

    No organism has evolved to generate nickel-62 as a waste product and likely never will. A “theory of evolution” that predicts organisms will evolve to generate nickel-62 is not one that I would find useful.

  12. #12 Guy
    November 16, 2009

    daedalus2u wrote: “A plant is a less effective way of dissipating solar energy than is a black body.”

    I appreciate your knowledge and interest in the thermodynamics of life. You might be interested in a paper on this subject I published a few years ago with colleagues.

    Hoelzer, G. A., E. Smith, and J. W. Pepper. 2006. On the logical relationship between natural selection and self-organization. Journal of Evolutionary Biology 19: 1785-1794.

    I agree with you that a black body on a rotating earth would effectively dissipate solar energy. However, I disagree with the following:

    “If evolution favored greater effectiveness at energy dissipation, then photosynthesis would never evolve. All evolved systems capture energy, which means they are less effective at dissipating that energy than a system that immediately dissipated that energy as heat.”

    I think the earth dissipates solar energy more effectively with photosynthesis than it would without photosynthesis. There is no mechanism I know of that could make the earth a black body as an alternative. Therefore, the energy gradient created by the earth’s shadow presented a potential for the emergence of effective dissipation mechanisms and life taps into that gradient as a fuel source. Your comments seem to assume something about efficiency (speed of dissipation) as well as effectiveness. I may agree with you on this assumption, but this is a pretty contentious issue right now. I think it will be important to make issues of efficiency explicit and to work them through.

  13. #13 abb3w
    November 16, 2009

    daedalus2u: By “impossible” I didn’t mean non-computable, just that human resources and abilities are insufficient to do it.

    …completely, anyway. Which, at essence, is the original point of what is “scary to you people about considering selection acting at multiple levels”. Most scientists like to work where the math has some vague hope of being addressed in human scope, more exactly than “the temperature in Kansas tomorrow will be somewhere between zero and 10000 Kelvins”. They don’t like math where after three months on a Cray you say “well, that was possible under current theory after all”.

    Multi-level selection increases the degree of non-linearity, which increases the extent you can only use approximations. This increases headaches.

    This, unfortunately, is life.

    daedalus2u: The problem is that the parameters that evolution is optimizing over are not limited to the ones that he is using, and many (virtually all) of the parameters are not known.

    Agreed; however, some thermodynamic parameters dominate in some cases. EG: individual selection is often locally dominant… if, perhaps, not always, and for limited values of “locally”.

    daedalus2u: No organism has evolved to generate nickel-62 as a waste product and likely never will.

    Depends on your definition of “organism”, and whether “big rip” or “big freeze” scenarios are correct. I would agree any cellular organism is unlikely to ever have any DNA-encoded mechanism for same.

  14. #14 piker
    November 16, 2009

    Download Guy’s paper at:

    Excellent work done there.

  15. #15 daedalus2u
    November 16, 2009

    Guy, the Earth’s albedo is 0.367, the moon’s is 0.07. Does that mean the moon is more evolved than the Earth because it dissipates solar energy better?


    Some asteroids have an albedo of 0.04. Does that mean they are even more evolved than the moon is? Does that mean there is some unknown life form that is much more effective at photosynthesis on the moon and asteroids that is dissipating solar energy that much more effectively?

    I know it doesn’t mean that, but that is what the hypothesis of “things evolve to dissipate energy most effectively” predicts.

  16. #16 Guy
    November 16, 2009

    daedalus2u wrote “that is what the hypothesis of “things evolve to dissipate energy most effectively” predicts.

    That is not even close to what I would mean by this statement. The Prigogine view, which I admire and largely embrace, suggests that gradients represent opportunities for the emergence of self-organizing systems. Life appears to be a classic example of a dissipating and self-organizing system. However, there is nothing in the theory predicting that life will emerge to dissipate any gradient, or that the gradients dissipated by live could not be effectively dissipated by alternative mechanisms.

  17. #17 daedalus2u
    November 16, 2009

    Having now read your paper I understand better where you are coming from. It is different than what was suggested in the paper linked to in comment #8.

    I think I subscribe to Prigogine view also.

    I completely agree that gradients are opportunities for self-organizing systems, and that all dissipative systems (which includes all organisms) require gradients from which they can extract free energy.

    I don’t disagree with anything you have on comment #16. I do see that at odds with what you said about photosynthesis and dissipation in comment #12 and in your paper.

    Life and other self-organizing systems can only acquire energy by extracting and preventing the immediate dissipation of energy in gradients. Living systems are not accelerating the dissipation, they are delaying it. Solar energy is not dissipated by being converted into chemical energy during photosynthesis, it is being captured. Capturing energy is completely different than dissipating it (in a thermodynamic sense). Energy captured is energy not dissipated.

    It is more accurate to characterize living systems as capturing the energy in gradients by preventing its dissipation.

  18. #18 Guy
    November 16, 2009

    daedalus2u wrote: “Living systems are not accelerating the dissipation, they are delaying it.”

    Well, I think living systems have to do both to exist. In fact, I would say that is true of all self-organizing systems. They “tax” channeled flow across a gradient and use this to fuel their construction and activities. Storage is necessary. However, dissipation is also necessary, because there would be no flow to tax without this service to the 2nd law of thermodynamics. Some of the energy that goes into living systems is stored and some is directly dissipated. In the end, it all dissipates. The net effect is an increase in the rate of dissipation, compared with the absence of life, with the interesting phenomenology of living systems emerging in the process.

  19. #19 daedalus2u
    November 16, 2009

    Guy, no, there is not an increase in dissipation. If we consider the moon, all the sunlight that falls on it right now is dissipated as heat. 100% of the energy is dissipated as heat. If we installed solar panels on the moon and captured that energy and used it to build a civilization on the moon, the separation of aluminum oxide into aluminum metal and oxygen would represent a storage of the energy from sunlight into the chemical energy of aluminum and oxygen. If you did an energy balance around the moon; by capturing sunlight and storing that energy in metallic aluminum, there would be less dissipation of the energy from sunlight because some of that energy would be stored. At steady state, after millennia of building, if the only energy source available was solar energy, at most 100% of that energy could be dissipated. That is not an increase over how much is dissipated right now. There is nothing that you can do to increase the dissipation of that energy to more than 100%.

    Life does not cause an increase in dissipation, there is a decrease. Life only exists because it captures energy being dissipated and uses that energy for its own metabolic processes.

    Life may take stored energy and dissipate it, as stored hydrocarbons can be oxidized, as iron can be oxidized, as organic compounds can be oxidized. But in virtually all cases, that energy is from biological processes in the first place which was captured instead of being dissipated.

  20. #20 Guy
    November 17, 2009


    I think we are making alternative hypotheses that can be tested in principle. I don’t accept your moon scenario as relevant, because the solar panels do not work like self-organizing systems. They are engineered to capture more energy than they dissipate in the process. My conjecture is that self-organizing systems, like living systems, dissipate enough energy over any increment of time that the net rate of dissipation is higher than would be the case in the absence of the system. This works under the 2nd law because for each time increment the system is effectively paying its dues. The second law would require any system that does not meet this criterion to degrade, rather than to build or sustain. Indeed, your solar panels would be degrading during the thought experiment you proposed. Growth and repair are fundamental to life in my view. I would concede that it is conceivable for a living system to degrade for a relatively short period of time and to recover before death is permanent. This might describe the situation with viruses, but it is hard for me to imagine this condition as long as there is an active metabolism. In any case, I think it is pretty clear that the net effect of living systems is to increase the rate of dissipation over the absence of living systems. If the alternative hypothesis is seriously plausible I think it would be very important to test between the two.

  21. #21 daedalus2u
    November 17, 2009

    Guy, solar panels don’t capture more energy than they dissipate. The best solar cells (multi-junction) are only ~30% efficient. That is they convert 30% of the incoming energy into electricity the rest is dissipated as heat. 70% of the energy is dissipated as heat. That is less than the 100% that is dissipated as heat when light strikes a black body.

    Suppose instead we used some living photosynthetic organism instead. Then maybe 1% of the energy is converted into stored chemical energy. 99% is still dissipated. The amount dissipated can never exceed 100%, so it can never be higher than in the black body case. Living systems capture energy instead of allowing it to dissipate as heat.

    The kind of system is irrelevant, self-organizing, living, inanimate, or even no system. It is thermodynamics.

    Living systems do need to dissipate energy. They need to capture the energy they dissipate from the environment. They can capture that energy from a dissipating gradient (as in capturing solar energy instead of allowing its dissipation as heat), or from stored energy (oxygen and oxidizable substrates).

    With an external source of energy (i.e. from the solar energy gradient), a system can continue to dissipate that external source indefinitely. Whether that system is inanimate (black body), living (photosynthesis), or mechanical (solar cells and robotics) is irrelevant. The thermodynamics is the same.

    There is a saying in chemical engineering, that “you can’t beat thermodynamics”. You can beat kinetics (i.e. with a catalyst), but you can’t beat thermodynamics. Thermodynamics is about end states, kinetics is about the path between states. Thermodynamics determines the states that are possible, but you need kinetics to find a path between them.

  22. #22 Guy
    November 18, 2009


    I agree with your thermodynamics, but you are missing my point. In the emergence of life on earth becoming a black body was not a thermodynamic alternative. As far as we know, there was no thermodynamically favored path that could have turned the earth into a black body with a dynamical positive feedback system. If such an alternative was equally available (the “adjacent possible” sensu Kauffman), given the local material constraints, I would expect that path would have manifested instead of life. If we agree that the black body path was not available, then my conjecture still holds. The emergence of life on earth increased the rate of dissipation compared with the absence of life. It also resulted in some storage of energy which, in addition to quickly spent free energy obtained by taxing the flow, is the payoff living systems could use to self-organize and behave.

    I don’t think this argument contradicts any of the thermodynamic points you made. Please correct me if I am wrong.

  23. #23 Sky Blue and Black
    November 18, 2009

    As an aspiring biochemist, I am happy to see such productive and respectful discussions taking place even where peer review does not edit or weed out disrespectful content. Although scientists will find themselves dead-set against other scientists on matters of theory, predictions of the future, or fundamentals, it is important to keep criticism polite so it can lead to the advance of knowledge rather than its stagnation while we all shout and refuse to listen. A few of the earlier comments were antagonistic, but the later ones are almost entirely polite, and informative rather than personal. It is important that the scientific community maintain this effectiveness of discussion.

  24. #24 daedalus2u
    November 19, 2009

    I posted a reply earlier, it had 2 links so it got held up in moderation.

    I found Kaufman’s ‘Prolegomenon to a General Biology’ on line and didn’t much care for it. I appreciate that it is an introduction to a longer piece, but I don’t like his casual (and I think wrong) application of information theory, Turing machines and quantum mechanics to evolution.

    Living systems do tend to self-regulate at critical points, that is where the power-law stuff comes in. Systems at a critical point are chaotic in a mathematical sense, where a differential change causes a macroscopic effect. At the thermodynamic critical point, the heat capacity and compressibility diverge, that is they become infinite. A critical point is a state of incipient instability. It is the best “default” state to have because the ease with which any other states can be reached is maximized. Neural networks self-regulate there so that a single neuron firing can cause a macroscopic change in the whole system.

    He is simply incorrect to state that an open system has an infinite number of possible states. If a system is finite, then it can have at most a finite number of possible states. That is not the same as the possible outputs of possible Turing machines being infinite. A Turing machine is a mathematical construct, it is not limited by being finite any more than the digits in pi are finite. Holding a physical circle in your hand is not the same as holding pi in your hand. You can’t measure the circumference and the diameter and then use them to calculate pi to 10^10 places. The model of a physical entity as a continuum breaks down.

    What that has to do with life and evolution is not at all clear. I think he is looking for a mystery when there isn’t one. There is nothing mysterious about the data needed to represent something after it has been efficiently coded. That is the definition of efficient coding. The “information” talked about in Shannon’s information theory is not the same “information” as is encoded in DNA.

    There is no need for a fourth law of thermodynamics needed to explain living systems. All living systems are already completely consistent with all known thermodynamics.

  25. #25 Guy
    November 20, 2009


    I appreciate your excellent post. I agree with most of what you wrote, including your criticism of Kauffman’s claims about information theory. I would simply comment that the rationale behind erecting a fourth law of thermodynamics, or revising our statement of the second law again, is not based on a problem with consistency. It is based on a desire to extend the explanatory power of the existing laws. Specifically, the aim is to capture the universally constructive aspects of thermodynamics. In my opinion, if thermodynamic construction through self-organization is a universal phenomenon, stating this as a thermodynamic law will be a very important step forward.

  26. #26 daedalus2u
    November 21, 2009

    This is the post that didn’t make it through moderation (without the links).

    If we consider the albedo of the Earth and moon, 0.37 and 0.07 respectively; what that means is that in the case of the Earth, 37% of the energy is not absorbed, but is reflected. That energy is neither captured nor dissipated, it is reflected. On the moon, only 7% is reflected, the rest is absorbed. 93% absorbed is a lot closer to a black body than is 63% absorbed.

    On Earth, the quantity of energy that is captured by photosynthesis is quite small, about 1%. That energy can either come from energy that would otherwise be dissipated or reflected.

    Energy that is reflected is not dissipated (in the thermodynamic sense that I was using, perhaps you were considering reflected energy to be dissipated?).

    In any case, living systems must capture free energy to power their own systems. But it is not necessary for the successful living system to always be the one that consumes the most free energy. Here are some values of albedo for various Earth surfaces.

    Snow is very high, around 0.8. Bare earth is lower, 0.17. Grass is higher at 0.25. The hypothesis of highest energy dissipation in a living system would predict bare earth is better at dissipating energy than is grass so bare earth should be the stable state. This value for soil is an average; values for individual types of soil are more variable, from between 0.048 to 0.402.

    With grass right in the middle between the two extremes which happened to be measure here, what is the expected stable state? The energy absorbed in photosynthesis is tiny compared to the changes in albedo that occur when grass covers different kinds of dirt.

    Thermodynamics does not determine a final state very strongly. For one state to evolve into another, the thermodynamics has to allow it. That only means that there must be some dissipation. Whether the dissipation is 0.01%, 10%, 50%, or 99%, all of those are “allowed” by thermodynamics. Which state will result depends on the path; the path taken depends on kinetics.

    The chemistry produced by the presence of living matter is simply another path.

  27. #27 daedalus2u
    November 21, 2009

    Guy, the problem with the idea that there is something like a “forth law of thermodynamics” as pertains to self-organizing systems is that there isn’t one.

    Many self-organizing systems do end up exhibiting power-law behavior, and that is a universal property of “active” systems, that is systems that are unstable, contain internal energy and can spontaneously collapse into a more stable state with the release of energy.

    An avalanche is an example, where the gravitational potential makes the pile of particles unstable. An activation energy causes movement of the pile, which releases energy which causes more movement. The size of the avalanches has a power-law dependence, where there are many more smaller ones than there are large ones.

    You probably could force fit a correlation of body size into a power-law metric if you made enough simplifications and ignored enough confounding factors. You will only get “clean” power-law correlations when you have “clean” power-law relationships, which only come from “clean” and simple interactions in active systems where the energy release couples to the activation of energy release. Avalanches are “clean” systems. Biological systems are not “clean” systems. Biological systems are extremely complex and unless that complexity can be ignored (which it usually can’t, and virtually never can be demonstrated to be ignorable), a power-law observation may be artifactual.

    I understand why people such as Kauffman want there to be some simplifying “law” to explain the complexity that is observable.

    Some biological systems do exhibit power-law behaviors, for example neuronal activity is chaotic and exhibits power-law behavior. The reason neural networks self-regulate there is because that is an optimum point to start from to switch rapidly to a different state.

    Virtually all physiological systems (when operating normally) also operate in a power-law regime, they exhibit chaotic behavior. For example the interval between heart beats is not regular, it is chaotic when the patient is healthy. As a patient’s health declines, the heart beat becomes less and less chaotic, that is it gets more and more regular. My explanation of this data is that because physiology comprises many non-linear coupled feedback pathways, they exhibit chaotic behavior (as all systems of multiple coupled non-linear parameters do); as health deteriorates and more and more of the control pathways become “out of range”, they drop out and the physiological process is controlled by fewer and fewer feedback loops. The control gets worse and worse and the heart beat is observed to get more and more regular.

  28. #28 Guy
    November 23, 2009


    I agree with all of your points, but I still think you are missing my point. At least, I didn’t see it addressed in your most recent post. The generation of power law relationships is a theoretical prediction and an empirical observation, but it is not a reason for invoking a 4th thermodynamic law. The reason, as I see it, is to recognize what is arguably a universality to the thermodynamics of self-organizing construction that was first (as far as I know) described by Prigogine. The traditional ways of stating the 2nd law have been about structural degradation as a means to increasing entropy. Self-organization in dissipative systems is a generic means to increase the rate of entropy gain in the universe. I personally find the argument that this represents a universal aspect of thermodynamics, and finally a DYNAMIC aspect, compelling if not yet proven. I have never heard an argument against its plausibility. Localized growth in structure through self-organization is at the very least a phenomenology that demands an understanding at the thermodynamic level. Life is clearly one such phenomenon, which is why I think it is so important for biologists to consider this perspective.

New comments have been disabled.