Is evolution “contingent” or repeatable?

October 3, 2016 • 10:15 am

Aeon is a nonprofit science and technology magazine that occasionally has some good pieces, though I’m not a frequent reader. However, several readers called my attention to a new piece by Dan Falk, a Canadian science writer, about whether or not human evolution was inevitable. Click the screenshot to go to the article:


Falk poses a strict dichotomy that many of you will have encountered before: was the evolution of humans (or “humanoids”: human-like creatures with high intelligence) inevitable, or was it due to a “stroke of luck”? This is the same debate that was once raged between Simon Conway Morris and Stephen Jay Gould. In his book Wonderful Life on the creatures of the Burgess Shale, Gould emphasized that humans were a “contingent” product, and if the tape of life were rewound, things could have turned out very differently. The early chordate Pikaia, for instance, could have gone extinct through “bad luck”, and hence no vertebrates and no us. The message of Gould’s book was that humans (and perforce other beasts) are here by accident.

In contrast, Simon Conway Morris, who is a Christian, has taken the view that the evolution of humans on Earth was inevitable, and if the tape of life were rewound, we’d still evolve. It’s pretty clear that his views are conditioned by his faith, because of course he sees God as having directed the process, with humans the end-product made in His image. (Also, Conway Morris’s argument based on evolutionary convergence—the repeated and independent evolution of some body plans, like those of fish, dolphins, and ichthyosaurs—doesn’t support his argument, as human intelligence is a one-off, not seen in any other animal.)

But Falk, like many others writing on this subject, is confused by the meaning of “contingency” and “accident”. I discuss all this in Faith Versus Fact, but I’ll give a precis here.

First, you have to clarify by what you mean by “rerunning the tape of life”. Do we mean starting the Earth over and over again with every molecule and particle in the same position, and then seeing what happens? If you’re a determinist, then I’ll maintain that “contingencies” and “accidents” are really determined circumstances, and so everything would pretty much evolve as it did over and over again—with one caveat (see below).

If you mean starting the Earth under different conditions, like different temperatures or different configurations of seas and continents, then of course evolution would not necessarily be repeatable, for the environment—a crucial component of natural selection—would be different. And that means that all we can say to the question “Was human life inevitable?” is “Who knows?”

Finally, you need to define what you mean by “humans”? Do you mean anything with the intelligence of humans, like hyper-cerebral octopi? Do the creatures have to have syntactic language? Do they have to recognize and worship God? (I suspect that Conway Morris would say “yes” to the last question.) Again, based on the one-offness of human intelligence, I don’t think we can say that human evolution was inevitable, even if you accept Conway Morris’s argument and the voluminous data he’s compiled on convergence in other animals. Again, we get a “who knows?” answer.

Now let’s answer the question as pure determinists. If you start the Earth over again with every molecule and particle in place, then evolution will repeat itself more or less exactly, as there are no contingencies, no “bad luck,” no accidents. All those things, like the asteroid striking the Earth, supposedly killing the dinosaurs, were inevitable consequences of the laws of physics.

Unfortunately, Falk, like many, mistakes “unpredictability” for “true indeterminism,” so he sees evolution as the result of a series of “accidents”. To wit:

But even if evolution has a direction, happenstance can still intervene. Most disruptive are the mass extinctions that plague Earth’s ecosystems with alarming regularity. The most catastrophic of these, the Permian-Triassic extinction, occurred about 250 million years ago, and wiped out 96 per cent of marine species, along with 70 per cent of land-dwellers. Gould examined the winners and losers of a more ancient mass extinction, the Cambrian-Ordovician extinction, which happened 488 million years ago, and found the poster child for biological luck – an eel-like creature known as Pikaia gracilens, which might be the precursor of all vertebrates. Had it not survived, the world could well be spineless.

But those mass extinctions were determined by physical laws, and are “happenstance” only in the sense that a well-informed human couldn’t predict them. Likewise this:

With the planting and harvesting of wheat and barley, and the domestication of livestock, it can seem like a short and perhaps inevitable step to the walls of Jericho and the pyramids of Giza. This doesn’t remove contingency from the equation. ‘There is nothing inevitable about the origins of agriculture,’ Chazan says. ‘It didn’t have to happen – but once it happened, it’s irreversible.’

Under determinism, of course, the origins of agriculture had to happen, and would happen again (along with the evolution of humans) under a true rerun of the tape of life.

So if we had a true rerun of the tape of life, with all the starting conditions repeated exactly, then, it seems, evolution would have repeated itself, and we’d have all the species, living or dead, that actually did evolve.

With one caveat.

And that is the caveat that if there is any true indeterminism in the history of life, of the quantum sort, then no, evolution wouldn’t repeat itself, even under identical starting conditions.

One such indeterminism is mutation. If mutation involves quantum effects, such as cosmic rays, then the process of mutation is fundamentally indeterministic. And if mutation is fundamentally indeterminstic, then it’s entirely possible that evolution, which is fueled by mutations, would also be indeterministic, and would not go the same way on a rerun. We don’t know the answer. Another spanner in the works: mutations are recurrent: the same mutations often happen over and over again, especially in large populations. So even if a single mutation changing a single DNA base is a single individual indeterminate, it may be statistically determined, so that again, evolution would go the same way under a rerun.

Finally, we’re only talking about Earth here. If you take other planets into account, all bets are off. There’s simply no way to answer the question of whether, somewhere in the vast Universe, humanoid creatures would inevitably evolve.

The upshot:

1). If by “rerun of evolution” we mean “a starting condition repeated exactly, with every molecule in place,” then yes, the evolution of all species, including humans, was inevitable—but only if mutations are not fundamentally indeterminate occurrences.

2). Under all other scenarios, including indeterminate mutations, the answer to the question “Was the evolution of humans inevitable?” must be “How the hell do I know?”

94 thoughts on “Is evolution “contingent” or repeatable?

  1. Wearing my pedant’s hat… wouldn’t every atom, and every subatomic particle need to be exactly “in place”?

      1. Once you frame it in terms of QFT, the determinism of the answer becomes clear.

        You may remember your school teachers posing the big conundrum of the nature of the photon and the electron: are they particles or waves? You probably weren’t told the answer: they’re waves. They only take on a particle-like appearance when you look at them macroscopically in certain contexts.

        We’re probably all familiar with magnetic fields, especially as easily visualized with a magnet and some iron filings. It’s an easy mental stretch to understand gravity similarly — as an universal space-filling field that has a value (strength) at every point. All you need to know about what something at a given point will do is know what the gravity field is at that point. That was one of Laplace’s big insights; you didn’t have to know the mass distribution of the entire Universe, as Newton thought, but just the state of the local gravitational field.

        Quantum Field Theory has taken that same idea and applied it to everything. Not only is there an universal magnetic field, not only is there an universal gravitation field…but there are similar universal fields for what most people think of as the fundamental particles and forces. And an electron, for example, is simply an harmonic vibration in the electron field. The strength of the electron field is the likelihood that your equipment will interact with the electron field at that particular location in spacetime.

        Consider now the double slit experiment. The electron gun fires off a single ripple in the electron field — and the “Quantum” part of “Quantum Mechanics” has to do with why the fields are quantized into discreet ripples of given strengths. The ripple heads towards the two slits and diffracts as it passes through them. The diffracted wavefront continues on to the detector — which, of course, is, itself made up of a mind-boggling number of similar quantum field ripples in the form of all its constituent particles.

        It is at this point that the remaining popular deep mystery of Quantum Mechanics becomes apparent. We know that the diffracted electron wave is spread out over the entire detector, but the detector only ever records the signal at one small location. The wave appears to have “collapsed” at one random spot and vanished from the rest of the detector.

        The favored explanation is that the wave doesn’t vanish from all those other spots, but, rather, the “interference patterns,” if you will, between the incoming electron and all the particles in the detector are such that the one site in the detector can no longer see the effects of electron at the other sites. Effectively, the waves “branch” and go their own separate ways.

        Thus, the paradox: the electron goes through both slits, but you’ll only ever be able to image it going through one side. At the same time, an almost-indistinguishiably-identical other “you” will see it on the other side. And both are equally the “real” “you,” just as a particular ripple in a wave traces its history back to a particular splash even as all the divergent cross-sections never reintegrate.

        So, a much more meaningful question to ask in the context of modern physics is not, “If we reset the particles and reran the tape would humans appear?” but, rather, “What proportions of other branches of the universal waveform that date back to a given time have humans?”

        It helps to consider these sorts of things in contexts of different levels.

        The proportion of branches which saw the electron go through whichever slit in any given execution of the experiment is 50% – 50%, as dramatic a divergence as you can come up with. But the proportion of branches in which the Sun continues to rise in the East is 100%. The proportion of branches dating back to yesterday in which at least one human observes the Sun rise in the East tomorrow will be 100%. The proportion of branches dating back a century in which at least somebody saw the Sun rise in the East today is also 100%, at least within any rounding I could type.

        But the proportion of branches dating back a dozen billion years ago in which humans here on Earth are observing the Sun rise…is certainly less than 100%. I’d guess it’s a larger percentage than many people might suspect, but it’s less than 100%.

        Another similar question worth pondering, at a completely different scale, is the number of technological civilizations alive in the Universe right now. We know it’s at least one. We can be reasonably confident that it’s numbered in the billions at least, and should not be surprised if it’s hundreds of trillions — and even such mind-bogglingly huge numbers would be so small in comparison to the rest of the Universe that we’d have no hypothetical hope of ever directly observing one.

        And we can make some reasonable assumptions about them. They will have biological evolutionary histories of their own, covering a span of at least a billion years or so.

        A significant fraction, perhaps even a majority, will be bilaterally symmetric. They will have manipulative digits. Some may be upright bipeds with a pair of manipulative limbs and a sensory cluster on top…but that’s the closest description you’d be able to make of any of them as being “humanoid.” The apex aquatic predators they’re familiar with will be torpedo-shaped. They will have social structures including strong moral codes, with distinct echoes between all of them but no universality.

        And any religions they have — and at least some of them will have religions — will be even more bizarre to us than our own ancient obscure cults. They will be even more perplexed by Christian missionaries than a Zoroastrian would be perplexed by a Scientology missionary.

        Sorry…that was rather longer than I initially intended…gotta get back to pretending to be productive at work….



        1. “What proportions of other branches of the universal waveform that date back to a given time have humans?”

          Don’t all branches date back to the same time? If not, then were there fewer branches in the past? Is that possible if there are an infinite number of branches now? (IOW, when did it change from finite to infinite)

          And if there are an infinite number of branches, wouldn’t the probability of at least one branch having humans be 100%? (after ‘restarting’ at a previous time).

          1. Sorry…I might not have phrased that clearly.

            Trace our own branch back to a given point in time — say, 1 billion years ago. There certainly would have been all kinds of other branches that had long since gone their own way by then, but we’re just looking at the great granddaddy of branches of the universal waveform that we can ourselves trace back to that time.

            Since then, that one branch has itself been branching like crazy, far beyond the ability of mere mortal minds to even pretend to comprehend.

            Of those branches, how many have recognizable humans?

            Is that any clearer?



        2. “What proportions of other branches of the universal waveform that date back to a given time have humans?”

          That was very well said. The Everett (“Many Worlds”) Interpretation has sounded a lot better to me ever since I learned that decoherence predicts a smooth yet rapid apparent “collapse” into “worlds” that for all practical purposes are separate, yet remain part of a single wavefunction.

        3. We can be reasonably confident that it’s numbered in the billions at least, and should not be surprised if it’s hundreds of trillions — and even such mind-bogglingly huge numbers would be so small in comparison to the rest of the Universe that we’d have no hypothetical hope of ever directly observing one.Actually…we cannot be reasonably confident in anything even close to those numbers. We might be alone in fact. Recent paper by A. Loeb in Journal of Astrophyics Aug 2016 comes to the conclusion that given the cosmic history, mass/age of sun with earthlike planet in habitable zone, stellar density and other factors that humanity is the first advanced civilization on the scene in our universe. Paper at; and faith/science commentary at .

    1. One such indeterminism is mutation. If mutation involves quantum effects, such as cosmic rays, then the process of mutation is fundamentally indeterministic. And if mutation is fundamentally indeterminstic, then it’s entirely possible that evolution, which is fueled by mutations, would also be indeterministic, and would not go the same way on a rerun.

      Even if initial states were exactly the same the universe wouldn’t turn out exactly the same, otherwise quantum states wouldn’t be probabilistic, they’d be deterministic.

      This using of the issues I have with the rejection of free will on the basis that if you rewound the clock you’d do precisely the same thing all over again. I don’t want to divert the thread but the ‘rewind the clock’ argument is neither evidence for nor against free will.

      Still, probability isn’t the same as randomness. Some outcomes are more likely than others, some are so unlikely we can reasonably ignore them.

      1. Even if initial states were exactly the same the universe wouldn’t turn out exactly the same, otherwise quantum states wouldn’t be probabilistic, they’d be deterministic.

        Jerry’s point still applies though (IMO): the question is whether the indeterminacy would impact evolutionary development. And the answer is, we don’t know.

        1. It’s pretty clear that quantum theory implies indeterminacy. But to the long term consequences of that indeterminacy are not really addressed by quantum mechanics itself – this would lie more in the application of chaos theory to evolution and the biosphere. If you like, quantum indeterminacy may kill a butterfly, but chaos theory is the way to understand the butterfly effect.

        2. Consider what it would mean to say that the indeterminacy would not impact evolutionary development. In effect you’d be saying that quantum mechanics can cause microevolution but not macroevolution.

          But we know that’s a false dichotomy. There is no sharp line between small-scale change and large-scale change: the latter is just the accumulation over time of the former. There’s no mechanism by which cumulative change can be limited to microscopic effects; given enough time, the divergence can become arbitrarily large.

          1. No, there’re clear limits to how large a divergence can be. Again again, no divergence is going to cause the Sun to rise in the West.

            I would challenge the thought experiment that says that all the air molecules can hypothetically “randomly” migrate to a far corner of the room and stay there. Molecular air movement is mechanically deterministic, not random. It is certainly chaotic at sufficiently small scales…

            …but consider an analogy of a billiards ball break. All sorts of possibilities can present, with most having small percentages of balls going in pockets. It’s at least close to impossible for an human-powered break to sink all the balls, and it is impossible for an human-powered break to sink all the balls in numerical order in a single pocket — the human can’t impart enough kinetic energy to the ball for the reaction to continue for the amount of time necessary for such an ordering. Why such is impossible is intuitively obvious, and the math will bear it out.

            The air in the room is like a billiards game but with umpteen brazilian orders of magnitude more balls. Chaotic and unpredictable at small scales, sure, but that most emphatically does not mean that all imaginable configurations are actually possible.

            There are limits, and we understand those limits superbly well. They’re collectively called, “physics.”




          2. There are places on the surface of Mercury where the sun does rise in the west on some days of the year.

            There are almost certainly extrasolar planets of Earthlike mass and orbit but with retrograde rotation.

            So clearly such histories are physically possible. The question then becomes: if we go back far enough in the formation of the Solar System, is there a branch point where some of the branches lead to such histories for Earth? I think it very likely there is, and I think you’d be very hard pressed to rule out that possibility based on what we know of physics.

          3. The question then becomes: if we go back far enough in the formation of the Solar System, is there a branch point where some of the branches lead to such histories for Earth?

            You’d have to go back to long before there was anything recognizable as a “solar system,” before even the formation of the primordial disk of dust whose center would eventually ignite. Celestial mechanics is about really, really big stuff.

            Think of the proverbial butterfly flapping its wings; there’s no way it’s going to result in the Earth’s axis shifting. And yet a butterfly is about as far beyond Quantum scales as the Earth is beyond butterfly scales. You could load up a potato canon full of butterflies and fire it off in the early Solar System and not have a measurable effect on the eventual axis of the Earth’s rotation, so the suggestion that Quantum-scale effects are significant really doesn’t play out.




          4. You could load up a potato canon full of butterflies and fire it off in the early Solar System and not have a measurable effect on the eventual axis of the Earth’s rotation

            I think your intuition is leading you astray here. In fact the movement of hurricanes has measurable effects on the Earth’s rotation. That’s mostly what leap seconds are for.

            The predictable clockwork Solar System of yore is a fantasy; the reality is far more chaotic. I don’t think there’s any basis in physics to rule out the possibility that quantum divergence could be amplified to planetary scale.

          5. In fact the movement of hurricanes has measurable effects on the Earth’s rotation. That’s mostly what leap seconds are for.

            The first sentence is true, but on the order of millimicroseconds. Leap seconds have much more to do with the tidal effects of the Moon slowing down the Earth’s rotation, with uncertainty coming from the speedup from the contraction of the Earth due to internal gravitational settling.

            And all that gets overwhelmed by the precession of the seasons, which is a result of the 20,000-year cycle of the wobble of the Earth. It becomes smoothed out to a day that’s…I forget, exactly, but maybe an hour or so longer today than when the dinosaurs were at the top of the heap.

            I don’t think there’s any basis in physics to rule out the possibility that quantum divergence could be amplified to planetary scale.

            You don’t have to know whether or not the fundamental physics permits it to know that it, in fact, is not possible. The experiment has been run, and, for billions of years, the Earth has rotated and orbited rather close to what it does today. The magnetic poles flip every hundred thousand years or so, but the orbital pole has never once reversed or significantly deviated from the current amount of tilt.

            Indeed, run the simulations for what will happen when the Sun goes through its red giant phase and expands past the Earth’s orbit…and the Earth may or may not emerge from the Sun’s atmosphere when it contracts, but it’s not going to flip over.

            To flip the Earth, you need to hit it with something Mars-sized at the right angle. Or you need the coordinated equivalent of microscopic changes. When you understand the magnitude of such impacts…you’ll realize that there’s even less chance of Quantum indeterminacy influencing the Earth’s orbit than there is of you winning every single lottery for the next four centuries.




          6. for billions of years, the Earth has rotated and orbited rather close to what it does today

            The approximate shape of the Earth’s orbit has remained fairly stable, but the Earth’s position in its orbit is unpredictable on a much shorter timescale (tens of millions of years). And if we’re talking about mass extinctions due to asteroid collisions, orbital position is what matters.

            I think it would be interesting to know how much quantum events contribute to those error bars. As far as I know, nobody has done that calculation, and until somebody does, I don’t think we can conclude that the answer must be zero.

          7. I think it would be interesting to know how much quantum events contribute to those error bars. As far as I know, nobody has done that calculation

            You don’t even need to break out pencil and paper.

            The net effect of any Quantum weirdness to shift the Earth is going to have to equal the same amount of change as is necessary for a classical mechanism to perform an equal shift.

            As you yourself have noted, it takes an entire hurricane to just barely produce enough of an effect on the length of the day for us to manage to measure with our most precise equipment ever made. We can safely rule out Quantum indeterminacy from spontaneously manifesting even a small whirlwind, let alone an entire hurricane — let alone something as dramatic as the presumed impact with a Mars-sized body that created the Moon but still didn’t manage to flip the Earth’s rotation.

            You might insist on holding on to something along the lines of cumulative changes over billions of years, but that still amounts to winning the lottery every year for centuries on end, with the lottery example not even coming remotely close.

            Yes, there are knife-edge-balanced phenomena that can have dramatic changes from small initial conditions. But those phenomena themselves don’t tend to be all that significant in the grand scheme of things.

            …again, consider my example of a quantum cookie / axe murdering spree….

            (f you really want to look for such an example, you have to go back to the very early history of the Universe, shortly after the Big Bang. If nothing else, the extraordinarily low entropy of the Big Bang should tell us that there was a great divergence of macroscopic histories at that point. But Inflation happened (and is still happening), and the number of possible outcomes decreased equally dramatically as entropy correspondingly increased. Eventually, with the heat death of the Universe, entropy will be so high that time will effectively stop. You could have all the Quantum indeterminacy you want in that setting, and still nothing is going to keep happening because there won’t be anything for it to happen to.

            …though, of course, it might be that it was from exactly such nothingness that the Big Bang itself spontaneously Quantumly indeterminately happened in, but that’s an entirely different story….




          8. We can safely rule out Quantum indeterminacy from spontaneously manifesting even a small whirlwind

            Give me some credit, Ben. I haven’t made any claims about spontaneously manifesting anything, let alone Mars-sized impactors.

            But if we’ve reached the point where you’re inventing irrelevant absurdities in order to shoot them down, you can do that without my continued involvement.

          9. But that’s my whole point.

            You’re suggesting that there exists or could exist some branch in the history of the Earth such that the Sun rises in the West, and that that arose through quantum indeterminacy. My point is that that’s on a scale far beyond hurricanes spontaneously manifesting through quantum indeterminacy.

            Maybe I’m misunderstanding the scale of what you’re proposing is possible through quantum indeterminacy. But my initial point with raising the “Sun rises in the East” example is that there really are limits to how far quantum weirdness can reach into the macroscopic world, and those limits are much more strictly classical than almost everybody realizes.

            To bring it back on topic again…I’d acknowledge that there might be cases where the particular mutation that grants immunity to a resistant strain of bacteria might be more readily identifiable as coming from quantum indeterminacy, but even that gets washed out really quickly once you run enough trials, and the end result is very difficult to distinguish macroscopically. By the time you get to population genetics of megafauna, you’re as far removed from realistic quantum effects as that spontaneous quantum hurricane.




          10. I agree with Ben. While not disputing the existence and real effects of indeterminacy, the scale at which organisms exist is large compared to the subatomic scale, and so what impacts organisms are the collective, statistical behavior of those particles. That collective, statistical behavior tends to adhere very close to the classical approximation, and so is (IMO) reasonably considered deterministic.

            Could indeterminacy have impacted evolution? Oh sure. The scenario is easy to imagine: a nuclear decay releases a gamma ray at one indeterminate moment rather than another. That gamma just so happens to hit a sperm/egg and cause a genetic mutation that is passed on. Had the decay occurred a millisecond later, it would not have hit that egg at just that location. The course of life on Earth is changed from that one indeterminate action. But can we say for sure that such contingencies were important to the evolution of life on Earth? Did my scenario ever actually happen? I don’t think we have the data to answer that question definitively one way or another.

          11. I don’t think we have the data to answer that question definitively one way or another.

            Oh, we do — and we know the frequency is given by the Born Rule.

            But we can calculate the Born Rule for such events, and it’s close enough to zero to be a rounding error. And, yet, the number of times when such an “experiment” is run is large enough that we get the statistical distributions of point mutations we observe.

            Working simultaneously with numbers both so small and large can lead to unintuitive results, but the net effect is typically nice and smooth statistical distributions if you zoom out to the proper scale.




          12. I grant that classical physics is an excellent approximation, good enough for us to land spacecraft on Mars.

            But it’s quite another matter to claim it’s accurate enough to produce negligible divergence on a timescale of billions of years.

            Once histories diverge, they must continue to diverge, because there’s nothing to keep them in sync, and there’s many more ways to be different than to be similar. The accumulation of random microscopic divergences (which are not rare; they happen continuously) inevitably add up to macroscopic divergence, for the same reason that entropy increases.

          13. Once histories diverge, they must continue to diverge, because there’s nothing to keep them in sync, and there’s many more ways to be different than to be similar.

            This is, of course, equivalent to a statement about entropy.

            First, again consider the sorts of difference we’re talking about — with my example of the cookies and the quantum coin toss being really good. The histories after you eat the cookie on the left or the right aren’t radically different. The histories after you eat second cookie and you go on an axe murdering rampage are radically different.

            There isn’t anything reaching across the histories to keep them in sync, obviously. But neither is there anything reaching across successive rolls of the dice to produce the nice Bell curves. We have no problem understanding that the emergence of such curves is inevitable with the dice, so we shouldn’t have any more problem understanding similar inevitability with respect to other macro-scale phenomena.

            Let’s say that you’re engaging in an analysis of dice rolls, one that spans trillions of rolls. And imagine you use the quantum coin flipper to manually change exactly one of the rolls to its complement — a 1 to a 6, a 5 to a 2, and so on. All the other rolls you leave unmolested. Can you seriously propose that you’d be able to differentiate between the eventual statistical analysis of the final results?

            Such is what you’re proposing with your single-quantum-event-dependent alternate histories, only the numbers are far more vast than mere trillions.




          14. Ben, in all of your posts here you’re inexplicably ignoring chaos theory, which is concerned with precisely the question in hand – what is the sensitivity of future states to small changes in initial conditions?

          15. Chaos is real, yes. But most systems aren’t chaotic — and chaos vanishes as the resolution of the analysis goes down.

            You might not be able to predict a single roll of the dice, but you can predict the statistical distributions of a sufficient number of them. The individual toss is chaotic, but the collection of tosses becomes more and more predictable the more times you throw the dice.

            And the macroscopic world tends to much more resemble dice throwing tournaments on steroids than a single coin toss whose outcome will determine whether Satan or Jesus gets to direct the next move.




          16. You seem to be using “chaotic” as a synonym for “random”. It is not. It has nothing to do with the distribution of random variables, and the law of large numbers is not applicable to the evolution of the state of a system through time.

          17. But it’s quite another matter to claim it’s accurate enough to produce negligible divergence on a timescale of billions of years.

            Why? Billions of years is trivial compared to 1E26 particles constantly interacting, which gives us classical approximations like temperature and pressure. Wait a billion years, ten billion years, a billion billion years, you still won’t find all the atoms in a room on one side of it. Prokaryotes are composed of (I’m estimating here) about 1E16-1E20 atoms. That’s more than plenty for them to obey classical approximations of quantum mechanics rather than being influenced by more oddball QM effects. We should no more expect a prokaryote to be impacted by entanglement or electron indeterminacy than we would expect a room’s temperature to be impacted by these things, for exactly the same reason: there are many interacting particles in both systems. More interacting particles than is needed for the overall system to behave classically.

          18. Ralph is right: it’s not about large numbers; it’s about sensitivity to small perturbations. A Geiger counter contains a very large number of particles, but that doesn’t prevent us from hearing the clicks of individual decays.

            Similarly, thermodynamics is of no use in predicting asteroid impacts, despite the large number of particles involved.

          19. Again, eric, you’re discussing quantum effects on the macro scale and arguing probabilities and the law of large numbers. This is completely irrelevant to the question at hand. The question is one of chaos theory – the evolution of a system through time. Is the long term evolving state of the system highly sensitive to a small initial perturbations?

            For example, quantum indeterminacy can quite obviously operate at the immediate macro level in whether a specific DNA mutation occurs. The question to be addressed then is whether evolution is a chaotic system – to what extent is there really a “butterfly effect” in evolution? Is it the case that if a particular key mutation doesn’t occur in one iteration of history, then evolution may play out completely differently? Or is evolution “robust” and non-chaotic, because a similar mutation will occur sooner or later anyway, sufficient to produce an approximately similar result?

      1. I must object to the omission of “octopuddlitarianistsisms” from that list.

        And the Indonesian is clearly incorrect; if that’s the form you’re going to use, it’d have to be “octopus octopus octopus octopus octopus octopus octopus octopus octopus octopus octopus octopus octopus octopus octopus octopus” for each pair of wee sea beasties.



        1. I like the imaginative greek: 16opus but that would only be 2 octopuses. Maybe myriadopus would be better.

          1. Hmmm…2 opuses would be a magnum opus, right? And a pair of magnum opera would probably be a maximum opus, thereby making the individual wee beasties magnetic maxi pads, if I’ve got the math right….



      2. Years ago having read Simon Conway’s book refuting Gould,I seem to remember that he said that human like intelligence evolved in the Octopus. Does anyone remember this ?

        1. The pedant in me needs to point out that octopus is Latinized Greek….the worst kind of Greek. So it’s Greek form is octopous. It’s a 3rd declension noun and in Greek it declines like this in the plural:

          Nominative: octopodes
          Genitive: octopodwn
          Dative: octoposi
          Accusative: octopodas
          Vocative (because we all call to the octopus): octopodes

          You can read more here

  2. Point mutations may be treatable statistically, but there are other sorts that are too rare. For example, the two rounds of polyploidization in the ancestry of vertebrates, which were clearly crucial in our evolution. Then again, they wouldn’t be considered quantum events. The question is whether there’s anything small enough to be subject to quantum indeterminacy but rare enough not to be subject to statistical prediction. I have no answer.

    1. There is also the importance of drift.

      Mutations need to not only arise but to also get fixed.

      And most mutations disappear very quickly even if they are advantageous.

      Also, the more complex and larger the organism, the weaker natural selection becomes, and the more important is drift.

      I.e. the environment played less of a role in shaping the evolution of humans than it did for that of bacteria. And we are down to molecules of DNA.

      Where, of course, the question again comes down to what is indeterminate as in affected by quantum effects, and what is determinate but unpredictable. And how much that influences the final outcome.

        1. What do you mean by “rare quantum events”? Every interaction between particles is a quantum event with a range of probabilistic outcomes.

          1. John can of course answer for himself, but remember that the Born Rule describes the frequency of observation of quantum phenomena.

            Run the double slit experiment long enough and you’ll eventually get an observation in one of the dark areas of the interference pattern, but, overwhelmingly, you’ll only make observations in the light areas. An observation of the dark area could most reasonably be considered a “rare quantum event.”

            It’s not at all difficult to imagine extrapolating that to, say, some singular radionuclide decaying at the exact moment necessary to cause some key particular mutation. But I would again suggest that that we already know that that sort of phenomenon is not the prevailing paradigm driving Evolution, because that’s not what we observe. Rather, the population dynamics work over large pools of such mutations.

            So, you might not be able to predict which mutation is going to lead to, say, antibiotic resistance in a particular bacterial line. But you can be overwhelmingly confident that one of a relatively small number of possible sequences that confers immunity will arise, and you can plot out the relative likelihoods of the various sequences by running the experiment enough times.

            And the mutations in this case don’t even have to be the obvious result of Quantum indeterminacy, any more than coin toss at the start of a ball game is. I mean, yes, of course, drill down far enough and it’s all Quantum…but even coin tosses are overwhelmingly Newtonian phenomena such that you can be confident that, as of the time the players take the field, the overwhelming number of branches have the same result from the coin toss. And even branches with different coin tosses are mostly going to result in the same win / loss result for the game, and few variations in sports leaderboards are going to result in differences in political election outcomes, and none of that variation is going to change where the Sun will rise in the morning.




          2. By “rare quantum events” I refer to events that can’t be treated the way Jerry refers to point mutations, that is, inherently unpredictable in the singular but common enough that they can be predicted statistically. Even if particular point mutations are the result of indeterminable quantum events, identical point mutations will occur at particular frequencies, enough so that their exact time of occurrence will not affect evolution. Polyploidization is an example of a rare event that individually could have a major effect on the course of evolution; if it were subject to quantum indeterminacy, that’s a problem for replaying the tape, but I suppose that it isn’t.

            Imagine that some process depended on the strength of radiation coming from a 10-gram chunk of uranium oxide. That process would be highly predictable, because the radiation would be too. Now imagine a process that depended on the decay of a singe uranium atom. Not so predictable any more.

            I have difficulty supposing that drift is a process of the latter sort.

          3. Drift happens at the level of chromosomes and their segregation.

            So if we start from the position of “Everything the same down to the very last molecule, replay the tape of life from there”, then you are correct, it does not matter that much.

            However, if we relax the constraints a bit to “Broadly the same planet”, then it probably matters a lot. The human genome would look very different if the tape is replayed because of drift, and whatever intelligence evolved could be phenotypically very different, if it ever evolved.

          4. Yes. Evolution may be chaotic, with a very sensitive dependence on initial conditions, but the only source of actual indeterminacy is at the quantum level. It all depends on what “the tape is replayed” means, as Jerry pointed out.

  3. Our evolution is a one time go around. Humans could obviously make a rerun if we all disappeared tomorrow, but there will never be another Queen Elizabeth.

    Charlie Stross has picked this theme before and most recently shared some ideas of what could come next which alternatively could be viewed as what could have been possible.

    Evolution if we all disappeared

    My vote is for raccoons, big raccoons. If they succeed us, I am not entirely confident they wouldn’t wreck everything, though they might get a head start if there are no JC or Mo coons to steer them shallow.

    1. I am not confident that we will not wreck everything. So maybe we will see a repeat and how it turns out in another billion years or 2. Depends how wrecked everything is when we are done with it. And try to figure out that fossil record.

    2. I’d like to see what kind of gloves Racoons would wear because their hands are already somewhat like gloves.

  4. The probability of intelligent life evolving in the present universe is >0. The probability of intelligent life evolving a second time, independently, therefore must also be >0. If the probabilities follow a Poisson, it should be possible to get a minimum estimate of the number times ‘we,’ or something like us, has evolved. Or am I all wet?

    1. QM can beat Poisson. Uncertainty relations allow it, cf. photon anti-bunching or sub-shot noise sensitivity. But I think this feature is mostly unconnected to what you point out, except to inform us that deviations to Poisson can exist.

  5. Do we mean starting the Earth over and over again with every molecule and particle in the same position, and then seeing what happens?

    But to achieve this you would have to *know* the position and velocity of every molecule and particle yet Heisenberg’s Uncertainty Principle says this is impossible. You would have to use supernatural powers to restart the Earth – in which case all bets are off, anything is possible.

      1. Regrettably (life would be much more straight forward) answers to hypothetical questions are worthless unless you can validate them empirically.

  6. By coincidence there’s an article on historical contingency in general and then applied to human history in the latest Philosophy of Science. The article (written a fairly well known philosopher of biology) makes the (correct) point that Gould didn’t want to address questions of “universal determinism” and but rather of “small differences make big differences”, which is correct. How correct, I don’t know how anyone would know exactly, but it does seem plausible that the initial conditions under discussion would have to involve not just the entire earth but perhaps the solar system and beyond, if as is suspected, comets (from, say, the Oort Cloud) are important. So lots of “tiny factors”.

    1. I think I just said what you said about the small differences below and therefore agree – the question isn’t really correct and the questioner is really asking a different one.

  7. Perhaps the question is wrong. I suspect what is really being asked is, what would it take for humans like us not to evolve or even intelligence like ours with a civilization, if in our distant past a few different tweeks occurred and what would those tweeks be? Maybe when we get AI and that AI can account for every variable, we’ll know.

  8. But those mass extinctions were determined by physical laws, and are “happenstance” only in the sense that a well-informed human couldn’t predict them.

    As I’ve argued before, even asteroid impacts may be subject to quantum uncertainty, given the chaotic nature of their orbits over geologic timescales.

    And cosmologists assure us that the very existence of our Solar System is a contingent outcome of quantum events in the early universe.

    The consequences of quantum indeterminacy aren’t confined to the subatomic world. They can and do have knock-on effects at all size scales. The universal wave function is constantly branching, and those branches diverge to produce different outcomes.

    Rerunning Schrödinger’s equation from the same initial conditions doesn’t produce the same single outcome; it produces the same ensemble of divergent outcomes. Which outcome we perceive is indeed a matter of happenstance.

    So to me, saying that the history of life on Earth is fully deterministic, with one caveat, gets things backwards. The world isn’t fundamentally classical, with a soupçon of quantum indeterminacy thrown in. The world is fundamentally quantum mechanical, and the appearance of classical determinism is in some sense an illusion resulting from our narrow perspective on one branch of the wave function.

    1. The only addendum I’d offer is that most macroscopic-scale phenomenon are much more stable to random fluctuations than most people tend to realize.

      No amount of Quantum weirdness of any flavor you might propose is going to change Jupiter’s orbital characteristics tomorrow by any measurable amount. And the same goes for Jupiter’s orbital characteristics centuries out. Yes, a billion years out there could well be some significant divergence…but, even then, it’s going to be clusters of different stable configurations rather than one giant smear of random stuff. And, even a billion years out, Jupiter is not going to spontaneously transform via Quantum indeterminacy into green cheese.

      If you want your own visceral, personal experience of this kind of stability, go find one of those online Quantum coin flippers that use an expensive piece of equipment to make some microscopic measurement that’s fundamentally indeterministic. Put two cookies on the table in front of you and commit yourself to eating the one on the left if the coin comes up “heads” and the one on the right if it comes up “tails.”

      And follow through with it. Congratulations, you’ve now just amplified a microscopic Quantum effect into a very easy-to-see macroscopic change. You will have eaten both cookies, yes, but in two different branches of the universal wavefunction, with each branch only having eaten the one cookie corresponding to the toss.

      Now, fire up the flipper again. This time, the choice is between eating the remaining cookie and going on an axe murdering rampage. Flip the Quantum coin.

      What percentage of branches of the wavefunction have you eating the second cookie rather than going on an axe murdering rampage, regardless of the way the coin comes up?

      Consider now the implications that has for percentages of all sorts of other macroscopic phenomena.

      Yes, there will be examples where something so trivially minor as a single flipped bit can have profound consequences. But, in such cases, the whole thing is so unstable that it’ll be obvious in retrospect that it could just as easily have gone the other way. For example, if you’re sharpening your axe right now as you read these words, you would have gone on that rampage sooner rather than later, regardless of the outcome of that coin toss.

      …or…Silly Season is upon us with full force. Ponder the magnitude of just how many Quantum effects would have to “randomly” happen in concert to change the outcome of the election. Even just looking at the Classical macro-scale picture, it comes down to a relatively small number of plausible scenarios of Electoral College counts — and you can be sure that not very many people will be flipping coins to determine their votes, let alone flipping Quantum coins.



      1. Ben, you’re completely missing the point of chaos theory. It has nothing to do with magnifying quantum-level indeterminacy into extremely low-probability macro indeterminacy immediately. It’s about the sensitivity of a system to small perturbations in initial conditions. It’s about how a tiny change today can affect an election in 10 million years, not the election in a months’ time.

        Quantum indeterminacy is almost a red herring.

        “Small differences in initial conditions (such as those due to rounding errors in numerical computation) yield widely diverging outcomes for such dynamical systems, rendering long-term prediction impossible in general.[2] This happens even though these systems are deterministic, meaning that their future behavior is fully determined by their initial conditions, with no random elements involved.[3] In other words, the deterministic nature of these systems does not make them predictable.[4][5] This behavior is known as deterministic chaos, or simply chaos. The theory was summarized by Edward Lorenz as:[6]
        Chaos: When the present determines the future, but the approximate present does not approximately determine the future.”

        1. It’s about how a tiny change today can affect an election in 10 million years, not the election in a months’ time.

          Cf A Sound of Thunder by Ray Bradbury. If you know this one, try substituting “Trump” for “Deutscher”.

        2. The question is what (relative to the problem at hand) counts as a sufficiently large perturbation.

          This is, ignoring the “roulette wheel” problem, the same as our discussions about quantum effects in the brain – except a few orders of magnitude even larger. (But drier and at lower temperature, so …)

  9. Given that South America produced a sabre-toothed Sparassodontid, Thylacosmilus, & then the gorilla-like lemur, Archaeoindris in Madagascar, I would say that convergent evolution means that circumstances are likely to produce similar creatures in similar niches, & why should the human niche be any different? There are good reasons to think that the human niche is different however, but I am sure we can all think of those things.

    Who knows?!

    1. “similar creatures in similar niches” is very broad. A kangaroo is about the same size and occupies a similar ecological niche as a deer. Sure, assuming quantum indeterminacy is significant I would still expect that any rewind of the world back, say, 20 or 30 million years would produce a 100-200 pound omnivore flexible enough to live in forests, grasslands, etc. Would that critter always descend from primates? Would it always be bipedal? My guess is probably not.

      1. It is very broad because these creatures differ in many ways as well. I agree, I am just trying to say that humans are not special.

    2. There are no marsupial primates, the elephant’s trunk evolved only once, and feathers evolved only once. For every example of convergence I can give you an example of something that could have evolved multiples times but didn’t

  10. And we must not forget the possibility that superdeterminism is true; superdeterminists do not recognize the existence of genuine chances or possibilities anywhere in the cosmos (wikipedia).

    It’s nicely explained by John Bell:

    “There is a way to escape the inference of superluminal speeds and spooky action at a distance. But it involves absolute determinism in the universe, the complete absence of free will. Suppose the world is super-deterministic, with not just inanimate nature running on behind-the-scenes clockwork, but with our behavior, including our belief that we are free to choose to do one experiment rather than another, absolutely predetermined, including the “decision” by the experimenter to carry out one set of measurements rather than another, the difficulty disappears. There is no need for a faster than light signal to tell particle A what measurement has been carried out on particle B, because the universe, including particle A, already “knows” what that measurement, and its outcome, will be.”

    Superdeterminism is not very popular.

    1. Superdeterminism is not very popular.

      And, yet, if you pay close attention to your mind and its functioning, you will discover that you most emphatically are not the author of your own thoughts nor director of your own actions. Rather, it’s all reactionary stimulus / response all the way down.

      Just consider for a moment the most recent thought that’s popped into your head — probably, righteous indignation at the thought that you don’t have any control over your thoughts. Where did that thought come from? Follow the chain and you’ll pretty quickly discover it was a reaction to something you just read.

      The next time you’re brushing your teeth or driving or doing some other mundane task, take a moment to reflect on just how much conscious control you really have over what your body is doing, and you’ll discover that you don’t actually have any conscious control over anything.

      Rather, all your consciousness is is a running commentary. Sometimes you get a commentary on the commentary, which is what we call self-awareness and consciousness. But even that’s just more stimulus / response.




    2. In fact, Bell’s explanation (and that or Wikipedia) are quite confusing, because the superdeterminism loophole is not really anything to do with free will in the sense that we usually discuss it here. It relates to the fundamental assumptions of statistical methods – random sampling and independence. When superdeterminism hypothesizes that the experimenter does not have “free will”, that just means that maybe the data gathered in an experiment are not really “freely” in the sense of RANDOMLY sampled, so that any statistical result derived from those data (e.g. Bell’s inequality) is invalid. It requires something that amounts to a “conspiracy” at the level of the entire universe, and if true it does not just invalidate the Bell’s Inequality experiment,s it invalidates pretty much all of science. It’s pretty much axiomatic that we can sample randomly, and if we can’t then all statistical inference of general laws about the universe is invalid.

      1. The fundamental question would be if the underlying process were as randomly chaotic as the digits of π appear to be or merely as complex as, say, the decimal expansion of 1/17. Superficially, the first dozen digits of each seem to be equally randomly chaotic and well-distributed — but the orderliness of the one becomes apparent very soon after.

        Or, instead, consider that you’re working with dice. You could roll them enough times and come up with the expected distributions. But you could also set them out in all possible combinations for just one complete sequence and get the same distributions. Now, consider a fractal setting-out of the dice…for each of the initial outcomes, put out another complete set, and so on. You get a rapidly-exploding number of results, with every single possibility realized.

        Now, imagine playing out such a process for thousands of times, and pick any random terminal result. Trace its hierarchy, and you’ll effectively get a random sampling that again reproduces the expected distributions. Yes, there will be outliers, with the most obvious examples being the ones on the ends where there are consecutive sequences of snake eyes and boxcars…but there’s exactly one of each amongst the umpteen brazilian others. And not a single instance where the total was 0 or 13.

        Entirely deterministic, all possibilities realized, and yet still with the full appearance of stochastic randomness, down to the nice Bell curves.




      2. Agreed. Superdeterminism is in the same category of “explanation” as “God did it”. It essentially abandons any hope of understanding why we observe the particular outcomes that we do, and simply asserts that things happen that way because that’s how it’s written in some cosmic script to which we have no access.

          1. Interesting, and I’d be curious to see how it turns out. But note that near the end of the post Hossenfelder admits that superdeterminism in general is unfalsifiable (“you can never rule out generic superdeterminism anyway, so why even bother”), which is exactly the category I think it’s in.

            Also note that in the comments, vmarko argues that it may be unverifiable as well, since the truth of superdeterminism would imply that experimental results are always subject to undetectable bias and therefore untrustworthy.

  11. Surely this is only an interesting mind game
    How can we say whether the tape of life if rerun could or could not produce the same result since we have no “outside” comparison that it could have run any differently.
    Only by the study of the evolution of creatures not of this earth compared with our own would even begin to answer this question.
    If we insist on saying point mutations, chance events etc would not be exactly repeated and therefore would inevitably have greatly different results we may be giving more credence than it deserves to the idea that life, evolution and us are so incredibly unlikely that the universe must have been created especially for us.

    1. I would say that experiments like Lenski’s make it less of a “only a mind game.” I’ll confess though that I haven’t followed his research beyond the 15 minutes of fame he had a few years back. So I don’t know whether his research supports the “the same basic thing happens every rewind” idea or the “every rewind is totally different” idea.

  12. This whole thing has been extremely helpful. I have thought about this question off and on for some time, and at last i know the best answer to this question is ‘I don’t know’.
    I am being serious. This really was helpful.

  13. “Under all other scenarios, including indeterminate mutations, the answer to the question “Was the evolution of humans inevitable?” must be “How the hell do I know?” I would add the word “Yet” holy books not withstanding, “Yet” is always the key word in any discussion about knowing.
    Of course, with the sixth great extinction well underway we may not be knowing anything about anything shortly.

  14. I think that, some time around now (give or take a few thousand millennia), ‘intelligent’ life would have evolved, probably up to our level of technical sophistication. (The scare quotes are intentional, have y’all seen reality TV recently?)

    But it might not have the same number of arms or legs (or eyes) as us.


  15. “Ponder the magnitude of just how many Quantum effects would have to “randomly” happen in concert to change the outcome of the election.”

    We don’t even know for sure who the candidates will be on election day. A simple ordinary unexpected car accident could change things significantly at any time.

    Who thinks that the car accidents that will occur on Oct 28, for instance, are already entirely determined? Or whose backyard will be getting rain that afternoon, while we’re at it? Not me.

  16. I think a lot depends on where ‘The Great Filter’ is. (See Tim Urban’s ), about Fermi’s “Where is everybody?”.
    If the Origin of life is improbable, a rewinding would not be likely to give us life, hence no ‘intelligent’ life, same if the filter is the symbiosis giving the complex eukaryotic cell. After that, I do not see clear filters in our evolutionary history.
    If the ‘Filter’ is still ahead of us, the evolution of intelligent life appears more probable.
    There is much convergent evolution (say e.g. camera-type eyes), although there are things that evolved only once (as Jerry pointed out) e.g. the eukaryotic cell. Still, since there is much ‘intelligence’ around, clearly from less intelligent ancestors, I think we have a good chance of intelligent life evolving in that case.

  17. Given all that i have read above and a lot too hot for my head, i also read that “there is no law of nature that says complexity necessary develops as systems evolve from low entropy to high entropy, but it can…”
    The Big Picture. Sean Carroll
    high entropy is what is required for evolution if my understanding is correct so in line with the post, we’re frickin lucky, if you want to call humans this, intelligent life.

  18. Hmm. Everything predetermined. Interesting idea.

    Is it supported by a solid combination of data and model or only by reasoning? If the later, it smells like religious belief.

    That events often surprise us is evidence against at least the idea that our models are adequate, may be evidence against predetermination.

  19. If everything under the Laws of Nature is determined, there ain’t nothing we can do about it, so lay back and enjoy the show, while you can.

  20. Under all other scenarios, including indeterminate mutations, the answer to the question “Was the evolution of humans inevitable?” must be “How the hell do I know?”

    They way to know that is by modeling outcomes. A fascinating example is the 3/2 orbital resonance of Mercury, which is near enough to the Sun to rather become tidal locked as Venus is close to be. As it turns out, there was a 30 % chance of a 3;2 resonance. So our system is not “unlikely” against a 5 % (2 sigma) outcome of individuals in a population experiment. (C.f. quality limits in biology and medicine.)

    We already know the outcome of this experiment. Populations of organisms have populated very different niches for 4 billion years. Only one lineage has qualified as “human”. Humanity in some form or other is an astronomically rare result.

Leave a Comment

Your email address will not be published. Required fields are marked *