Reviewers' comments
Reviewer 1: Eric Bapteste (Université Pierre et Marie Curie)
I might be strange but I often like to read Eugene Koonin's papers. This one, about the origin of replication and translation mechanisms within a particular cosmological framework, is vastly metaphysical. Biologists might appreciate that it addresses the limits of application of classical evolutionary thinking to early life stages in a pretty original way. Developing rigorous metaphysical arguments – especially when dealing with cosmological scenarios- is however very tedious. It probably requires a solid philosophical background to be done convincingly, and most likely the collaboration of both evolutionists and philosophers to write such an ambitious manuscript. A collaboration of that kind would help clarifying the meaning of many of the terms used and help presenting what specialists of the field consider to be the important notions and issues in cosmology, possibly coupled with a fair introduction for biologists about the alternative cosmological main models.
In my short review of this paper, I will assume that, although there is no doubt Eugene Koonin is a very bright evolutionist and a powerful thinker, he can not be considered as an expert on cosmology. For this reason (maybe), I feel that, if there are positive aspects in this manuscript, there are also important sources of concerns in it.
Author's response: There are a great many issues touched upon in this not so short review, and replying in full would, effectively, take another paper. Furthermore, to write such a paper properly, one might indeed need to be (or closely collaborate with) a professional philosopher. I will therefore restrict myself to a few brief comments on specific points, and a closing statement on the relationships between physics and metaphysics, and between the Anthropic Principle and Intelligent Design; additional relevant discussion can be found in my response to Krakauer below.
As a researcher, I welcome the multiplication of alternative scenarios (metaphysical – when coherent- or scientific) to address scientific questions. This pluralistic approach is probably the best way to test our favourite scientific theories by challenging them, in any case to keep them alive and debated.
Eugene Koonin's present appeal to cosmology in order to address a complex evolutionary issue tells me that, as a scientist, he feels that there is currently a problem with the scientific theory aiming to explain the origin of replication and translation. In other words, his approach says something both about him and about the possible limits of our discipline on that topic. In this review, I will simply try to rephrase what serious problem Koonin has identified according to me, and I will argue that I am afraid his answer to this problem might open too broad an avenue to the supporters of intelligent design, as it is currently formulated, and thus does not satisfy me as such as an alternative to the theory to the RNA world.
Koonin's problem and solution
Koonin's problem (the central problem of biology, he claims! (cf.p.6)) is that traditional evolutionary models can not help conceiving the emergence of replication and translation systems: for him, they would be too complex to have evolved independently of a primordial replication-translation system by natural selection acting on ribozymes alone. To solve this issue, Koonin proposes that a coupled complex system of translation and replication emerged instead by chance and by anthropic selection. After what he assumes that Darwinian selection took off and that the traditional evolutionary thinking can be applied safely to analyse the rest of the biological evolution.
To overcome the fact that the likelihood of the emergence of a complex system by chance alone is low, Koonin's multiplies the universes where such a phenomenon could be observed. Although some event e is deemed to be rare in one universe, in a multiverse framework, there must be one universe where such e happened. In his own terms: "spontaneous emergence of complex systems that would have to be considered virtually impossible in a finite universe becomes not only possible but inevitable under MWO, even though the prior probabilities of the vast majority of histories to occur in a given O-region are vanishingly small" (p. 6).
This position raises itself many concerns (biological and philosophical):
- Is it really true that traditional natural selection could not explain the emergence of the replication and translation systems?
- In any case, why should we assume that the emergence of translation and replication had to be coupled in the first place? Would not it simplify the "central problem" (i.e. make one of these questions amenable to classical evolutionary thinking) if both of these issues were uncoupled? For instance, assuming replication evolved first, could not it be possible that translation evolved progressively under Darwinian selection?
- Is a transition from anthropic selection to Darwinian selection possible and likely?
- Is it possible to assume such a transition without taking the high risk to reintroduce teleology everywhere in the evolutionary field?
Deeply, I agree with Koonin that explaining the origin of replication cycles in general challenges any sort of thinking based on natural selection and goes beyond the classical evolutionary theory, in the sense that Darwinian evolution needs replication to happen and evolutionists need replication cycles (and descent) to reason. Explaining the origin/cause of the phenomenon of replication is thus a big problem. Yet, I disagree with Koonin that natural selection and cycles of replication have to happen late in the cosmological history. To me it (only) takes a population of imperfect replicators of whatever nature for the natural selection to act, and some models on early life suggest how molecules, crystals of some sort could replicate on different inorganic substrates. Once replication is in place, and from a traditional selectionist perspective, it does not seem impossible that emerges a biological system of minimal complexity from which the process of biological replication itself could evolve under natural selection. In other words, I would locate the so-called "central" problem one step earlier than Koonin does in his manuscript: if we (evolutionists) were asked to explain replication in general, we might get in trouble, and we might be seriously tempted by metaphysical arguments.
Koonin bravely tries to tackle such a deep conceptual issue, using metaphysics where, according to him, science does not seem to work, but I am afraid his present (and arguable) solution, although fairly underlining one of the limits of traditional evolutionary thinking, could open a huge door to the tenants of intelligent design.
My general issues with the manuscript
Philosophy or biology or both?
At various locations, precisions are needed to allow a real understanding of the first (cosmological) section of the paper
Maybe because my background in both biology and philosophy is limited (I have a phD in both), I have a general issue with several parts of the manuscript, which should be either more rigorously developed or should be significantly shortened. At the very least, I believe that to be truly understood the author must propose -very early in his manuscript- clear definitions (in a box, or in the main text) of the following terms:
Cosmology, probable, possible, multiverse, anthropic selection, Darwinian selection.
Author's response: I took up this suggestion and included some of such key definitions in a Box. There was no need to define Darwinian selection, in my opinion, but the less common key notions, and those common ones that are reinterpreted here are defined.
He can not just assume that every reader shares his conceptions of the meaning of these different notions. He should also clarify what in biology are "macroscopic histories" by giving biological examples of it. In general, he should make clear when he is using a word in its philosophical sense or in its more common meaning. For instance, I am not sure what Koonin means by "repeated" in his sentence: "each history permitted by conservation laws of physics is repeated an infinite number of times in the multiverse" (p. 5). Does he mean each history is duplicated, exactly identical, an infinite number of times (i.e. the very same organism reoccurred in the very same context endlessly)? Depending on what the source informing his opinion was (philosophical or not), there might be a possible confusion. Indeed for the process-philosophers (who come to my mind and who have tried to develop a cosmology), repetition never means repetition of the same, but all the time repetition of something different (see Deleuze for instance).
Author's response: Let us be clear: not a single term in this paper is used in any specific philosophical meaning. The cosmological models and concepts discussed here are physics not metaphysics, even as they have important philosophical implications (see my response at the end of this review for a more general discussion).
First major concern: should we reintroduce teleological processes to explain what can not be explained by natural selection?
Biological systems replicate, yet Koonin tells us we can not know how they happen to do it in the first place and that we need to seek for an answer outside of the realm of traditional evolutionary thinking to address this question. He calls in multiverse and anthropic selection.
I have particular trouble with this second notion.
If Koonin really wants to promote the anthropic principle the way he defines it – ("According to the anthropic principle, the only "reason" our O-region has its specific parameters is that, otherwise, there would be no observers to peer into the universe", p.5)-, he seems to be assuming no less than a major role for teleology to explain life biodiversity (namely here that the evolution of observers to peer into the universe is the root cause of evolution). In general, teleology means that processes are directed by their goal-to-be-achieved, that their real causation is the later consequence the structures developed in the evolutionary process will lead to. It certainly makes sense in a context of replication (like descent with modification and natural selection) that a selected effect (i.e. the possession of a phenotype) can act as a cause on the evolution of future organisms of the same lineage (i.e. if the development of an eye fulfils a function of perception, therefore it can be selected if it increases the fitness of perceptive organisms in the population). However, assuming that teleology is acting before the replication process and natural selection are started is equivalent to assume that there is an a priori reason for which there has to be one day, on earth, organisms with eyes, whose goal will be to see. Unfortunately, such a position does not distinguish itself from the conception of the supporters of intelligent design either. ID people could always claim that life evolves for a reason (they could also enrole Darwinian selection as a mean to achieve this higher goal/reason).
Thus, I want to precise that I do not see what's really brought in by invoking such an "anthropic principle" (but trouble). The idea that evolution would be oriented toward consciousness is a recurrent cosmological option indeed but it is definitely not the only one possible, and certainly not mine. A deeper philosophical exploration of cosmological theories would most likely have showed that evolutionists do not have to embrace a strong anthropic principle as if it was an integral part of the cosmological package introduced here (too roughly) by Koonin.
Yet, maybe Koonin himself does not care so much for this anthropic principle, as I suggest below.
Author's response: this is a very, very serious misunderstanding of the anthropic principle. The idea that anything is "oriented toward consciousness" is pernicious nonsense. The revised version of the paper includes a brief debunking of the so-called "strong anthropic principle", in response to this unfortunate misstatement and a similar one made by Krakauer. I have to be frank here: to me, this mistake invalidates much (not all) of the critique in the review. There is no teleology at all involved in my approach in this paper. No teleology. It is the opposite of teleology. More on this in my closing response to Bapteste.
Second concern: assuming Koonin's model, are there strong evidence of the transition from anthropic selection to Darwinian selection?
Koonin thinks that anthropic selection is inevitably replaced by Darwinian selection once a sufficiently efficient mechanism of replication/translation has been put together. Thanks to such a transition, the teleological order of things would thus be replaced by the Darwinian order of things, under the action of natural selection. (In other words, once his problem is solved, Koonin does not need alternative models to evolutionary thinking anymore). Unfortunately, as long as he does not carefully introduce and clarify his anthropic selection principle, I think that he might be unwillingly opening a huge door to ID supporters to enter into the very heart of the evolutionary theory.
What guarantee do we have indeed that Darwinian selection would replace anthropic selection?
What guarantee do we have that some anthropic selection would not persist, having some effects on the living world, although of lesser importance than those of Darwinian selection? What guarantee do we have that it would not be still acting on and then be responsible, here and there, for some lucky natural "miracles"? (i.e. the birth of a pandemic, a spontaneous generation, a benevolent mutation for superior organisms, any kind of oddities (toward consciousness), since after all the anthropic principle does not say if the observers to peer (that peer? peering?) into the universe have to come sooner rather than later, does it?) ?
Author's response: Again, there is absolutely no teleology involved, I could not insist more strongly on this point. Beyond that, however, we certainly cannot have a guarantee that the anthropic principle was not involved in subsequent evolution because it definitely was. Certainly, in the context of the infinite multiverse, there is an enormous number of worlds where eukaryotes have never evolved...or animals etc. The history of life is riddled with contingency, and there is nothing new about this notion. Not that I am so keen to rely on authority but leading scholars of evolution from Jacob to Monod to Gould advocated the importance of contingency with great force. Gould's metaphor of rewinding the tape of evolution – and finding that even major events would not be the same – is, perhaps, most vivid. A prominent philosopher, Dan Dennett, also wrote about this in "Darwin's Dangerous Idea" and even claimed that evolution is governed by deterministic chaos. An unsettling description, perhaps, but essentially true, I believe. Thus, the anthropic principle remains relevant throughout the history of life although, after the breakthrough stage (see Figs. 1 and 3), it is, largely, subjugated by the Darwinian process. I included a brief comment to this effect in the revised manuscript.
Worse: what guarantee do we have that there won't be ID people to claim that, as the very important evolutionary biologist E. Koonin showed, Darwinian selection is a secondary player in cosmology, and is itself a force evolved for a reason, a chosen by product of anthropic selection, in which case everything evolving under Darwinian selection evolves in fact under the eternal drive of the anthropic principle? (As I read it, in this paper Koonin did not prove that anthropic principle and natural selection were not working hands in hands, and that the later was not following the former for a reason of higher order).
For all these questions could be exploited very meanly by ID people, Koonin is in my view very naïve to think that he can call in teleology to start the process of evolution and call it off subsequently: if teleology is called in (a view which I doubt is worth supporting in science and that I feel is dangerous), unless teleology achieves the goal it is needed for, there is no reason why teleology should suddenly stop being relevant in the middle of the way... In fact, a teleological process deemed to be non successful is no more a force, as by definition it will fail to achieve its goal, and then why to invoke it as a force in the first place?)
Does Koonin really want to claim that "the basic structure of the [genetic] code results from anthropic selection in as much as only codes with a certain minimal level of robustness would allow the appearance of a functional replicase in the breakthrough system" (p.13), and take the risk to be badly misinterpreted? Obviously, he is smart enough to foresee this kind of problems and the reason why he claims he does not worry about it (p.18) is perhaps that this kind of "anthropic selection" is not really what he has in mind.
Third concern: is our world Koonin's brave new world?
Maybe, instead of invoking such a strong anthropic selection, Koonin is arguing along much simpler lines. I suspect he is in fact introducing a model from which teleology could be entirely absent to justify that, if everything with a low probability is possible somewhere, even though it has a low probability in a given universe, providing a high enough number of universes, everything -including the oddest phenomena- becomes necessary somewhere. Then, complex structures for replication have to emerge. By chance, we happened to be in the "right" type of universe where this very emergence took place, says Koonin. It is just maths: the initial paradox can be solved by an axiom: it had to be so or we would not be here to tell. Now, some people (me included) might find this position to overcome the paradoxical origin of replication a little bit dry (and panglossian).
Author's response: So Bapteste, after all, understands what I am trying to say. Is the whole basis of the criticism above the mistaken identification of anthropic principle with its ridiculous strong version? Very regrettable if so. In the revised manuscript, I stress this point in the text and in the Box. Hopefully, this eliminates any cause for confusion. Now, whether or not the concept developed here is dry is, of course, just a matter of taste. However, the characterization of my position as "Panglossian" cries for a comment. I believe that the worldview presented here is, actually, anti-Panglossian. Indeed, in an infinite multiverse with a finite number of histories, there is no chance that we live in the best of all possible worlds. There is an infinite number of worlds that are incomparably better, even those where Elvis is still alive in 2007 as wryly noticed by Garriga and Vilenkin (refs. 1 and 2). Thus, I believe that here I am relegating Dr. Pangloss to irrelevance, continuing along the lines of Gould and Lewontin's famous San Marco paper.
Especially if this kind of claim is applied again and again to several complex problems we are (and we will be) facing. As Koonin notices, even if it is unlikely, one could indeed propose that the whole life organization of a universe is just the necessary outcome of such a model. That's why in his paper, a central question subsequently becomes to decide up to which point complex life phenomena to be explained on our planet were necessary rather than the result of Darwinian selection (i.e. at which point traditional scientific explanations become valid again). Regarding that issue, Koonin takes a strong position: when the conditions for Darwinian selection are satisfied (namely once biological replication and translation systems have been set up), necessity leaves our world. (Someone with a different "central" problem might thus observe that, when Koonin does not need necessity anymore to play the role of the explanans of his own biological problem, he drops it; yet maybe another "central" problem still challenges traditional evolutionary thinking, should we invoke the necessity and the mutiverse for it one more time?).
If what I described is what Koonin had in mind rather than the anthropic selection, the good news then is that since such necessity is strictly supported by statistics, he can propose that things are the way they are because they had to be so, without needing to invoke observers of the universe and to open the door of evolutionary biology to ID people.
Author's response: I appreciate these points. The wording in the original manuscript was unnecessarily rigid, in an attempt to strongly emphasize the transition. I made modifications to acknowledge the important role of contingency at subsequent stages of evolution as well.
Yet, scientists could wonder if this principle position (use necessity when the Darwinian model does not apply, else use the Darwinian model) is the only and best null hypothesis to deal with issues regarding early life. They could notably question:
- how much time is needed for everything to happen somewhere in the multiverse (and if this is a realistic time line from our human perspective),
- if this multiverse scenario is significantly more likely than chance events on a single universe,
- if it is that impossible for populations of ribozymes to be self-sustaining and self-replicating?
Koonin tries to address these questions with some toy statistical models, yet I am very skeptical that toy probability calculations can prove decisive in that matter. (For instance, Whitehead's cosmology (a solid philosophical reference but not an easy read) talks a lot about societies and populations, nexus of all kinds of elements, which could have emergent properties, and could offer interesting breakthroughs, the probabilities of which are probably impossible to derive mathematically).
To summarize, it is legitimate that Koonin tells us where he feels comfortable applying traditional evolutionary thinking. It is legitimate that he tries to propose possible explanations to go beyond that limit. Yet, I am not convinced that there is more than intuition in this paper and that he could prove that our world (the explanandum) is the brave new world he, as a scientist, would be happy with.
I am thus interested -and amused- that, in this paper, Koonin ends up multiplying the universes to solve a biological issue, when I remember how much, on other occasions, he enjoys to invoke parsimony (the assumption that one should not multiply beings without necessity). I would be more satisfied however when he would have made his own views on ID and the anthropic principle clearer in a revised version of this manuscript.
Author's response: Considering the main points of Bapteste's comment, I believe I need to summarize my position on three issues: i) physics versus metaphysics, ii) the nature of the anthropic principle, iii) the relation of the present concept to Intelligent Design (ID). Obviously, the first point can be discussed at great length, so I will just say it a nutshell. The concept developed here is a purely scientific one, positioned at the interface of physical cosmology and straightforward evolutionary biology. If some aspects of the paper seem unusual and counterintuitive, that is mainly because such are the key features of the MWO model. Of course, the MWO model has profound philosophical (both metaphysical and epistemological) implications (see refs. 2 and 13 for discussion), and the biological implication developed here has additional ones. There is nothing unique in this: when the fundamental aspects of the structure of the world are concerned, physics and philosophy merge. In the most obvious example, foundational research in quantum physics had been like that for the last 80 years.
I have already made several statements on the anthropic principle above, but I think a more definitive brief explanation should be useful. Once again, we deal here only with the weak anthropic principle that has nothing to do with any teleology. Moreover, it is a fairly trivial statement, "pure math" as Bapteste puts it. In an infinite multiverse, where all possible scenarios actually realize, albeit with very different probabilities, we naturally find ourselves in one of those, extremely rare O-regions where the conditions are conducive to the evolution of complex life forms. This is all there is to anthropic selection which it is not any kind of active selection process. This concept is the very opposite of any kind of teleological scenario: there is nothing special about our region of the universe except that it belongs to the relatively small biophylic domain of the parameter space of the multiverse (e.g., refs. 14,15), and even within that domain, a region with such properties is rare (the fact that this region is very special to us because we live here is scientifically irrelevant). Effectively, under this concept, the conditions for the onset of biological evolution emerge by the power of large numbers, without the involvement of any special interactions let alone any directional process. I think this is the ultimate anti-teleogical stance. To conclude the discussion of the anthropic principle, it might be helpful to emphasize, once again, its crucial link to the multiverse model. In a solitary universe, as depicted by the classical Big Bang model, anthropic principle would amount to enormous luck; that, indeed, would be a Panglossian world. In the infinite multiverse, the element of luck is removed, and the emergence of an infinite number of instantiations of life is guaranteed as long as life is compatible with the laws of physics (the one example we are familiar with proves that it is), even if the biophilic regions of a universe are enormously far between. I would like to quote a rather categorical statement of Leonard Susskind (the inventor of the original version of string theory) on the connection between the multiverse (megaverse, in his parlance) and the anthropic principle: "Without the idea of a megaverse of pockets, there is no natural way to formulate a sensible Anthropic Principle" (Ref. 8, p. 300).
The main text of this paper contains a clear statement on Intelligent Deisgn (ID) but, since this is a serious concern for Bapteste (and I agree that ID is an important, even if meta-scientific, issue), I will reiterate and further reinforce my position that directly follows from the reasoning explicated in this paper. The above discussion of the anthropic principle implies an unequivocal sentence for Intelligent Design; let us spell it out. As indicated above, in an infinite multiverse, anthropic selection guarantees the emergence of systems of whatever complexity that are required for the biological evolution to take off (see Fig. 3). This scenario is watertight: first, chance/anthropic selection, then biological evolution. There is no gap here where the ID wedge could fit. Properly interpreted, the anthropic principle is a death knell to ID.
Obviously, anthropic principle is often misinterpreted as providing support to religious (and otherwise teleological) believes (regrettably, this includes Bapteste's review). On this issue, I cannot resist quoting the recent book by Richard Dawkins: "It is a strange fact, incidentally, that religious apologists love the anthropic principle. For some reason that makes no sense at all, they think it supports their case. Precisely the opposite is true. The anthropic principle, like natural selection, is an alternative to the design hypothesis. It provides a rational, design-free explanation for the fact that we find ourselves in a situation propitious to our existence" (Dawkins, R. 2006. The God Delusion, Houghton Mifflin, Boston-NewYork, p.136).Just like Dawkins, I find it difficult to identify the exact reasons behind the confusion (though see further discussion in my response to Krakauer's review). The possibility that the ID crowd interprets this paper as support for their cause is one of Bapteste's main concerns. Will they, actually? No doubt they will! However, the only way to prevent them from doing so is to stop publishing research on any hard problem in evolutionary biology and somehow declare these problems solved. The ID folks do no research themselves, so they apply all their considerable intellectual resources to turn published scientific work upside down and claim support for ID (it happened to several seemingly innocuous papers of mine, to my considerable amusement). I believe evolutionary biologists should not and actually can not worry about this, only about their own papers being correct and coherent.
Answers to all your Biology Questions
