Randomness in Evolution: Valid and Invalid Usage

[jimbob]

Who is trolling, and why does articulett think so?

She seems to be the main person who doesn't have anything to contribute in this discussion, except to make claims about other posters.

Wowbagger, do you see what I am getting at wth the random factor being more important over geological timescales?

I am not critising evolutionary thory for being random, but there is a fundamental difference in the type of prediction and confidence between random and nonrandom systems.

It is possible to be very confident in predicting the Earth's orbit, position and velocity 100-million years from now. It isn't possible to do the same for an evolutionary course over this timescale into the future. This isn't just because it is predetermined but hard to predict, but that this evolultionary course isn't yet determined and will be significantly affected by future random events.

In stable environments or where there are constant selective pressures, there will be adaptation to these pressures, some of which are more probable than others. Over geological timescales the environments are not constant, and when they change, and the ecosystems are plastic, these ecosystems are subject to random changes that then set the course of evolution until another unstable period when random factors again become important.

Long periods of stable evolution, punctuated with "points of inflection" where ecosystems are plastic and subject to significant random alteration, the outcome of which will determine the general shape of the ecosystem and thus evolution until the next disruptive event.

Whether the event itself is random isn't really the point. It wipes enough of the slate clean so that the biological interactions can "set" significant selection criteria, and this criteria will depend on nonlinear interactions and be significantly altered by random events.

 

[cyborg]

Whether the event itself is random isn't really the point.

Then why bang on about Quantum Mechanics if:

It changes nothing.

?

 

[sol invictus]

nonlinear systems tend to diverge over time i.e. these differences get magnified not averaged out.

jimbob, a quick comment. You're using "non-linear" as a synonym for "chaotic". That's not really correct. Non-linearity does not always imply chaos, and I can think of linear systems that exhibit chaos.

For an example of the first, take air resistance. An object moving though air experiences a drag force that is proportional to velocity squared (so it's non-linear), but the results are quite predictable (it slows and stops). No chaos.

For the second, take a single frictionless particle bouncing around in a container with complicated walls. Perfectly linear, but chaotic (change the initial position a little and you make a huge difference in the position a little later).

Evolution - as I think we all now agree - is unpredictable (and probably chaotic) in many of its details, but predictable in some details and in its macroscopic behavior. Calling something like that "random", full stop, is obviously a mischaracterization.

 

[jimbob]

Whether the event itself is random isn't really the point.

Then why bang on about Quantum Mechanics if:

It changes nothing.

My full quote was:

Whether the event itself is random isn't really the point. It wipes enough of the slate clean so that the biological interactions can "set" significant selection criteria, and this criteria will depend on nonlinear interactions and be significantly altered by random events.

The event might have been inevitible, but its consequences weren't.

Back to the KT impact.

Whether the impact itself* was random, was not my point.

The consequences of the KT impact were random. Yes it led to the "Rise of The Mammals (TM)" but it needn't have; other truly random events were important. The impact, and other events where the envitonment changed significantly wiped the slate clean, and sensitised the ecosystem to developing in directions which had been affected by random factors, i.e which organism began randomly adapting to a particular niche, and how this affected the fitness landscape for other organisms.

The rise of grasses was another event that significantly altered the fitness landscape, and I would argue that such an occurrance could also lead to a sensitivity to random factors.

Over geological timescales I think it makes mores sense to describe the course of evolution as random. In short timescales, and with stable environments, I would agree that it isn't. However if people are discussing the evolution of humanity, which is a popular subject, then this evolution has been significantly affected by random factors.


*There is evidence that cometry Near Earth Orbits are chaotic, which would make the impacts themselves sensitive to random events.

 

[jimbob]

Sol, I'd agree with most of what you are saying with a few caveats:


{nitpick}If something like air resistance (or energy) is proportional to the square of the velocity, then I would say that this is still a linear system. {/nitpick} However there are examples, supporting your point, where nonlinear systems are nonchaotic, even the classic pendulum which we were all taught as simple harmonic motion, isn't really a linear system as it relies on the approximation:
sin(x) -> x , for small values of x...

However the types of feedback loops in biological systems, (and ecosystems, which is what defines the evolutionary landscape) will mean that aspects of evolution are sensitive to chaotic systems.

My maths isn't good enough to know if all the interactions can even be analytically described, but there would be discontinuities, and significant sensitivty to random factors.

Evolution - as I think we all now agree - is unpredictable (and probably chaotic) in many of its details, but predictable in some details and in its macroscopic behavior. Calling something like that "random", full stop, is obviously a mischaracterization.

I'd agree with that statmenet, however I am saying that calling it "nonrandom" is also wrong. Especially over geological timescales, when there are periods where the evolutionary landscape is "randomly" altered.

 

[sol invictus]

{nitpick}If something like air resistance (or energy) is proportional to the square of the velocity, then I would say that this is still a linear system. {/nitpick}

Well, you could say that, but you'd be wrong. The differential equation has a term in it that's the square of the velocity. When you solve for the position as a function of time, you're solving a non-linear equation. Multiplying the solution by a constant does not result in a new solution, you cannot superpose two solutions, and you cannot write the equation as a linear operator acting on an unknown function. It's non-linear.

However there are examples, supporting your point, where nonlinear systems are nonchaotic, even the classic pendulum which we were all taught as simple harmonic motion, isn't really a linear system as it relies on the approximation:
sin(x) -> x , for small values of x...

Yes, that's another example.

 

[articulett]

Ah... it's time for me to point out once again:

"Having random components does not a "random process" make".

Not that it matters for those who desperately wish to believe this to be so.

 

[jimbob]

Well, you could say that, but you'd be wrong. The differential equation has a term in it that's the square of the velocity. When you solve for the position as a function of time, you're solving a non-linear equation. Multiplying the solution by a constant does not result in a new solution, you cannot superpose two solutions, and you cannot write the equation as a linear operator acting on an unknown function. It's non-linear.

Yes I had a brainstorm about that part

ETA: but my reasoning about the rest is unaffected.

 

[jimbob]

Sol, back to my points about chaotic systems.

They are significantly affected by random events.

Doesn't this mean that an asteroid in a chaotic Near Earth Orbit is also going to be significantly affected by random events, possibly to such an extent that it might hit either the Earth or Moon (or neither), with vastly different effects on the course of evolution.

Do you see my point about random factors being important over geological timescales?

 

[sol invictus]

Sol, back to my points about chaotic systems.

They are significantly affected by random events.

Doesn't this mean that an asteroid in a chaotic Near Earth Orbit is also going to be significantly affected by random events, possibly to such an extent that it might hit either the Earth or Moon (or neither), with vastly different effects on the course of evolution.

Do you see my point about random factors being important over geological timescales?

Sure - at any moment a giant asteroid could come along and smash the earth into tiny pieces, killing all life but some anaerobic bacteria (and thus rather drastically changing the course of evolution). Or intelligence could evolve, develop thermonuclear weapons, and engage in a nuclear war. Or the sun could capture a wandering star and go nova. Or a meteor strike could make all the giant lizards go extinct, allowing some furry rodents to take over the world and eat lots of chicken.

But you could say the same things about your pendulum - you claimed it was deterministic and predictable, but what if someone comes along and yanks it off its peg on a cosmic-ray-to-the-brain inspired whim?

The whole discussion is ridiculous. Everybody here (just about) agrees on the phenomena in question - we're just arguing over a word. But this is a complex process, and you just can't capture it very well with one word - you have to say a few more. Tough.

 

[articulett]

Jimbob... you argument is the equivalent of arguing that a random number generator is itself random... or it produces numbers "randomly" because we can't predict the outcome in advance and some small random change to the code can change the whole outcome.

No matter how you slice it, that still doesn't make random number generators "random" nor does it mean they generate numbers "randomly"... you can say that if you don't mind being unclear, but why would you?

And why would you insist on using such an unclear explanation in regards to evolution? Go to the other thread where a creationist is wondering how scientists could think this all came about randomly and see if you can convey any actual information about how order comes about from the randomness while insisting that evolution IS random.

The process that brings order to randomness is not, itself, random.

The Brazil nuts don't rise to the top of the mixed nuts randomly... they rise through a gravitational process, whereby the little peanuts fall between the spaces and end up on the bottom. If you want to call this random-- be my guest... but it make you look like a boob. Why in the world would you need to convince someone this is an intelligent way of describing such a process? And that goes triple for evolution.

You go on and on spinning these semantics as long as you have an audience. It's maddening to new people, I'm sure.

 

[Walter Wayne]

Ah... it's time for me to point out once again:

"Having random components does not a "random process" make".

Not that it matters for those who desperately wish to believe this to be so.

And it has been pointed out to Articulett before that having non-random components does not a non-random process make.

And her comment about random number generators is nothing like what Jimbob is arguing, and I assume from her description she is talking about pseudo-random number generators.

Walt

 

[Wowbagger]

Wowbagger, do you see what I am getting at wth the random factor being more important over geological timescales?

I see the point that small random effects could be magnified over time. Which is one reason why it becomes more difficult to predict the future the further into the future you try to look.

But, that has nothing much to do with what I am talking about: Which has more to do wtih the general process acting over time.

I know you were not criticizing evolution, jim. But, I suspect others on here, might be.

 

[articulett]

No one here (except T'ai) is actually criticizing evolution (except in a backhanded way)... they just refuse to convey it a way that is understandable to those who don't understand evolution... leading me to wonder why they are so keen to convey evolution in a certain way, when all the experts find such descriptions obfuscating and misleading.

It is the same as insisting that a random generator is, itself, random...
Why would you do that if you were trying to convey what a random generator is or how it uses code to generate random numbers... how a known process can bring meaning to randomness...

Why would anyone insist on calling a random number generator "random" simply because it produces random numbers, the numbers are unpredictable, and a "random" change in the code would produce different results. The nonrandomness of a random number generator has nothing to do with whether a single event could change (or could have changed) the outcome! Whether a given sequence was "inevitable" per Jim's tangential argument is irrelevant to whether the random number generator (or evolution) is random in itself.

Jim continues to argue this inevitability point as though it means something in how we describe the process of evolution. It's backwards. It obfuscates. It's an awkward attempt to have it make sense to yourself to call evolution "random"-- or to convince yourself that scientists think this all came about "randomly".

The thing that brings order to the randomness (in the case of evolution, it's "selection") is not, itself, random... nor is the process that involves it. Having random components does not a random process make.

He wants you to agree that he's making sense in his description of evolution... that he's clear about something or conveying information to someone in a useful way. He wants to believe that he is more clear than the actual experts multiply quoted in this thread. He wants others to believe he has as much expertise on this topic as he seems to imagine.

 

[cyborg]

Well I find it odd that jimbob doesn't, to the best of my recollection, have any particular issue with genetic algorithms despite the fact that they don't really have much choice on the whole other than to deal with pseudo-random numbers. (I.e. it is technically a completely detereministic system).

Things must be random because otherwise they are determined - and ****, we can't have that because humans must have happened in that case! Creationists would love that!

The fact that it changes nothing about the evolutionary explanation doesn't seem to get through at all. You simply can't bring yourself to agree with this can you jimbob?

 

[jimbob]

I am not sure what cyborg's point is.

I agree that evolution is the explaination, it is just that descibing the situation as nonrandom, and denying the significance of random events is also wrong.

Random effects are probably of little importance over timescales of millions of years, but over longer timescales, they are likely to become more important.

The random events alter the selective landscape, in other words (and grossly simplifying) they change the selection criteria randomly.

If the selection criteria are subject to random alteration over geological timescales, then saying evolution is nonrandom*, is misleading.

SIngle words are going to be wrong.

As Sol has said it can't be explained in a single short soundbite.

Articulett, would you agree, that over geological timescales, random events have played crucial roles in the evolutionary course of the ancestors of almost any organism alive now?

Sure - at any moment a giant asteroid could come along and smash the earth into tiny pieces, killing all life but some anaerobic bacteria (and thus rather drastically changing the course of evolution). Or intelligence could evolve, develop thermonuclear weapons, and engage in a nuclear war. Or the sun could capture a wandering star and go nova. Or a meteor strike could make all the giant lizards go extinct, allowing some furry rodents to take over the world and eat lots of chicken.

But you could say the same things about your pendulum - you claimed it was deterministic and predictable, but what if someone comes along and yanks it off its peg on a cosmic-ray-to-the-brain inspired whim?

The whole discussion is ridiculous. Everybody here (just about) agrees on the phenomena in question - we're just arguing over a word. But this is a complex process, and you just can't capture it very well with one word - you have to say a few more. Tough.

But the point is that these events have happened in evolutionary history, whilst not in any of the pendulum demonstrations I have seen.

*which to me implies that there is nothing important about randomness

 

[Earthborn]

While Evolution could be random in the sense of quantum uncertainty, it would be invalid to criticize it as random

I think everyone in this thread agrees with that. Calling evolution "random" is not criticism. It is describing how it works, for some definitions of random.

It tries to predict emergent patterns in life forms, as precisely as possible

That depends on what aspects of evolution one is studying. Many paleontologists are not predicting emergent patterns, but rather describing what paths evolution took in the past.

they just refuse to convey it a way that is understandable to those who don't understand evolution...

You keep saying that, but you haven't shown why calling evolution "non-random" makes it easier to understand to those who don't already understand it, and neither have you shown why calling evolution "random" hinders understanding.

It is the same as insisting that a random generator is, itself, random...
Why would you do that if you were trying to convey what a random generator is or how it uses code to generate random numbers...

As Walter Wayne said, you are talking about pseudo-random number generators. There are also "true random number generators". A random number generator is not necessarily code, it can also be hardware. A roulette wheel is also a random number generator, and I think everyone agrees that roulette is random (which of course does not mean it is unreasonable to predict that the wheel will go round instead of move in random directions; just because there are elements that are non-random does not make the process non-random)

 

[jimbob]

they just refuse to convey it a way that is understandable to those who don't understand evolution...

You keep saying that, but you haven't shown why calling evolution "non-random" makes it easier to understand to those who don't already understand it, and neither have you shown why calling evolution "random" hinders understanding.

Especially if anyone askes you about whether the KT event was important in the evolution of humans?

"Evolution is nonrandom but random events can alter evolutions course"?

 

[cyborg]

I am not sure what cyborg's point is.

You're obsessed with large red fish.

 

[Earthborn]

Well I find it odd that jimbob doesn't, to the best of my recollection, have any particular issue with genetic algorithms despite the fact that they don't really have much choice on the whole other than to deal with pseudo-random numbers. (I.e. it is technically a completely detereministic system).

Genetic algorithms would work just as well with true random number generators, it is just that would make little practical difference.

Genetic algorithms are usually used to evolve code towards specific predesigned criteria, and the code that that is declared unfit is not bred at all. It is usually the equivalent of artificial selection. People using genetic algorithms are breeding digital lifestock, killing or castrating any beast that does not fit the criteria.

With real natural selection, the criteria for fitness are never entirely clear and are constantly changing in subtle ways and not so subtle ways, and there may be many different evolutionary strategies that can be followed. And just because one individual organism is less effective at reproducing itself does not necessarily mean it is not reproducing at all.