What evidence is there for evolution being non-random?

[Dr Adequate]

I just think that which local optima it tends to is up for grabs, and highly dependent upon which mutations happen first, the happenstance of which genes turn up mixed together first, arbitrary environmental interactions and genetic drift.

Yes, evolution has random elements. So does backgammon, but the last time I met someone who told me it was "a game of chance" I beat him 13 games to 0.

If it was very stable all life would have evolved together as one homogeneous species, instead of branching and diversifying.

Not if there is more than one environmental niche.

Check out the TIERRA program. That shows diversification from uniformity, and the outcome is fairly predictable, in that one can say what sort of diversity will arise.

 

[CFLarsen]

Given that each organism "evolves" independently

No, it does not.

Organisms evolve depending on not only environmental conditions, or mutations, but very much in cooperation/competition/symbiosis with other species.

 

[cafink]

Dr. Adequate, your Backgammon analogy is simple and elegant. I think that's the perfect way of looking at the problem.

 

[Walter Wayne]

Thank you.

Now that we have a definition, we can have a discussion. To use my philosophy minor for some use, your argument appears to be the following:

1 - Systems which have identical initial inputs can have different outputs. (In other words, non-determanistic).
2 - Evolution is such a system.
C- Therefore, evolution is random.

I wouldn't call that my argument. For one I talk about two major components of evolution, mutation, and selection. I also go briefly into some of the characteristics of the system would lead to random outputs with random inputs, (the fact the each random input isn't independent of the last, the interdepedency of various species, the non-static environment ...)
1 is just a definition of random, 2 is the conclusion of my argument and C is basically a repeatition of 2.

Your argument is sound, but I disagree with your premises. Specifically, 2. You are including all of evolutionary history as your "system", but only setting the first set of inputs as the same.

I have said, I beleive twice, that even with all non-bioligical inputs the same (meteors, continental drift etc.)

So lets simplify your example into three scenarios:

1) A point mutation occurs. Assuming equal ratios of nucleotide mutation frequiencies, and assuming identical initial conditions, do you think the same mutation would occur a second time?

Not necessarily, specifically because population don't always have all the time in the world to adapt. We are looking at a fairly steady time for humans, but what about populations that are vulnerable. A small breeding population that has a few generations to establish itself either out competes its neighbours or perishes. It doesn't get to roll the dice a thousand times, so mutations don't become inevitable. If a mutation has a one in a million chance it will happend about 6000 times in the next human generation, the same can't be said of smaller breeding populations that occur during history.

Rolling a six on a six-sided die is easy. Just set there and keep rerolling. But if you have to roll it on the next six tries, well now you have a 33% chance of failure.

So while I expect a system of many die-rolls to be predictable, lesser numbers are not.

2) A defined level of variation exists in a population. Selection changes the frequencies of alleles in the population over time (i.e. evolution). Assuming identical initial conditions (i.e. the same defined variation), do you think we would see the same change in allele frequencies occur a second time?

This doesn't invoke the random element of mutation. For the purposes of this argument I am stipulating it is a deterministic process.

3) A population of E. coli exists in which all the members contain a plasmid which encodes a Kanamycin resistance gene. In this gene is a transposon (transposable element), thus switching this gene off. Growing colonies of this E. coli strain on growth medium which contains Kanamycin leads to the arisal of Kanamycin resistant colonies, as a result of the transposon 'jumping' out of the gene, thus switching it back on. When this happens is unpredicable. Given identical initial conditions, do you think the same ratio of Kanamycin resistant colonies would arise?

The number of E. coli around makes me expect that variation in number would be pretty minimal. Again your dealing with a vary large population.

 

[Meadmaker]

So, is that a good layperson's explanation? Or am I wrong somewhere? ('Cause if I'm wrong I'll fix it.)

Is there any other term used in the sciences that we are so concerned about the layperson's understanding? Maybe we should ask the layperson to discuss quantum mechanics?



However, be that as it may, there is one sense of the word "random" that is proper English usage and most certainly does not describe evolution. That is the sense in which people think that all outcomes are equally likely. In mathematical terms, they confuse "random" with "uniformly distributed". Some people will insist that if you have a loaded die, the outcome is not random because one side is more likely to come up than the others. Evolution is a bit like a loaded die. Not all forms of a gene are equally likely to survive. Does that make it non-random? If you insist that the outcome of a loaded die is non-random, then evolution is non-random as well.


Even then, your (jimlintott) example of the layperson's understanding isn't quite correct. I think the layperson would say that a fair die was "truly random" (a phrase with no coherent meaning, despite it's common use) even though there was no possibility that it would turn into a parrot and fly away. Even for the layperson, not every conceivable outcome need be possible, such as dogs producing kittens. An uninformed layperson might insist that evolution is not random because not every organism has an equal chance of survival. It's a misunderstanding, but that might be something people think.

Oh, well, maybe all this talk about the meaning of "random" isn't all that important. When describing evolution, why not use the language the way the layperson uses the language, instead of its more precise technical meaning. After all, the whole thing's only a theory.

 

[Walter Wayne]

Are you asking whether evolution is random with respect to the universe or random with respect to the Earth? It's clearly not random with respect to the Earth, because, as you say, organisms evolve to be more fit in their environments. You might argue that evolution is random with respect to some larger environment, such as the universe, because there is much less interaction between Earth's organisms and the universe as a whole.

I'm not sure what you mean by increasing the fitness of individuals, but randomly.

~~ Paul

There is more than one way to become more fit in their environments. And with limited resources you don't get to find all the ways to become more fit. Which fit solution emerges is random.

 

[Walter Wayne]

Yes. I also made the graph.

Do you understand it?

Really. Were you right in any way?

Yes.

 

[Paul C. Anagnostopoulos]

There is more than one way to become more fit in their environments. And with limited resources you don't get to find all the ways to become more fit. Which fit solution emerges is random.

Well, it's a random distribution with highly weighted probabilities. I can become fit to climb trees by evolving large claws or by evolving an inflatable balloon appendage, but the probabilities aren't equal.

As you say, though, certain aspects of the result of evolution are random, within constraints. But it is misleading to state that evolution is random, because any fit result is not random with respect to the environment.

Much of the problem has to do with the layman's understanding of random. We say that any process with a random input is a "random process," but saying that to a layman is grossly misleading.

~~ Paul

 

[Walter Wayne]

I see that most of this thread has degenerated into a bickering about the meaning of the word "random". Well, it is clear that if you told someone that "evolution is random", you'd be giving them an entirely false impression.

And if you said "evolution is non-random" you'd also be giving them a false impression. Some idiot IDers played with a little equivocation, and instead of pointing out this blatant dirty word play some biologist and skeptics have decided to except there definition which is useless to describe any process other than whirlwinds creating 747s in junk yards.

The line of reasoning leading to the claim that "evolution is random" would seem susceptible to a reductio ad absurdum. If evolution is "random" because the mutations on which selection acts are "random", then we must regard evolution by artificial selection as being "random" too, on the same basis. What's sauce for the goose is sauce for the fundie propaganda.

Actually, no one's argument is that evolution is random because the mutations on which selection acts are random. I have made some specific points about the characteristics of selection which are sufficient to make the process random if the mutations are random.

Even T'ai Chi's argument goes beyond mutation is random therefore evolution is random.

 

[Dr Adequate]

Do you understand it?

Yes. I made it, I wrote the accompanying article, is there something you don't understand?

Yes.

What way would that be?

 

[Dr Adequate]

And if you said "evolution is non-random" you'd also be giving them a false impression. Some idiot IDers played with a little equivocation, and instead of pointing out this blatant dirty word play some biologist and skeptics have decided to except there definition which is useless to describe any process other than whirlwinds creating 747s in junk yards.

It's no use blaming the fundies, they didn't invent the English language --- as I have to remind them from time to time.

Actually, no one's argument is that evolution is random because the mutations on which selection acts are random. I have made some specific points about the characteristics of selection which are sufficient to make the process random if the mutations are random.

Can you list any of these points which do not apply to evolution by artificial selection?

If not, my point still stands.

For example, you have just written:

"There is more than one way to become more fit in their environments. And with limited resources you don't get to find all the ways to become more fit. Which fit solution emerges is random."

Substitute "useful" for "fit".

 

[Walter Wayne]

If the collisions of gas molecules against the walls of their container is random, then air pressure must be random too. So how can air pressure in your tires hold your car up off the road?

If the outcome of a pull on a slot machine is random, then the money the casino makes (or pays out) must be random too. So how can casinos predict their profit margins?

If the directions that neutrons are emitted by a nucleus undergoing radioactive decay are random, then a fission chain reaction must be random too. So how is it possible to control a nuclear reactor?

If passive diffusion of a molecule across a membrane is random, then osmosis must be random too. So how can the net flow through a membrane be precisely repeatable for a given salt concentration?

If individual variation is random, and some accidents of birth and death unrealated to genotype are random, then evolution must be random too. So how can complex inderdependent genes arise?

The answer is the same in all cases: The statistical characteristics of a large number of individually random events are not random.

Respectfully,
Myriad

To avoid long windedness and going into each example, why did you chose only illustrations including independent random variables to illustrate a problem including heredity.

Example, throw a million fair six-sided die. Probability that all 6 rolls (1 through six) occured at least once, 1 (or close enough). It is for all intents and purposes determistic.

Now add dependence. Throw 12 fair six-sided dice in the first "generation". In the second generation, we double the number of dice, but they are no longer fair. Odds are now waited by the distribution of the first twelve throughs. That is if 3 of the 12 throws came up as a four, then a roll would have a 1 in 4 chance of having coming up four, while if 2 of the 12 rolls came up five, then a roll would have a 1 in6 chance of coming up five.

Repeat this, doubling each "generation", and waiting the probability of outcomes of rolls by the population distribution of the previous generation. Eventually you will have a "generation" of over 1.5 million die rolls. What is the probability of all 6 rolls occuring at least once in that 1.5 million trials.

Well, the probability of all 6 numbers occuring in that first generation is less than 1 in 3. Since subsequent generation probabilities are inherit probability distribution from previous generations, once a number doesn't appear in a generation it is gone forever. So that final 1.5 million trials will more likely than not contain only a subset of the six numbers.

Walt

 

[thaiboxerken]

Seems to me that some people want to call an entire process random if any part of that process has a random element. It's like calling a person who told a lie, one time, a liar but not calling him a truth-sayer even though he's told the truth more often.

 

[Walter Wayne]

Given that each organism "evolves" independently, I suspect that it's not as "up for grabs" as you seem to think.

Individuals aren't selected independently unless they don't compete for resources, the same with respect to species evolving.

If there are two or more local optima, the usual outcome of evolution is to tend to both of them.

At once.

We call this "speciation."

That assumes that said optima are reachable without mutation. A small population can't rely on randomness of mutation to reach every available local optima.

 

[Walter Wayne]

Yes, evolution has random elements. So does backgammon, but the last time I met someone who told me it was "a game of chance" I beat him 13 games to 0.

Backgammon has intelligent elements to it, and given sufficient disparity in intelligence one gets a somewhat trivial system.

Evolution doesn't have intelligent elements to it.

 

[Walter Wayne]

However, be that as it may, there is one sense of the word "random" that is proper English usage and most certainly does not describe evolution. That is the sense in which people think that all outcomes are equally likely. In mathematical terms, they confuse "random" with "uniformly distributed". Some people will insist that if you have a loaded die, the outcome is not random because one side is more likely to come up than the others. Evolution is a bit like a loaded die. Not all forms of a gene are equally likely to survive. Does that make it non-random? If you insist that the outcome of a loaded die is non-random, then evolution is non-random as well.

An odd thing is that people who insist that a loaded die isn't random sometimes say that the sum of two rolls a fair die is random even though there is a bias to 7. I've noticed the inconsistency more than once.

 

[Schneibster]

Let's try this again.

First, fitness landscapes are at minimum pseudo-random; they are determined by exigent circumstances. For example, what does the path of a comet have to do with the shape of the fitness landscape? Not much- unless it hits the Earth. Then it has a great deal to do with it. So again, what do you mean by "random?"

Now, genes. The production of novel genes occurs by several means, all of which are again, at minimum, pseudo-random, and some of which are truly random- for example, the fastest spermatozoan happens to be the one that had a chromosome altered by a passing cosmic ray. The speed of the spermatozoan doesn't depend on this alteration. So an egg is fertilized, and the altered chromosome happens to be expressed. That leads to a novel phenotypical characteristic that leads to the organism surviving the above-mentioned comet strike.

And yet again, what do you mean, "random?" It looks pretty random to me.

 

[Taffer]

I wouldn't call that my argument. For one I talk about two major components of evolution, mutation, and selection. I also go briefly into some of the characteristics of the system would lead to random outputs with random inputs, (the fact the each random input isn't independent of the last, the interdepedency of various species, the non-static environment ...)

Fine. But only some of the "inputs" of evolution are random. Specifically what form variation takes. But selection acts on the same variation the same way every time, given the same environment.

1 is just a definition of random, 2 is the conclusion of my argument and C is basically a repeatition of 2.

It's called an argument. 1 and 2 are premises. 3 is your conclusion.

I have said, I beleive twice, that even with all non-bioligical inputs the same (meteors, continental drift etc.)

You miss my point. You set the initial conditions the same, but only the initial conditions, not all conditions. If something is truely random, even if all the conditions are the same, the output is not predictable. But, with evolution, if the variation is the same, the output is predictable.

Not necessarily, specifically because population don't always have all the time in the world to adapt. We are looking at a fairly steady time for humans, but what about populations that are vulnerable. A small breeding population that has a few generations to establish itself either out competes its neighbours or perishes. It doesn't get to roll the dice a thousand times, so mutations don't become inevitable. If a mutation has a one in a million chance it will happend about 6000 times in the next human generation, the same can't be said of smaller breeding populations that occur during history.

Rolling a six on a six-sided die is easy. Just set there and keep rerolling. But if you have to roll it on the next six tries, well now you have a 33% chance of failure.

So while I expect a system of many die-rolls to be predictable, lesser numbers are not.

This has nothing to do with my first situation. I simply said that, do you think that given the same initial conditions, the same mutation would occur a second time? In other words, do you think there are any predictable parts to mutation?

This doesn't invoke the random element of mutation. For the purposes of this argument I am stipulating it is a deterministic process.

So you are saying that in the second situation, it will always occur?

The number of E. coli around makes me expect that variation in number would be pretty minimal. Again your dealing with a vary large population.

Again that is not what I said. These are very specific situations, so please stick to what is outlined. In the third situation, it is not variation which is causing the evolution of Kanamycin resistance, but the transposition of a transposon which has been inserted into the Kanamycin resistance gene.

 

[Taffer]

Backgammon has intelligent elements to it, and given sufficient disparity in intelligence one gets a somewhat trivial system.

Evolution doesn't have intelligent elements to it.

Can you describe the difference between "intelligent elements" and those which are not?

 

[Dr Adequate]

Backgammon has intelligent elements to it...

And random elements. Is there any chance you could address the relevant point of the analogy?

You might as well complain that backgammon, unlike evolution, takes place on a green baize board.

Evolution doesn't have intelligent elements to it.

Evolution by artificial selection does, in fact, have intelligent elements to it, and I notice that you have not answered my question.

Do you have any criterion by which you can claim that evolution by natural selection is random and evolution by artificial selection is not?