Randomness in Evolution: Valid and Invalid Usage

[sol invictus]

I was under the impression that those random fluctuations averaged out at our scale ?

Right - they do. That's why some "random" processes are actually highly predictable.

Are you arguing that "deterministic" should mean "predictable" or am I misunderstanding you ?

The word "deterministic" is usually applied to mathematical models for the world in which every interaction is precisely predictable given exact knowledge of the initial conditions. However such models are often chaotic and therefore totally unpredictable if you don't have an infinitely big computer with infinite precision and infinitely precise knowledge of the initial conditions. Since you never do, such systems are never predictable, despite being technically deterministic.

So "random" can be predictable, and "deterministic" can be unpredictable. That's why you cannot simply say "evolution is random" and then stop - it's just not that simple, and the statement is either meaningless (but correct using a naive definition of "random") or simply wrong.

Why are we still discussing this? Does anyone (other than mijo) actually disagree with anything I'm saying here?

 

[Earthborn]

That is precisely the point I was making above - many "truly random" processes are highly predictable

So when you said "Random means unpredictable" you meant to say "Random means highly predictable" ?

 

[Wowbagger]

I figured it was time I chimed in here. I have NOT had time to read every post in here, since my last post in here. But, I think it is time we tried to establish some common ground, and see if we can spot where our disagreements stem from.

First question: Do we all agree that not all definitions and usages of "Random" apply to describing and studying Evolution? We might not all agree on which ones are which, but at least we can agree that, in general, there are some that do, and some that do not?

Good.

Now, here's the key challenge: We should try to establish some way to figure out which definitions are valid and invalid, for Evolution, that we can all agree on, and then see how each definition and usage stacks up against those standards.

My suggested method is simple: The scientific method. Mathematical constructs are good and all, but it is really in the testing and observing that we find out if things are useful for obtaining scientific knowledge or not.

When arguing about these matters, ask yourself: How does this help us gain more knowledge about life forms?

Unless someone thinks they have a better idea.

 

[sol invictus]

So when you said "Random means unpredictable" you meant to say "Random means highly predictable" ?

*Sigh*. No, that's not what I meant and not what I said.

I really don't see how to re-phrase what I already said in any way that will make it easier for you to understand, so I'm not going to try. If you're actually interested in this topic rather than just trolling, I suggest you re-read my posts and see if there's anything you actually don't understand.

The world is an interesting and complex place, and the words we use to describe it are often imprecise. Sorry about that.

 

[zosima]

man I had the best post on this and I thought I hit the post button before I left work, I hope its still in my browser tomorrow.

So when you said "Random means unpredictable" you meant to say "Random means highly predictable" ?

The short answer is that while any n+1th value of a random variable must be unpredictable, by definition, but the statistics of a random variable are very regular, and insofar as they serve as input to an algorithm they can produce very regular results. A moment of a particle distribution like temperature, pressure, density, velocity, is really a statistical statement.

But I think the confusion here results more from the conflation of a random variable and a random variable dependent process or an algorithm that takes a random variable as an input. These sorts of things are part of the probabilistic polynomial complexity class in computer science, and the algorithms that are studied, taken to the limit, are deterministic.

 

[Belz...]

The word "deterministic" is usually applied to mathematical models for the world in which every interaction is precisely predictable given exact knowledge of the initial conditions. However such models are often chaotic and therefore totally unpredictable if you don't have an infinitely big computer with infinite precision and infinitely precise knowledge of the initial conditions. Since you never do, such systems are never predictable, despite being technically deterministic.

Thanks. I think I understand what you're getting at.

 

[zosima]

Here's the rad post I forgot to post.

*Sigh*. No, that's not what I meant and not what I said.

I really don't see how to re-phrase what I already said in any way that will make it easier for you to understand, so I'm not going to try. If you're actually interested in this topic rather than just trolling, I suggest you re-read my posts and see if there's anything you actually don't understand.

The world is an interesting and complex place, and the words we use to describe it are often imprecise. Sorry about that.

I think the way to say it is statistics of random variable are highly predictable. In computer science, random variables have uniform distribution, and this is imho the purest definition of random. In science random variables often
follow other distributions(Gaussian, Boltzmann). This is a slight deviation from truly random, because we can say with confidence greater(or less) than the_length_of_some_interval/the_range_of_our_variable whether a variable will fall in some interval. That said, since we cannot predict the exact value any better than in a uniform distribution, this too is a very good definition of random.

Much of this discussion is conflating random variables and processes that take a random variable as an input. Such a process can take a random variable as an input and still be highly predictable. Chaotic processes are processes for which this is not the case, but these processes are not the norm.

As to whether the path of evolution is random, well thats tricky, and I have a bunch of conflicting thoughts on the issue. One thing is for sure, not all phenotypical changes are equally likely, this means that random variable serving as input to the process has a non-uniform distribution, so it isn't random in the broadest possible sense. (If the change requires two mutations, it will be much less probably than one)

 

[jimbob]

Indeed wowbagger.

I would define an outcome as "random" if identical starting conditions could lead to significantly different outcomes. I think this would mean that any cahotic physical system would be random by this description, as the infinite precision would mean that quantum fluctuations would eventually have a significant difference, in a way that isn't the case for non-chaotic systems. ETA: Sol, is that still the understanding? IIRC it was when I was an undergrad, but I tended to concentrate on solid-state physics...

I agree that there are situations where, despite the inherently random nature of the process, there are aspects that are highly predictable.

Talking about the process, mutation is essentially "undirected" with respect to the direction of evolution. It is needed as a source of variation, but the "clever" part is natural selection.

I have posted many times before why I think natural selection is a random process, in a similar way as a bent game of dice is random. Basically traits modulate the chances of reproducing, but for any individual trait, the odds are still stacked against it surviving the first generation. I don't just believe that there are chaotic systems involved that affect natural selection, e.g. the weather, but that these chaotic systems are also affetced by truly random quantum events.

Furthermore I would say that as organisms affect the selective environment for other organisms in their ecosystem, then there are complex feedback loops in place which depend on random events. To me this implies that if you were able to have universes with identical starting conditions, the course of evolution in each "Earth" would be different. Looking at the diversity in life just tens of millions of years after each large extinction event, and the differences between each one, I would say that not only was humanity just a lucky accident, which only seems special to us, but that it was only pretty recently when it became inevitable that our ecological niche would be filled (Neanderthals being an obvious alternative candidate tjhat could fill "our" niche).


If there is not any inevitibility about what would evolve, and indeed whether some of the less probable* niches are filled, then I would say that it is perfectly valid to talk about the results evolution as being "not-predetermined"**.


I think the random, yet somewhat predictable nature of natural selection is important as I have heard some people (not sure if they were sincere or creatinists) asking whether all animals that "survive" are "fitter" than all that don't. My answer is no, because it is like a loaded dice game, but over time the odds will out.


*Given the timescales, and the length of time that there have been large mammals, it could be argued that our niche (as a, "fire-using social animal") might have been potentially fillable for some time before it was.


**I do agree that describing evolution as "random" is misleading, due to the connotations, unless you try to explain what is meant by this. However, I also think that "nonrandom" is also misleading, and probably even more wrong, but then I am happy talking about random events being highly predictable in certain fashions...

 

[Paul C. Anagnostopoulos]

I do agree that describing evolution as "random" is misleading, due to the connotations, unless you try to explain what is meant by this. However, I also think that "nonrandom" is also misleading, and probably even more wrong, but then I am happy talking about random events being highly predictable in certain fashions...

How about "nonrandom with respect to the current environment"?

Really, that's what it is.

~~ Paul

 

[mijopaalmc]

How about "nonrandom with respect to the current environment"?

Really, that's what it is.

~~ Paul

No, it is "biased with respect to the environment". Bias does not actually imply non-randomness.

 

[jimbob]

How about "nonrandom with respect to the current environment"?

Really, that's what it is.

~~ Paul

Yees sort of...

I'd prefer a longer statement that where selective pressures are constant, then there will be optimisation to that environment.

My quibble with your statment is that it could be read as saying that in a constant environment, optimistion will always happen in the same manner.

What is determined is that "the problem will be solved". What form of "solution" isn't predetermined, even in a constant environment. Convergent evolution shows that similar "solutions" do reoccur, but other "solutions" can also be viable and occur.

This also depends on the resolution.

In the link about cool-adapted bacteria evolution, all 12 flasks did demonstrate evolution and adapt to the cold. However different biochemical solutions were used, as some were also did better than the parent strain in the warm, whilst others did worse.

 

[Meadmaker]

That's why you cannot simply say "evolution is random" and then stop - it's just not that simple, and the statement is either meaningless (but correct using a naive definition of "random") or simply wrong.

Why are we still discussing this? Does anyone (other than mijo) actually disagree with anything I'm saying here?

I agree. The issue, for me, is that even if you don't stop, there are people who draw unwarranted inferences from the comment.

During the Dover trial, it came out that Miller's biology textbook, in one edition, had the phrase, "Evolution is random" in it. (IIRC the exact phrase was "Evolution is random and unpredictable". It was "evolution is random and ..." something.) Miller explained that he thought it was very unfortunate, and that the phrase had simply been slipped in by a collaborator and not caught in editing.

I thought Miller's response was unfortunate. The biology book contained all the context necessary to understand the phrase as written, and to realize that the author was emphasizing that there was no plan or design. For that purpose, and with the context to make it clear, "random" is quite a good word for it. We shouldn't be afraid of it. I think he should have left it in the later editions as well.

Oh, but wait. It's THEIR word, and they will undoubtedly twist it, and blah, blah, blah.

 

[zosima]

If we believe Stephen J Goulds claims on punctuated equilibrium I think that would suggest that evolution is highly non-random, or at least non-random given the environment.

His claim is that species and the distribution of genes in a gene pool don't change for millions,sometimes billions of years.. When they do change it is because something in the environment is putting pressure on the environment of the species. For example,there is a mass migration, an abrupt climate change, an invasive species, or an asteroid impact. All the sudden the species more quickly(in 1000s to tens of thousand of years, some argue less) to another equilibrium in their gene pool. Basically the rate of mutation is fast enough, by far, and sexual selection misses genes quickly enough, that any consequence of them averages right out.

That would imply a good analogy between a species and a gas. A gas in a closed container at equilibrium, has energy added, it quickly shifts to a new equilibrium, the size of the container increases or decreases and again a quick shift to a new equilibrium. A land bridge forms and a population quickly moves to a new and deterministic distribution of genes, but what genes the offspring of any single creature will end up with is as difficult as predicting the location of a particle in a gas.

Of course not all evolutionary biologists agree with punctuated equilibrium, but if you do, then evolution is random in the same way PV=nRT is random.

 

[mijopaalmc]

Of course not all evolutionary biologists agree with punctuated equilibrium, but if you do, then evolution is random in the same way PV=nRT is random.

But that doesn't mean that evolution by natural selection is not fundamentally random.

 

[zosima]

But that doesn't mean that evolution by natural selection is not fundamentally random.

I'm really not sure what you mean by fundamentally random. I stated in a previous post that I believe random a good definition is unpredictable and uniformly distributed. I also stated that I don't believe any phenomenon in the universe is really random, not even quantum physics. I'm perfectly comfortable with a non-local hidden variable theorem or a super deterministic theory, which excludes random behavior.

By my definition a gas should be called fundamentally random, because we describe it probabilistically. We can't prove that all the gas won't migrate into a single part of a container, and my positions on the quantum physics don't exclude that possibility. However, its just not significant because it is just so improbable.

 

[articulett]

I'm really not sure what you mean by fundamentally random. I stated in a previous post that I believe random a good definition is unpredictable and uniformly distributed. I also stated that I don't believe any phenomenon in the universe is really random, not even quantum physics. I'm perfectly comfortable with a non-local hidden variable theorem or a super deterministic theory, which excludes random behavior.

By my definition a gas should be called fundamentally random, because we describe it probabilistically. We can't prove that all the gas won't migrate into a single part of a container, and my positions on the quantum physics don't exclude that possibility. However, its just not significant because it is just so improbable.

Mijo has a very strong need to see evolution as random He cannot cede that it might not be useful or informative to say otherwise. Your explanations are patient and stellar. But I suggest you read this thread http://www.internationalskeptics.com/forums/showthread.php?t=82155&page=65 or at least pieces of it... because all the people who insist that evolution is "not nonrandom" appear there and have the same non arguments. All biologists, and people who actually teach and explain evolution to others recognize that your explanation is correct and useful and clear... as is Dawkins whom mijo imagines he's more of an expert than (he has no expertise in evolution at all as one can see by any of his posts here, but he appears to imagine himself as having expertise in both evolution and probability.) So it's not you.

Any thread where anyone says "evolution is not random"-- the same people appear saying the same things trying to convince someone or themselves that it really is. Every one tries to explain very carefully thinking that they might make progress, but it never happens. This thread from over a year ago will show you how the same people have identical "non arguments" to support the same garbled way of obfuscating understanding. Interestingly, it's the very same technique used by creationists to obfuscate understanding of natural selection by claiming that scientists think "all this life stuff just came about randomly"

Yes, there is no plan--no designer needed... but it's far from "random"-- it's a progressive accumulation of the best reproducing DNA over time... a long, long time. Darwin's theory was about natural selection... he had no way to know if the variations in pheontype were "random" or not. They don't need to be "random" for evolution to work.

 

[cyborg]

I see no one read the paper I linked to.

 

[mijopaalmc]

I see no one read the paper I linked to.

And I'm guessing that it said that evolution by natural selection was not random because it was "biased", "causal", "constrained", "directional", "equiprobable", or "predictable".

When will you learn that you are just proposing straw men or succumbing the way in which the creationists use random?

 

[cyborg]

And I'm guessing that it said that evolution by natural selection was not random because it was "biased", "causal", "constrained", "directional", "equiprobable", or "predictable".

No.

When will you learn that you are just proposing straw men or succumbing the way in which the creationists use random?

When will you learn something about the words you are using? Never if you refuse to read.

 

[zosima]

Actually, the problem seems to be that the majority of the phenomena that people who argue that evolution is non-random are not in and of themselves random, so the fact that we observe them in evolution by natural selection does not make it inherently non-random. By the way, my original argument, which seems to have been lost in the recesses of time, was that there is no evidence that evolution is non-random. Such a lack of evidence does not mean that evolution is therefore random, but my main point of confusion was why people were so adamant that evolution was non-random when the evidence they cited said nothing of the sort.

I cited evidence that evolution is non-random. Whether it is random or not depends on the evidence supporting punctuated equilibrium. There is plenty of evidence supporting punctuated equilibrium, but then again if you want you can always argue about the fossil record.