Randomness in Evolution: Valid and Invalid Usage

[Belz...]

Determinism is possible only in extremely simple and totally unrealistic mathematical models, like the two-body problem in Newtonian gravity.

And yet everything averages out so that our predictions work remarkably well, every time. Sounds deterministic to me, no matter how random the underlying physics are.

 

[sol invictus]

And yet everything averages out so that our predictions work remarkably well, every time. Sounds deterministic to me, no matter how random the underlying physics are.

Every time? Can you given an example of a prediction like that?

Certainly there are things we can predict with high accuracy, but the predictions are always probabilistic.

There may be a semantic issue here - the word "deterministic" has a very specific technical meaning which is not compatible with randomness in the underlying physics.

 

[Dancing David]

It ain't evolution as we know it without mutation. Genetic drift is also part of evolution, but usually considered seperate from natural selection.

Walt

Thats very silly, why are artic foxes white? Any difference in genetics which influences reproduction is subject to natural selection.

Where did your notion come from?

 

[Dancing David]

And saying that chance wasn't involved, also misses a significant part of the story, especially as the odds of an individual cod fry reproducing are of the order of 500,000:1 against. The fry could have traits that make it a thousand times more likely to reproduce, and it would still only have about a 0.2% chance of reproducing. If it did breed, and half its fry also had these traits, then you would expect about 500 offspring to reproduce, and the trait would spread.

Of course, in reality, beneficial mutations are only likely to give a few percent advantage. This is enough for evolution to work, but whether an individual trait spreads form its original parent has to be largely influenced by chance.

And natural selection doesn't care. It is not about the individual, although they are the ones who reproduce or not. A species is composed of individuals, the mean traits of a species are not significant for the individual.

yes fish economics are a high output low probability , even worse for bug economics.

Natural selection doesn't care about the chances of an individual, that is strange deterministic thinking.

Natural slection happen every time an organism reproduces.

 

[Dancing David]

Back to the OP.

Mutation: Random and haphazard.

Natural slection: Probabilistic (due to chaotic systems) but not haphazard. (The "game" is played with loaded dice).

Evolutionary Direction: Stable for long periods of time, but affected by random events. Probabaly chaotic.

Just like Mijo and Wayne, you are hung up on mutation, gosh , it is NOT the only means of genetic variability. It is not the only factor in variability that effects natural selection.

Why are you three obsessed with it?

Why are artic foxes white?

 

[Walter Wayne]

Thats very silly, why are artic foxes white? Any difference in genetics which influences reproduction is subject to natural selection.

Where did your notion come from?

What is your issue with my quote. Is mutation not a component of our current understanding of evolution? Does it not create the new "information" that natural selection acts on.

Just like Mijo and Wayne, you [jimbob] are hung up on mutation, gosh , it is NOT the only means of genetic variability. It is not the only factor in variability that effects natural selection.

Why are you three obsessed with it?

Why are artic foxes white?

Hung up on mutation? That's hillarious.

Articulett left it out of her "useful definition" of evolution all together, and claimed natural selection was responsible for iteration of information (not heredity) and "increasing" complexity (not mutation). I'm not hung up on it, rather I am giving it its proper place. How anyone can claim to have an expertise on the subject of evolution, but not realize the roles of the different components of it is beyond me.

Both Jimbob and I are also arguing that it is the properties of natural selection, that make the process random at larger scales. I have even argued elsewhere, that even if mutation was determistic, evolution would still be random except in the most technical sense of it.

Walt

 

[sol invictus]

Articulett left it out of her "useful definition" of evolution all together, and claimed natural selection was responsible for iteration of information (not heredity) and "increasing" complexity (not mutation). I'm not hung up on it, rather I am giving it its proper place. How anyone can claim to have an expertise on the subject of evolution, but not realize the roles of the different components of it is beyond me.

I think her (?) point is that mutation is not necessary for evolution. Imagine starting with a population with some genetic variability present in the genome, but with no mutations possible. That population will evolve under natural selection, perhaps towards more complexity (for example if the more complex organisms present are more fit, or through simple sexual reproduction).

Obviously mutation was a crucial part of the evolution of life on earth up to this moment, but it is not necessary for evolution per se, and it is not responsible for many of the examples of evolution we see occurring on shorter time scales.

 

[Belz...]

Every time? Can you given an example of a prediction like that?

Certainly there are things we can predict with high accuracy, but the predictions are always probabilistic.

There may be a semantic issue here - the word "deterministic" has a very specific technical meaning which is not compatible with randomness in the underlying physics.

Well, for one we can send probes to pluto and they actually get there.


As I've learned from my time here, "random" and "deterministic" seem to be able to mean anything we want!


So, honey, pass the deterministic, please. No, the other one.

 

[sol invictus]

Well, for one we can send probes to pluto and they actually get there.

Always? You think if we sent 100 probes to Pluto, they'd all get there? Considering the that NASA has difficulty even converting metric to english, I doubt that very much.

As I've learned from my time here, "random" and "deterministic" seem to be able to mean anything we want!

Yes .

But there really is a technical definition of determinism, and probabilistic systems don't meet it. Actually "random" is much harder to define - it's not really a technical term at all, contrary to what mijo likes to claim.

 

[Wowbagger]

Well, I suppose there can be technical definitions of the word "random" in different contexts. It's just that there are a lot of different contexts.

 

[sol invictus]

Well, I suppose there can be technical definitions of the word "random" in different contexts. It's just that there are a lot of different contexts.

I think it's just very hard to define precisely. I checked a standard textbook on probability theory and it didn't define it at all - in fact it essentially never used it (there were only two instances of the word, both informal).

You can define a random variable in a formal way, as here, but as we've seen that doesn't provide a useful definition of "random". And even mijo, after months (or years?) of harping on this, still didn't understand even that not-so-useful definition, as the exchange regarding the wave function a few pages back evidenced.

Actually it might be interesting to open a thread on randomness - it's really a very slippery concept - and that slipperiness might have something to do with the ridiculous length of this discussion.

 

[cyborg]

As I have proposed before algorithmic descriptions provide a much more tangeable notion of randomness.

In short this is:

Sequences for which the shortest algorithm (with input included) for producing them is the sequence itself are random.

 

[sol invictus]

As I have proposed before algorithmic descriptions provide a much more tangeable notion of randomness.

In short this is:

Sequences for which the shortest algorithm (with input included) for producing them is the sequence itself are random.

Well, that's a good approach, but you're not there yet. Any given finite sequence can be compressed at least a little. So you have to define this as some kind of limit, for an infinite number of sequences or infinite length ones. That's where it gets subtle.

 

[cyborg]

Any given finite sequence can be compressed at least a little.

No it can't.

Simple counter-example for you - compress this sequence:

1

 

[Dancing David]

What is your issue with my quote. Is mutation not a component of our current understanding of evolution? Does it not create the new "information" that natural selection acts on.


Hung up on mutation? That's hillarious.

Articulett left it out of her "useful definition" of evolution all together, and claimed natural selection was responsible for iteration of information (not heredity) and "increasing" complexity (not mutation). I'm not hung up on it, rather I am giving it its proper place. How anyone can claim to have an expertise on the subject of evolution, but not realize the roles of the different components of it is beyond me.

Both Jimbob and I are also arguing that it is the properties of natural selection, that make the process random at larger scales. I have even argued elsewhere, that even if mutation was determistic, evolution would still be random except in the most technical sense of it.

Walt

You and the other two are hung up on mutation, it is NOT the only thing that natural selection acts upon.

It ain't evolution as we know it without mutation.

is your quote.

Do you really think that artic foxes are white because of a mutation? Do I really have brown eyes because of a mutation (both my parents have hazel eye).

there are variations in the genome that don't require mutation as the basis of vaiability. I can speel it out for you if you want, I will ask again:

How does mutation make artic foxes white?

Answer the question.

 

[Belz...]

Always? You think if we sent 100 probes to Pluto, they'd all get there? Considering the that NASA has difficulty even converting metric to english, I doubt that very much.

Yeah but all kidding aside, the probes that wouldn't get there wouldn't not because pluto skips an orbit, or anything, but because of a technical difficulty or unknown variable such as collision, etc.

 

[Wowbagger]

I think it's just very hard to define precisely. I checked a standard textbook on probability theory and it didn't define it at all - in fact it essentially never used it (there were only two instances of the word, both informal).

I know that when Dawkins has used it, it meant something similar to "directionless" and/or "indifferent", but I wonder how he would define it more precisely. Hmmm...

Actually it might be interesting to open a thread on randomness - it's really a very slippery concept - and that slipperiness might have something to do with the ridiculous length of this discussion.

Another one?! I suppose a more generic thread, on what randomness actually is (aside from its relevance to biology), might be useful.

If any long tirades about evolution crop up in it, we should be sure to point them back here.

 

[jimbob]

Yawn

You're talking as if you're providing any new information. You're just not getting what is being said and I'm getting tired of explaining myself.

Well you have never explained how your approach is useful in answering those sort of questions.

If you are saying that evolution doesn't need probabilistic selection to work, that might be arguable*, however I have been under the impression that we have been discussing how evolution works, within known biology. In this example, it works probabilistically, fitness and luck both play a part.

*It would still be wrong though for the following reasoning.

For darwinian evolution it is necessary and sufficient for there to be imperfect self-replication.

If there is, then in any finite system, there will eventually be a resource limitation, thus there will be competition amongst the replicators, which will lead to natural selection, even without any other form of natural selection.

The replicators will thus affect the selective landscape for the surrounding replicators. This type of feedback loop is charateristic of certain types of chaotic system.

If the evoutionary landscape is chaotic, then the evolutionary direction is chaotic. And if it is chaotic, it is wrong to say that it is completely nonrandom, as over long enough timescales, it is random.

The selection is also likely to be chaotic for the same reasons.

 

[cyborg]

For darwinian evolution it is necessary and sufficient for there to be imperfect self-replication.

Oh, here we go again.

If the evoutionary landscape is chaotic, then the evolutionary direction is chaotic. And if it is chaotic, it is wrong to say that it is completely nonrandom, as over long enough timescales, it is random.

You still haven't understood a singular thing I've said.

Come back to me when you understand what I mean by "algorithmic complexity" then we can discuss the notions of "chaos", "random" and "completly non-random" with some sort of formalism rather than, "well, we all know what that means".

 

[jimbob]

Yeah but all kidding aside, the probes that wouldn't get there wouldn't not because pluto skips an orbit, or anything, but because of a technical difficulty or unknown variable such as collision, etc.

We are talking about long timescales here:

According to wiki, Pluto's orbit is unpredictible beyond about ten million years.

I'd imagine that chaotic NEOs are on a slightly shorter timeframe, having a higher orbital frequency, but even ten million years in the context of the 3.8 billion that life has been on Earth is sufficiently short for it to be considered random.

Again I'll reiterate, you can make predictions, but they need to be probabilistic if talking about suitably long timescales.