Evolution Under Fixed Constraints; Or, Why Random Changes Don’t Cut It

Not actual human beings.
Not actual human beings.

See if this analogy makes sense. A ramp onto which you loose a ball, which can roll down and fall into one of three slots. The configuration is such that the ball must go into one of these slots. It cannot get stuck or jump off the ramp. Suppose, like a pachinko machine, the ramp is covered in plastic to prevent leakage.

Now something will cause that ball to go into one of the three slots, but suppose you’re no expert in physics and can’t figure out the equations of motion. So you don’t know the cause.

But can you agree with me that the ball must go into one of the three slots? As a matter of logic, or information, given the setup, the probability of any slot is 1/3. The probability didn’t cause the ball to end up anywhere; chance isn’t a cause. That the ball could potentially be in one of the slots also wasn’t a cause.

Randomness isn’t a cause, either. Neither “chance.” The ball did not fall “randomly” in any causal sense. It fell because of the physics; our ignorance of the physics means nothing.

The ball had to go into just these three slots. Why? Well, that was the way the setup was designed.

And there it is. The “D” word.

The system from its start to its end could be said to have evolved. Evolution in action! But it could have only evolved in certain set ways. The ball could evolve only into one of three configurations. It could not have evolved into fourth slot, for one doesn’t exist. It could not have evolved into a petunia.

This is a stretch of the word evolution, however, which is why it is only an analogy. But ordinary evolution cannot be different in manner, though of course it differs in mechanism.

When DNA undergoes a mutation it must do so only in certain constrained ways. Like the ball, it must fall into prescribed “slots.” Some thing or things will cause the mutation. Nobody might know these causes, or even the range of possible “slots”. This ignorance is called “randomness”. But randomness isn’t ontological. There is nothing physical there.

These facts are why explanations of evolution are often screwy. Take this example from The Scientist:

Are mutations truly random?

Do genetic mutations really occur at random spots along the genome, as researchers have long supposed? Maybe not, according to a study published online today (January 13) in Proceedings of the Royal Society B, which proposes a mechanism for how new mutations might preferentially form around existing ones.

Truly random has no physical or biological meaning. Here it seems to be a stand in for uniform. Whether mutations are uniform along a genome, or express a preference for other sites, some thing or things still causes each one, and under some constraints. It is the duty of the biology to discover whether these causes are biased towards this or that direction. That is, what the “slots” are and their natures.

Another frequent mistake is made by those anxious to argue evolution is “unlikely”. For example, the paper The Mathematical Impossibility Of Evolution by Henry Morris, Ph.D. Morris’s argument is along the same as we met yesterday, when a fellow calculated the probability you existed was nearly impossible.

All probability—a measure of our information—is conditional on the premises supplied. Change the premises, change the probability. Morris starts with the central error of his opponents, “Mutations are random changes in genetic systems”, as if randomness is a cause, from which he concludes “Since random changes in ordered systems almost always will decrease the amount of order in those systems, nearly all mutations are harmful to the organisms which experience them.”

This is a theoretical conclusion, not one based on observation. (I recall people used to say bumblebees couldn’t fly because aerodynamical theory forbade it.) It seems to also be based on the premise that random is not only a cause but also uniform in action. That would be like saying that while we can only see three slots, theory demands there be four thousand (maybe there are slots besides the three which are too narrow, but which theory insists are options).

Just like “you exist” cannot be that impossible since you, in fact, exist, evolution cannot be impossible because it, as observations prove, exists. The nature of evolution is badly misunderstood on both sides.

We still have not succeeded purging the mysticism of randomness from probability. Frequentist statistical theory is rife with it. But then so are many interpretations of quantum physics.


I was reminded of this after popping over to Mike Flynn’s place and seeing this and this. Travel day today and tomorrow.


  1. Mike Ozanne

    “(I recall people used to say bumblebees couldn’t fly because aerodynamical theory forbade it.) ”

    It’s one of those becoming less wrong things. When the Wright brothers finally ended the development phase that put the Flyer airborne. They had discovered that pretty much the entire body of human knowledge concerning aerodynamics from aerofoil shapes and wing chords to the physical properties of air (The Stokes Number), was wrong. Not just “close enough for government work” or “subject to fine tuning” but mangled bleeding and dead in twisted wreckage wrong. Once they had improved the tech to the point where engineering design was possible, some attention was paid to the properties of wide bodied short winged fast flying animals. After the odd fatal interlude (look up the GB Racers) it was concluded that many modes of insect flight couldn’t be explained under our current knowledge of aerodynamics. Since then we’ve invented the helicopter and developed an understanding of rotary wing mechanics. We’ve also discovered the concept of the relative viscosity of air for small objects. Between these two the flight of the bumblebee is no longer an impossible thing that you can witness before breakfast, our knowledge of aerodynamics is less wrong and hopefully will become even less wrong in the future.

  2. Sander van der Wal

    A random mutation can very well be caused by a cosmic ray. The cause is there, but you do not know which bit of the DNA it is going to hit, or who’s DNA. You also do not know what exact change it is going to be. But here you can probably calculate the possible outcomes and their frequency.

    Regarding cosmic rays, humans have a builtin detector. Its the stuff on top of your head, and gray is the color that signals detection.

    For something more modern, a digital camera is sufficient.

  3. TinyCO2

    “But can you agree with me that the ball must go into one of the three slots”

    Not if it was hit by a meteorite. Not if the machine is operated so many times the fabric of the machine or the ball changes.

  4. Gary

    Maybe part of the problem is that we too frequently use the nouns “random” and “randomness” instead of using the adverb form “randomly.” That implies that it’s a thing instead of a description of how something happens. A subtle difference, but telling.

    The online Etymology Dictionary says this about the word’s origin:
    “having no definite aim or purpose,” 1650s, from at random (1560s), “at great speed” (thus, “carelessly, haphazardly”), alteration of M.E. randon “impetuosity, speed” (c.1300), from O.Fr. randon “rush, disorder, force, impetuosity,” from randir “to run fast,” from Frankish *rant “a running,” from P.Gmc.

  5. James

    What is your opinion on using an analysis such as Dr. Morris’ to cast doubt on a particular set of hypotheses?

    If a set of evidence/assumptions yields incredibly small chances for an outcome to occur, and yet we observe all kinds of that outcome, then it would appear to me to build a case against the assumptions. In this context, if assuming purely uniform mutation chances does mean that you should see much more degradation of genetic information, and yet we observe the opposite (such as a wide array of genetic information), then it makes it reasonable to say that the uniform mutation chance assumption is probably wrong.

    I think this is the point of Dr. Morris’ paper, although maybe not phrased that way. There are also many assumptions baked-in, such as what constitutes a degradation of information, etc.

  6. Sheri

    Someone forgot to tell the caterpillar that just hatched as a butterfly in my jar that it couldn’t roll outside the expected options.

  7. MattS

    Sander van der Wal,

    “A random mutation can very well be caused by a cosmic ray. The cause is there, but you do not know which bit of the DNA it is going to hit, or who’s DNA. You also do not know what exact change it is going to be.”

    The fact that we do not have enough knowledge/information to make more than a probabilistic prediction does not make something random.

  8. Sander van der Wal

    In the case of cosmic rays we will not be able to obtain the necessary information. Cosmic rays come in at the speed of light, you “cannot see them coming”. The events that generate them are far away, so monitoring these sources up close in order to study them won’t be feasable for the foreseeable future. A lot of these rays were created by sources that do not exist in the local universe (quasars, one billion light years or so) which means that the sources we can see farther away will be gone by the time we get there.

    Other sources, supernovae, can be much closer to home. Predicting which stars will go supernova is not that hard (they are rare and massive), but when it is going to is still hard.

    So, while these obstacles are being overcome, I’m sure people will be happy using a probabilistic prediction model.

  9. David L Hagen

    Your tinker toy examples are well chosen.

    Re: “how new mutations might preferentially form around existing ones.”
    Mutations preferentially forming in some regions rather than others only provides a small increase in local mutation frequency. It does not address the issue of reaching the exceedingly rare functional biological configurations by random mutations.

    For technical mathematics publications on the limitations of Darwinian searches (random mutations) “generating” ordered biologically functional configurations see publications by William Dembski and Robert J. Marks II at the Evolutionary Informatics Lab.

    Please clarify your objections to Morris distinguishing between random mutations and intelligent agents in forming ordered systems, versus mutations or “entropy” degrading existing designed or ordered systems.

    Re: “Since random changes in ordered systems almost always will decrease the amount of order in those systems, nearly all mutations are harmful to the organisms which experience them.”
    This is a scientifically accurate statement describing the observed impact of mutations on current biological “ordered” systems.

    For evidence see plant geneticist John C. Sanford:
    Genetic Entropy & The Mystery of the Genome (Mar 2008) ISBN-10: 0981631606
    and biochemist Michael J. Behe
    The Edge of Evolution: The Search for the Limits of Darwinism (Jun 5, 2007) ISBN: 0743296222

    On the complexities of genomic information vs disorder, William Gitt explores five levels of information. See: In the Beginning was Information CLV

  10. MattS:

    Your assertion that “The fact that we do not have enough knowledge/information to make more than a probabilistic prediction does not make something random” violates the principle of entropy maximization. Under this principle, the entropy of the inference to the way in which the outcome of the event will occur is maximized under mathematical constraints expressing the available information. The availability of information affects the probability value. One of the consequences is Laplace’s rule of succession: If a given outcome was observed in x of n trials of an experiment, the value of the probability of this outcome is (x + 1)/(n + 2). Given that there have been no trials, the outcome is completely random.

  11. Have any of you folks read “Darwin’s Radio” by Greg Bear? It gives an interesting (if untested) mechanism for sudden jumps in phenotype and genotype–retroviruses operating on the 40% (??) parts of DNA that are supposedly junk. Not scientifically verified but plausible I would think.

  12. Sander van der Wal

    @Bob Kurland

    Yes, I did, couple of years ago. No idea whether it is plausible, but my guess would be that it is not. The consequences of science in his novels are weird, Fantasy-like. And not the Middle Ages-rerun Fantasy. Or the Tolkien rehashes.

  13. MattS

    Sander van der Wal,

    “I’m sure people will be happy using a probabilistic prediction model.”

    Neither I nor Mr Briggs has objected to using probabilistic prediction models. All we have objected to is calling real outcomes random.

    Terry Oldberg,

    “Given that there have been no trials, the outcome is completely random.”

    No, given that, the outcome is unpredictable. That is not the same thing as saying it is random. Real world outcomes are not and can not be random. The fact that we do not and in many cases can not ever posses enough information to make deterministic predictions does not change that fact.

  14. The complexity and improbability of it all may be a little hard for humans to take, but if you think we’re improbable, what must you think would be a chance of there being a God???


  15. MattS:

    I used “random” in the sense in which it is used in statistics as reported by the Random House College Dictionary, copyright 1999. The definition is: “of or characterizing a process of selection in which each item of a set has an equal probability of being chosen.”

  16. @JimmiesRusted
    Hey, all I can think of is Mark Twain’s aphorism:
    “There are three kinds of lies: lies, damn lies and statistics!”

  17. MattS

    Terry Oldberg,

    “of or characterizing a process of selection in which each item of a set has an equal probability of being chosen.”

    This can not be true of any real physical process. As Mr. Briggs has pointed out repeatedly in his blog, this definition can only apply to our uncertainty about the outcome of the process, not the process itself.

  18. @Jimmie Rustled
    Here’s a quote from an article on “The Corner” (National Review Online–yes, I know it’s conservative) about that infamous study:(let’s see if I can do the html for blockquote):

    The New York Times and the media are widely reporting the results of a new study of 500 children raised by same-sex couples. Researchers are aghast that such a shoddy piece of work, which is based on a convenience sample of parents who volunteered to be surveyed, and which relies solely on these parents’ reports to determine child well-being, should have been published in a reputable peer-reviewed journal. The author is facing a petition and statement signed by hundreds of reputable academics questioning his moral character and scientific integrity, and the editor of the journal who published such a thing faces similar attacks and enquiries about whether the peer-review process was compromised.

    Oh wait, the study shows children of same-sex parents do better than the average child, after adjusting for income. Never mind.

  19. Bob, so same-sex couples have higher income too? Wow. Must be pretty with-it folks, huh?


  20. Jim S

    “But can you agree with me that the ball must go into one of the three slots?”

    Utter nonsense. The track or the ball will wear out (or decay to dust) after enough runs.

    Hmmmm, sounds like errors made in genetic replication….

  21. John Moore

    I fear that this discussion is a bit pedantic, at least regarding common usage. Many things are effectively random – we cannot predict them and many we cannot go back and find any specific cause. With a rigorous philosophical analysis, perhaps most of this randomness goes away (ignoring quantum physics), but that just doesn’t apply to the normal language usage where a “random” cause really just means a cause that isn’t sharply and narrowly fixed or understood.

    Anyway, regarding evolution… random mutation is one of the weaker mechanisms. Sexual gene mixing is much more powerful, giving a “reason” for the wide existence of sexual reproduction, even though it is far more expensive and more risky. Another is inter-species exchange of genetic information, which is one reason that antibiotic resistance spreads so quickly. Beyond that, there are various mechanisms which can cause large chunks of DNA to be rearranged (perhaps by a “random” cause, but with a different and potentially far more powerful effect than the single base flip of a typical random mutation).

  22. Sander van der Wal

    What happens if we replace the board by one with 10**100 instead of 3 slots? And instead of the slots being straight, deep and narrow, they are intermittently shallow and deep, they split and come together again, and have the occasional hole in them?

  23. Sander van der Wal ( 9 July 2014 at 1:57 AM):

    Thanks for posing the interesting question. In the system that you describe, once it comes to rest the ball can be in states that are of large number. Each of these states is a way in which an outcome can occur for the ball. You describe no constraints on entropy maximization. Maximization of the entropy of the inferred way in which the outcome will occur without constraints yields the conclusion that each way has the same probability value. In units of bits, the entropy is the logarithm to the base 2 of the number of ways.

  24. Nullius in Verba

    “Since random changes in ordered systems almost always will decrease the amount of order in those systems, nearly all mutations are harmful to the organisms which experience them.”

    That’s what the theory says. But of course, the appearance of design does not arise from mutation. It arises from the selective death-before-reproducing of all those organisms that are not good designs. Mutation decreases order, but natural selection increases it. It’s called “Evolution by Natural Selection”, and not “Evolution by Mutation” for a very good reason.

    Evolution is a work of sculpture; the chisel is wielded by Death.

    Or perhaps ‘topiary’ is a more accurate analogy? The scythe is an agricultural implement after all…

  25. Nullius in Verba


    Evolution and selective breeding are the same thing. People had known about artificial selection for a long time, and it had even been proposed as a possible origin of the species before Darwin’s time. It was Darwin’s insight that the natural propensity of ‘better’ designs to survive could provide this selective breeding selection naturally, without human intervention. Evolution by Natural Selection works pretty much the same way as Evolution by Artificial Selection (i.e. Selective Breeding), except that it is an animal’s own ability to survive that decides which animals get to breed.

    Breeding forms a new species when different lines get far enough apart that they can no longer interbreed. Although it can be a bit ambiguous for a while, with mules (species cross-breeds that are infertile) and ring species (where animals living as neighbours can cross-breed, but those at the far extremes of the range cannot).

  26. Sheri

    NinV: Where does genetic manipulation (GMO) fall — is is that a ball jumping the track?

  27. Nullius in Verba

    “NinV: Where does genetic manipulation (GMO) fall — is is that a ball jumping the track?”

    If I understand the analogy right, that would be looking around for slot machines where the ball is in the winning slot, detaching that ramp and slot, and attaching it to your own slot machine.

    That’s assuming you’re talking about genetic engineering, of course. Conventional breeding is a form of genetic manipulation, too, but a bit more uncontrolled and dangerous. For example, the practice of using radiation and chemical mutagens to induce mutations at a faster rate is like shaking the ramp vigorously to cause the balls to jump about, and then the cross-breeding to separate the desirable from the undesirable traits is like linking chains of ramps and slots together into a maze to indirectly steer the balls to their destinations, like one of those tilt-puzzles where you have to get all the balls into the holes just by tilting and shaking the box.

    It’s a bad analogy for evolution, because there’s nothing in it to correspond to natural selection, which is what causes the effect we’re interested in. It only describes the mutation part. But I guess Briggs was really talking about randomness, and the example chosen to illustrate it was incidental, so I’m not arguing.

  28. Sheri

    NinV: I kind of thought it’s not “standard” evolution. With all of the modifications we are making to genes, however, this could basically render natural selection no longer needed (though it will occur in some places where humans don’t touch things). Then evolution becomes not chance but choice. With all the talk of “remaking” dinosaurs and gene manipulation, evolution as chance and natural selection seems to be heading for obscurity. Not immediately, of course, but that is the direction I see this going. (Humans are also notorious for trying to eliminate natural selection, trying to save the weak and damage critters and places because they tug at heart strings. So, again, natural selection being revamped by humans. Now, if we consider humans to be part of the process, then some revisions to the theory may be needed to account for creatures that overcome the chance part.)

    I’m just thinking this through, not necessarily arguing a point. It’s something future generations will deal with.

  29. Sheri, not all genetic modification–breeding–done by humans has been for good. I’m thinking mainly of deformed canine breeds–German shepherds with sloping backs and deformed hips, shi-tzus, pekinese, and other breeds with bulging eyes, long backs and consequent spine problems….All the hip problems in big dogs.

  30. There’s a fine science-fiction series by Nancy Kress–“Beggars in Spain”, “Beggars and Choosers” and ??? that shows the unintended consequences of genetic manipulation.

  31. Sheri

    Bob K: Agreed. I have had mostly pure bred dogs and all had health problems unique to the breeds. They are adorably cute but very bad genetics. Of course, cute sells!

    Will check out the Nancy Kress series.

  32. Re GMOs. There are actually natural processes of gene transfers that are similar. Viruses for example can pull genes from one animal and transfer them to a different species. This is actually fairly common in plants, less so in animals.

Leave a Reply

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