What evidence is there for evolution being non-random?

If you really wish to show me that you just don't get it, please keep posting like this. You really don't understand what Cyborg and Myriad and I have been saying do you?

Actually, I do, and I think that you are avoiding what I am saying. You can talk all day about how the existence of stochastic models of evolution doesn't make evolution a stochastic process and how scientists who study evolution as a stochastic process are only speaking "synechochically", but you are ignoring the fact that there are scientists who refer to the entire process of evolution as a stochastic process when they are not talking about modeling. For example, Bell is very explicit in his description of evolution by natural selection:

Bell (1997) said:
In every generation better-adapted individuals will be more likely to survive and reproduce. This is only a tendency, however, not a deterministic rule. A snail living in an English hedgerow is less likely to be eaten of its shell is striped rather than plain.But it is not very likely to survive in any case; it may be eaten by a shrew, or die of heatstroke or starvation; it may even be eaten by a bird after all. Selection is a process of sampling. The variation of characters among individuals ensures that the sample that reproduces is a biased sample of the population as a whole, but its composition cannot be precisely specified in advance. But there is nobody responsible for selecting snail at the bottom of hedgerow, and no individuals, no matter how well-endowed has any guarantee of success, only a greater or lesser chance. Richard Lewontin once prefaced a lecture on this topic with a quote from Ecclesiastes: the race is not alway to the swift, nor the battle to the strong; but time and chance happen to both.

The nature of evolution as sampling implies that evolution is a stochastic process that is subject to sampling error. The composition of a population at any point in time will be determined by three factors. One is historical, the composition of the generation from which it descends. The second is selection, which tends to increase some kinds of individual and decrease others. The third is chance. The actual composition of the population will inevitably differ from what we expected based on descent and selection, because the life of each individual is a historically unique succession of events who eventual outcome is influenced by a multitude of factors. The next generation is formed in a stochastic, or probabilistic, fashion from the success and failure of many such lives. We may be able to predict its average properties with some assurance, but its composition will fluctuate to a greater or lesser extent in ways we cannot predict or account for.
(emphasis mine)

Strangely, this is exactly how the sites that the "non-randomites" promote describe evolution except they adamantly refuse to call it "random" or "stochastic" (e.g., "Evolution means that life changed 'by chance.'", But it's not random either!, Misconceptions: Life Changed by “Chance”, Misconceptions about natural selection).

Now, why do identical explanation receive opposite and contradictory descriptions?
 
Actually, I do, and I think that you are avoiding what I am saying. You can talk all day about how the existence of stochastic models of evolution doesn't make evolution a stochastic process and how scientists who study evolution as a stochastic process are only speaking "synechochically", but you are ignoring the fact that there are scientists who refer to the entire process of evolution as a stochastic process when they are not talking about modeling. For example, Bell is very explicit in his description of evolution by natural selection:

(emphasis mine)

Strangely, this is exactly how the sites that the "non-randomites" promote describe evolution except they adamantly refuse to call it "random" or "stochastic" (e.g., "Evolution means that life changed 'by chance.'", But it's not random either!, Misconceptions: Life Changed by “Chance”, Misconceptions about natural selection).

Now, why do identical explanation receive opposite and contradictory descriptions?

You haven't read a single word I have written about this have you? How you have arrived at such an entirely distorted view of what I have written is totally beyond me. I have never once not called that part of the process that we do describe using probabilities as anything but random (except to point out where the definition of non-random from mathematics contradicts the common use of that term).

But, hey, whatever floats your boat. If misunderstanding your fellow human being is your thing, then you've successfully made a new enemy.
 
You haven't read a single word I have written about this have you? How you have arrived at such an entirely distorted view of what I have written is totally beyond me. I have never once not called that part of the process that we do describe using probabilities as anything but random (except to point out where the definition of non-random from mathematics contradicts the common use of that term).

But, hey, whatever floats your boat. If misunderstanding your fellow human being is your thing, then you've successfully made a new enemy.

I'm sorry you like to think like articulett, but your repeated insistence that "adaptive optimization" is "the non-random part of evolution", regardless of whether the process itself is random*, is as misguided as it is wrong. "Adaptive optimization" in random systems is an emergent property of the mathematics that govern randomness (e.g., laws of large numbers, convergence of random variable, and (central) limit theorems). In other words, "adaptive optimization" is as random as the mutation, drift, migration, and selection processes that create it, and it is the properties of probability theory and statistics, which are also in and of themselves random, that require the emergence of apparent non-randomness.

*Note: I am using "random" only in the sense "[o]f or relating to a type of circumstance or event that is described by a probability distribution" and no other sense.
 
This thread probably ran out of usefulness about 50 pages ago. I;, like almost every other long term participant, am just repeating myself with different words. If my self control holds, this will be my last post to this thread. However, I cannot resist one last comment.


I have no reason to doubt the truthfulness of anyone who has participated here.
 
"Adaptive optimization" in random systems is an emergent property of the mathematics that govern randomness


Yes, that's the whole friggin' point. It is an emergent property, and as such is something different from the its components. Wetness is not precisely predictable from a single water molecule.

In other words, "adaptive optimization" is as random as the mutation, drift, migration, and selection processes that create it, and it is the properties of probability theory and statistics, which are also in and of themselves random, that require the emergence of apparent non-randomness.

No. And this is the point that you are missing. If the adaptive optimization is random, then show me the range of optimizations. There must a range, a probabilistic range, of optimizations. But, if there is, then we don't have an optimization. That is the whole friggin' point, once again. Probability theory is useful for discussing some of the nuts and bolts of this process because we don't have all the necessary information to describe it in any other useful way. It is an imprecise way of describing an abstraction that we have created in our minds in the first place. As far as nature is concerned, there is no such thing as 'evolution'. There are organisms surviving and dying and reproducing or not reproducing. All of that is completely determined by the physical forces in nature. We abstract this thing we call 'evolution' and describe how certain groups live and die by using our crude approximations that we call probability theory. The output is one thing. Optimization to an environment.

Frankly I think Meadmaker is right. There is no sense in discussing this further.
 
Ichneumonwasp-

I'm sorry but you entered the discussion with the philosophical presupposition that "nothing is truly random", and that is what I think is our sticking point. I'd be willing to discuss the possibility that evolution could be non-random, but it is seems doomed to failure if the underlying assumption is that nothing is truly random.

On a completely different note, I would like to apologize to you if the thing that I have said implied that you were diametrically opposed to my approach in every respect. Having gone over you posts in the thread, I now see that you have cautiously agreed with me several times. Again, I am sorry, for being unnecessarily confrontational.
 
Ichneumonwasp-

I'm sorry but you entered the discussion with the philosophical presupposition that "nothing is truly random", and that is what I think is our sticking point. I'd be willing to discuss the possibility that evolution could be non-random, but it is seems doomed to failure if the underlying assumption is that nothing is truly random.

On a completely different note, I would like to apologize to you if the thing that I have said implied that you were diametrically opposed to my approach in every respect. Having gone over you posts in the thread, I now see that you have cautiously agreed with me several times. Again, I am sorry, for being unnecessarily confrontational.

How about this post? Is it random or non-random? Would it make any sense to describe it as "random"? If so, what would that add to our knowledge?
 
How about this post? Is it random or non-random? Would it make any sense to describe it as "random"? If so, what would that add to our knowledge?

Apparently if you are trying to imply that "nothing is truly random" then your conversation with mijopaalmac is doomed to failure.

If you are not going to allow him to assert that "there is no evidence for evolution being non-random" and define that garble as he sees fit-- then you are just not able to comprehend his depth when he spins out more nothingness in the name of scientific rigor or something like that. (He's the one who gets to ask the leading questions that imply the answer--not you, silly.)

And thanks for quoting him so I can add, that I do not think Ichneumonwasp-
"entered the discussion with the philosophical presupposition that 'nothing is truly random'", as Mijo alleges-- but merely to answer the OP question as many have tried to do not realizing that Mijo had the answer and the answer meant that he would call anything random that was related to probability in any way.
 
Last edited:
Apparently if you are trying to imply that "nothing is truly random" then your conversation with mijopaalmac is doomed to failure.

If you are not going to allow him to assert that "there is no evidence for evolution being non-random" and define that garble as he sees fit-- then you are just not able to comprehend his depth when he spins out more nothingness in the name of scientific rigor or something like that. (He's the one who gets to ask the leading questions that imply the answer--not you, silly.)

Yeah, but you're evil and people don't like you...

...so, since everything in the universe seems to contain random elements, nothing can ever be said to be non-random. Not a single thing, ever.

You know, when people insist on defining and applying terms in such a way that they can easily stretch to cover everything and anything, I see their argument as being basically meaningless. Once you expand a term to mean everything, they in cannot be used to explain anything.

Saying that anything with random elements cannot be called "non-random" is really saying that everything is random. Once you say that, then the word itself is useless, and serves only to obscure the issue and make meaningful discussion impossible. Certainly, that seems to be the ultimate goal of our creationist "friends".
 
Cyborg, are you arguing that humanity was inevitible?

No - ir's more complex than that - but that does not make the only viable alternative that it must be a 'random' occurrence

Think about a program hat takes an input. Essentially this equates to a collection of funcions, each one of which would have an 'implicit' input. You could see the inevitable outcomes therefore as a function of the input.

Now the output on such a program could easily be be determined by a 'seed' and it could be devised in such a way as that it appears probabilistic. The problems come if you cannot access that 'seed'...
 
On a completely different note, I would like to apologize to you if the thing that I have said implied that you were diametrically opposed to my approach in every respect. Having gone over you posts in the thread, I now see that you have cautiously agreed with me several times. Again, I am sorry, for being unnecessarily confrontational.

Thank you. I think I understand your reaction because of the way the thread has evolved and the way that you have been attacked.

I'm sorry but you entered the discussion with the philosophical presupposition that "nothing is truly random", and that is what I think is our sticking point. I'd be willing to discuss the possibility that evolution could be non-random, but it is seems doomed to failure if the underlying assumption is that nothing is truly random.

Actually I don't make that presupposition. I generally precede comments concerning determinism with "if" statements. Sometimes I leave them out, though, which I realize can cause confusion. But what we see in nature appears determined. From what I understand, the general consensus is that the quantum indeterminacy at lower levels is "washed out" in what we see at higher levels so that we can speak of a determined universe (not at a fundamental level but at an operational one). I am fully aware that the very idea of determinism is possibly wrong, but at the level where organisms interact it seems to be THE force.

Ultimately it doesn't matter for this discussion, though, for a couple reasons.

I fully grant the description we use to discuss the abstraction of evolutionary change as random. But that description is a description. There is nothing fundamental about it. We often mistake the mathematics for the "thing itself" -- a practice that Edmund Husserl warned against around the turn of the century. The fundamental level is still one organism fighting for survival. The abstraction we create of many organisms we can only speak of in probabilistic terms. To talk of the nuts and bolts of this process we absolutely need that description. In fact, I think we are relatively handicapped without it. But I think it is important for us all to realize that it is a "mid-level" description (if that makes sense) -- not the "thing in itself" of the organisms or the ultimate emergent property we call 'evolution'. If, however, you want to argue that indeterminacy (due to quantum fluctuations) impacts the evolutionary process, then you must admit that it impacts everything, so everything is random and the only non-random elements are mathematical abstractions. If everything is random, then it doesn't make sense to speak of evolution as random. It is simply part of everything.

And when it comes to the emergent property of 'evolution', well this is just one more example that shows that the emergent property is not the components that make it up. Just as the random interactions of water molecules produce wetness. We speak of the water molecules as random. We don't speak of wetness as random. We speak of the movements of gas molecules in a container as random but do not speak of their aggregate activity -- the emergent property of pressure -- as random.

Lastly, the whole idea of using 'random' in this discussion, as I have said before, doesn't add that much when you really look at the process. In fact, it is almost a truism since we cannot even speak of the nuts and bolts of evolution without recourse to variation, which is described in probabilistic terms. But, again, that is part of the abstraction and not the "things in themselves".

This is really just a long-winded way of me saying -- drop the word fundamental from your descriptions and I think we are mostly on the same page. It may just be me, but fundamental (to me) implies a discussion of the basic, bottom-level elements of the process. The really basic elements of the process are probably determined (one organism fighting for survival). There is no way for us to know for sure unfortunately but that is how it looks. The abstraction (many organisms competing with variability being a core component), however, only makes sense in descriptions that employ probability. I fully agree (and always have) that a description of evolutionary processes at the level of discussing species survival and change - how Darwin described it - requires the concept of 'random' (mathematical definition) because one of the central tenets is variability. That is why the literature is replete with such descriptions.

Do those distinctions make sense?
 
A quick response to Cyborg's post #2601.

That is the essential difference between our positions, I think.

I am saying that it isn't just that we can't access the seed value, which would be a pseudorandom system.

It is my contention that some of the seed values are truly random, quantum effects being magnified nonlinearly to affect e.g. weather (but not climate). A probabilistic treatment allows you to still produce meaningful analyses in such a situation.

I think this is useful. The addition of the qualifying phrase "tends to" to Articulet's statements does not seem to be much of a penalty for a gain in accuracy.
 
Last edited:
Actually I don't make that presupposition. I generally precede comments concerning determinism with "if" statements. Sometimes I leave them out, though, which I realize can cause confusion. But what we see in nature appears determined. From what I understand, the general consensus is that the quantum indeterminacy at lower levels is "washed out" in what we see at higher levels so that we can speak of a determined universe (not at a fundamental level but at an operational one). I am fully aware that the very idea of determinism is possibly wrong, but at the level where organisms interact it seems to be THE force.
See my previous point about weather and climate. I would say that the adaptative traits are to the climate, but it is weather events that actually perform the selection, and I would argue that these are truly probabilistic.
Ultimately it doesn't matter for this discussion, though, for a couple reasons.

I fully grant the description we use to discuss the abstraction of evolutionary change as random. But that description is a description. There is nothing fundamental about it. We often mistake the mathematics for the "thing itself" -- a practice that Edmund Husserl warned against around the turn of the century. The fundamental level is still one organism fighting for survival. The abstraction we create of many organisms we can only speak of in probabilistic terms. To talk of the nuts and bolts of this process we absolutely need that description. In fact, I think we are relatively handicapped without it. But I think it is important for us all to realize that it is a "mid-level" description (if that makes sense) -- not the "thing in itself" of the organisms or the ultimate emergent property we call 'evolution'. If, however, you want to argue that indeterminacy (due to quantum fluctuations) impacts the evolutionary process, then you must admit that it impacts everything, so everything is random and the only non-random elements are mathematical abstractions. If everything is random, then it doesn't make sense to speak of evolution as random. It is simply part of everything.

And when it comes to the emergent property of 'evolution', well this is just one more example that shows that the emergent property is not the components that make it up. Just as the random interactions of water molecules produce wetness. We speak of the water molecules as random. We don't speak of wetness as random. We speak of the movements of gas molecules in a container as random but do not speak of their aggregate activity -- the emergent property of pressure -- as random.

Lastly, the whole idea of using 'random' in this discussion, as I have said before, doesn't add that much when you really look at the process. In fact, it is almost a truism since we cannot even speak of the nuts and bolts of evolution without recourse to variation, which is described in probabilistic terms. But, again, that is part of the abstraction and not the "things in themselves".

This is really just a long-winded way of me saying -- drop the word fundamental from your descriptions and I think we are mostly on the same page. It may just be me, but fundamental (to me) implies a discussion of the basic, bottom-level elements of the process. The really basic elements of the process are probably determined (one organism fighting for survival). There is no way for us to know for sure unfortunately but that is how it looks. The abstraction (many organisms competing with variability being a core component), however, only makes sense in descriptions that employ probability. I fully agree (and always have) that a description of evolutionary processes at the level of discussing species survival and change - how Darwin described it - requires the concept of 'random' (mathematical definition) because one of the central tenets is variability. That is why the literature is replete with such descriptions.

Do those distinctions make sense?

Yes
Articulett said:
If the question was "was humanity inevitable?" then we'd be having a different question with different words, and I wouldn't get involved in such a fatuous discussion. Whether a betting man could have predicted vertebrates has nothing to do with the non-random aspects of evolution nor the OP. What you think of as valuable or useful descriptors has nothing to do with what actually is useful or valuable... nor does it answer the OP. Many have answered the OP with the notion that Natural Selection is NOT random and the key to really understanding evolution. If you and your buddies don't like that explanation try to get something published in peer review, because all you seem to be doing is confusing yourself and others and pretending to understand something that you sound very clueless about. If you want to believe that there is nothing nonrandom about evolution, be my guest. But consider that maybe you are missing something and you might be poorly qualified like Behe in explaining the concept to others.

But the liklyhood of particular optimisations occuring can be inferred form the fosssil record. Eyes, all over the place, ditto flight.

Even assumoing that only placental mammals were capable of evolving into technological, social animals, that still leaves only one occurance in about 30-odd MY. I would say that this was because the liklyhood of that particular evolutionary niche being filled, within any defined timeframe, was lower than the niche of large herding herbivorous animals.

If particular ecological nihes being filled are probabilistic in nature, then I feel happy in saying evolution is random.

That optimisation occurs is inevitible, what optimisation is not.

I have not seen any treatment of natural selection that I would not call probabilistic, including Dawkins.

How else do you account for a quantification of a selective advantage?
 
See my previous point about weather and climate. I would say that the adaptative traits are to the climate, but it is weather events that actually perform the selection, and I would argue that these are truly probabilistic.

Entirely possible, but not something that we could say we know. We don't really know to what extent the truly random inputs of quantum indeterminacy could or have impacted this process. It is very, very likely that at early stages, as the molecular machineries were being assembled, that there was some, if not considerable, influence. At the level of description today, however, I think it is much less likely.


Adding 'tends to', I think, is the proper approach. We don't (and can't) know all the inputs into these very complex systems.


That optimisation occurs is inevitible, what optimisation is not.

Yep*, that was my point above which was only a less elegant re-statement of Myriad's excellent post.

ETA* Actually I think I overstated the above. If we had more knowledge than we actually do, we should be able to predict with reasonable accuracy what optimizations will occur. We know with resonable accuracy not just that pressure arises from interacting gas particles but what sort of pressure we will see with a particular gas at a particular temperature in a particular volume. If we knew the equations that described the optimization process we should be able to predict, just as with gas pressures, what optimizations will occur.

I have not seen any treatment of natural selection that I would not call probabilistic, including Dawkins.

How else do you account for a quantification of a selective advantage?


Yep. The nuts and bolts at that level of description requires probabilistic language (and I, for one, still prefer the word 'probabilistic' simply because it has not been misused in the way that 'random' has been in the past). At the far ends of spectrum, though, -- that optimization is inevitable on the grand scale and that we are really only speaking of organisms that are essentially working in deterministic frameworks (probably) -- we can use deterministic language.

The problem as I see it in this discussion (which is similar to discussions I have seen about free will) is that different camps are describing different aspects of the issue as a whole. That is why I mentioned forests and trees.
 
Last edited:
I was going the fact that the Central Limit Theorem says the Normal (or Gaussian) distribution has the form:

[latex] $N(\mu,\frac{\sigma}{\sqrt{n}})$[/latex].

Therefore as the sample size [latex] $n$[/latex] increases the standard deviation approaches 0. This may be a result of the Law of Large Numbers, but it is also obviously intimately connected to the Central Limit Theorem.

Well, the Central Limit Theorem (CLT) is a much stronger result than the Law of Large Numbers (LLN). Imagine that we have [latex]$n$[/latex] independent and identically-distributed samples [latex]$X_1,\, ...,\, X_n$[/latex] from some probability density function [latex]$f(x)$[/latex].

The LLN then states that
[latex]$\frac{1}{n}\left(X_1\, +\, ...\, +\, X_n\right)\, \rightarrow\, \mu$[/latex], where [latex]$\mu$[/latex] is the mean of the pdf [latex]$f(x)$[/latex].

The CLT, on the other hand, states that if we take the quantity
[latex]$Z\, =\, \frac{\sum_{i=1}^n X_i\, -\, n\mu}{\sqrt{n}\, \sigma}$[/latex], where [latex]$\sigma$[/latex] is the standard deviation of [latex]$f(x)$[/latex], then the pdf of [latex]$Z$[/latex] converges to the standard Gaussian distribution:
[latex]$\frac{1}{\sqrt{2\pi}}\, e^{-x^2/2}$[/latex].
(This is the most common way that the CLT is stated, although it is equivalent to the way you've stated it.)

It is true that the CLT implies the LLN. However, there are more auxiliary assumptions that must be satisfied for the CLT to hold than for the LLN to hold, and so when making the kind of point you are making, most people refer to the more generally-applicable LLN rather than the CLT.
 

Back
Top Bottom