Randomness in Evolution: Valid and Invalid Usage

[mijopaalmc]

zosima-

It is very easy to accuse someone of making things up when you in fact don't understand the topic. So I would appreciate you telling me where you feel I have made things up so that I can explain to you the sources of my knowledge.

 

[zosima]

Look mijo, I don't want to argue with you about whether or not you answered the questions. You said some things that were really quite unclear and I asked you for clarifications several times. That was my attempt to understand, I can't read your mind, so if you don't answer my questions directly, it is very difficult to communicate. If you don't explain why you say things, I assume they're false.

It seems you claimed that that the use of the word "random" in evolutionary biology is somehow inconsistent with the use of statistics to prove their hypotheses. I wanted to understand your claim. Telling me that evolutionary biologists compare statistical experimental data to the null hypothesis does not tell me anything about how you have inferred an inconsistency. It is just a definition of one of many statistical techniques used in many sciences. So at best this is a very oblique, inadequate answer to my question. But a more pragmatic and realistic assessment is that you chose not to answer and instead chose to recite a definition. This makes suspicious that your original statement lacked merit. To put in terms that are more apropos to the subject matter: In absence of any evidence to the contrary, I'm forced to believe the null hypothesis about your credibility with respect to the statements I queried upon and am tempted to generalize to all your statements.

So I'll ask it again, what did you mean by this?

The problem is that statistical hypothesis testing (even in maximum parsimony methods, especially when the number of taxa is above 8) used in evolutionary biology has a much wider usage of random that the aforementioned one, and it is therefore inconsistent to state that evolution by natural selection is non-random while making the assumptions of randomness necessary to perform any number of statistical hypothesis tests.

and this:

I disagree I agree that using the definition of "random" as "equiprobable" is consistent with all the statistical assumptions that evolutionary biologists make to demonstrate evolution.

Note the bolded text. See the ambiguity?

My second question, also unanswered. Lets say we have a system where we are quite certain about the details that we care about on a given level of description, but are not certain about details on subordinate levels of description. Do you understand how some people might find calling this system random a little bit misleading?

For example. I can say with a high degree of certainty that my cat is on my lap. Am I 100% certain of my cat's position? No. Am I aware of what the internal components of my cat are doing? No. Is it misleading to say my cat's position is random? I would say yes. Do you agree? Do you understand why I think this? Do you see how this applies to the subject matter?

I was just trying to assess if you understand where everyone else on this thread is coming from or if you genuinely are confused by why we are so insistent that calling evolution random is, at best, misleading. If you don't understand I see a reason to continue, if not I'd be happy to stop. I'm not trying to trick you. I'm just trying to assess whether it is beneficial to continue.

Also,
Mijo, forcing me to ask a question four times just to get a response is asinine and frustrating. Then you claim that because I'm trying to make an attempt to understand what you are saying, I must not understand the subject matter. That is unreasonable. I was trying to communicate to; be polite, but don't try to imply I don't understand statistics, Mijo. Remember just 3 pages ago you demonstrated a complete ignorance as to how certainty applies to probability and the differences between a finite set, a countably infinite set, and an uncountably infinite set. To say that I'm going out on a limb to be nice,to respect your position regardless of how well you understand the subject matter, is an incredibly generous understatement. So please, lets not play the who knows what game. I would just appreciate it, if you could directly answer my questions. I promise Mijo these aren't trick questions. You only make yourself look either silly or paranoid when you dance around them.

 

[zosima]

In lieu of any response, again, I think we can consider this thread over. Mijo's last minute assertions about evolutionary biologists misusing their statistics are conceded as completely without basis in evidence or reason.

 

[mijopaalmc]

zosima-

I'm just a little baffled about how you could not understand that statistics are based on probability distributions of samples statistics.

Have you taken a basic statistics course?

 

[mijopaalmc]

Heres' the deal:

The point of inferential statistics is to make a statement of about a descriptor of a population without having to measure said descriptor for every individual in the population. In other word, inferential statistics strive to describe an unmeasured parameter of the population using a measure statistic of the sample. Now, regardless of the specific statistical test chosen, that sample statistic thus calculated follows its own probability distribution, giving one a probability of finding data that is as extreme as or more extreme than than the observed data purely by chance. Now, this entire process depends on the fact that the statistic calculated from the sample data is in fact a random variable and can take on many values depending on certain characteristics of the the data, which in turn requires a broader definition of "random" than "equiprobable".

 

[jimbob]

Is there agreement that if the environment is stable, then there will be adaptation to that environment?

My contention is that some of the forms of adaptation that take place will be random, and these will close up certain avenues, and open others.

On an aside about to Zosima's statement about the eye.

As far as I can think of, arthropods alone have evolved at least three different types of "advanced" eye:

• The lobster eye (collimators, which is the approach used for X-ray optics BTW).
  
• The insect compound eye
  
• The spider eye(s) which I think are simple eyes with lenses

Then there are the mollusc eyes (the scallop being a bivalve must have evolved its vision system seperatelyto the cephalopods)

The point being that all these variants "solve" similar problems, but the solution is effected by completely different means.

ETA.

I would consider it valid to talk about the separate evolution of the eye as either random or (almost) inevitable, depending on the level of detail. that there are eyes is very likely; what form this eye will take in any particular organism is dependent on which set of utations occuced first, and which initial direction the evolution took.

 

[zosima]

zosima-

I'm just a little baffled about how you could not understand that statistics are based on probability distributions of samples statistics.

Have you taken a basic statistics course?

I have taken courses in statistics as well as higher math. But remember we decided to drop the ad hominems about who knows want because we don't want you to be embarrassed. All I would ask is that you make your point clearly, there is no reason to be incredulous about people's inability to understand you. I've asked several times what you mean.

Whats really frustrating, is that I ask a question nicely, and you refuse to answer, I wait for a while, you ignore it. You won't just answer a question, I knew that if I added that little post afterwards stating the issue was closed that I could get you to come back and answer. You really are quite predictable.

So thanks for finally getting around to answering my question. Let me summarize your argument. Evolution is modeled using random variables. The term 'random variable' has the word random in its name. Thus evolutionary biologist are wrong to call evolution non-random. I think its pretty easy to see the flaw in this argument. The technique of using random variables to model our imperfect knowledge of a process doesn't say anything about the process itself. For example in the Monty Hall problem we use random variables to describe the probability of winning, but the actual process is predetermined. Nothing changes because of the way we model it. It seems obvious to me that there is a difference between a random variable and a random process. Evolution as a science uses the former, evolution is not the latter. We're done.

On eyes:

I would consider it valid to talk about the separate evolution of the eye as either random or (almost) inevitable, depending on the level of detail. that there are eyes is very likely; what form this eye will take in any particular organism is dependent on which set of utations occuced first, and which initial direction the evolution took.

Man, I was hoping this thing was over. I'm worried that if I respond, this thing will blow up again. But here it is.

I see what you're saying, but characterizing evolution as reaching a point where its like "heads: compound eye, tails: vertebrate eye" is a little off base. There are two separate issues. One different types of eyes are better suited for different conditions. Two, once an eye shows up it is conserved. So after evolving they may be modified by evolution to suit purpose that would actually better served by the eye available in another taxonomic branch.

Looking at arthropods:
At some point the originator of the arthropods developed photoreceptors. The environment pushed it towards a compound eye. The reason it went this way probably had to do with balancing all sorts of different constraints, developmental architecture, effective sensation, minimal energy requirements. Once it showed up it stayed. One type of spider eye and the lobster eye developed from the compound eye. The other type of spider eye is a pretty basic photoreceptor. I have no idea why spiders at one point in their evolution required two types of eyes, but apparently they did.

We can see how these different solutions today suit different purposes. Spiders catch their prey without site and only really need vestigial eyes. Lobsters need to be able to see in very low light environments so they developed the reflective eyes that maximize their capacity to see in low light conditions. The compound eyes have a huge amount of redundancy. If you look at many insects, it seems clear that producing many redundant copies seems to be their ecological niche.

The point though is that saying that it "really could have gone either way", is conflating the importance of trait conservation and the long history of a species, with randomness. At each point traits develop due to the pressures at that time and they accumulate over a long history. While each individual generation is subject to randomness, over a number of generations the law of large numbers applies and the path ends up going the same way, regardless.

If there are any points in evolution that are random, in the sense that things can go either way, they are few and far between

 

[jimbob]

I see what you're saying, but characterizing evolution as reaching a point where its like "heads: compound eye, tails: vertebrate eye" is a little off base. There are two separate issues. One different types of eyes are better suited for different conditions. Two, once an eye shows up it is conserved. So after evolving they may be modified by evolution to suit purpose that would actually better served by the eye available in another taxonomic branch.

{nitpick}
The eye is only conserved where there is a selective advantage. Loss of sight in lightless environments is also inevitible.

I also don't like the phrase "So after evolving they may be modified by evolution to suit purpose" Although I do sometimes use similar phrases. To me this implies that there is a "guided direction" to evolution that I would dispute. Evolutionary Algorithms in engineering are to a purpose, but evolution itself isn't.

I think it can hide a bit of the story.

Here is an example of what I mean:

Devil's Advocate: ""mammals evolved fur from modified scales to regulate temperature" or "fur evolved because modified scales with fur-like features gave a selective advantage of improved temperature regulation"..."



{/nitpick}

I am also saying that there are also positive feedback loops: the physical/biochemical trait facilitates certain behavioural traits or lifestyles, which in turn modify the selective pressure. An animal with rudimentary sight might begin to use such sight for hunting/scavenging, one without this wont. There are a whole set of niches opened up by the presence of the light-sensitive cells.


I would disagree about the "rudimentary" spider sight:

Different versions seem very well adapted:

especially jumping spiders:

When hunting, the eyes of jumping spiders see in three different ways, using three different sets of eyes:

• The spider first senses movement of distant prey with the side eyes (PLE), which provide a blurry wide-angle image.
• Once movement is detected, the spider turns in that direction and locks onto the moving prey with the large, middle front eyes (AME). These eyes provide a clear, focussed telephoto image, probably in colour. The spider can track moving prey both by body movements and by using muscles to internally swivel the elongated eye capsules so that the light sensitive retina of each eye remains locked on the prey.
• While the spider stalks closer, it uses the side front eyes (ALE) judge the distance to the prey. When it judges the prey to be close enough (about 2 - 3 cm), the spider leaps

There are several other types of vision system within arthropods.

And at least two different types of system within molluscs. For example the scallop has a "solution" that resembles a compound eye in function if not in form.

That adaption will occur is inevitable, what adaptions, and how; indeed, which niches are created and occupied are not.

If this wasn't the case then ecosystems would tend to look very similar once re-stabilizing after each extinction event. Now you could argue that this is the case to a certain extent, rhinos and triceretops are/were both large herbivores, but there are also significant differences. Only one could possibly have eaten grass. Indeed the emergence of grasses must have had a disruptive effect on the whole ecosystem. Ditto flowering plants.

Life got along without flowering plants for hundreds of millions of years, so I would consider that there must be a window of tens of millions of years when such plants could have emerged.

If you are more interested in the differences, then I guess you are more prone to talk about the random aspect, if you are more interested in the similarities, then the inevitible aspect is more interesting.

Completely denying one or the other is wrong. Being an engineer with a physics background, and dealing with highly probable but still random events also pushes me towards emphasising the random aspect.

 

[mijopaalmc]

I have taken courses in statistics as well as higher math. But remember we decided to drop the ad hominems about who knows want because we don't want you to be embarrassed. All I would ask is that you make your point clearly, there is no reason to be incredulous about people's inability to understand you. I've asked several times what you mean.

I have made my point clearly and you have ignored it. The practice of statistics is based on making inference about the relation between populations and sample or two or more different samples, which requires understanding that quantities calculated from a specific data set almost certainly will differ from quantities calculated from another data set even if it is taken from the same population. This therefore necessitates comprehending that making a statement about the data can only be done by considering how the sample statistic is distributed in a large number of similar experiments. That you seem not to think that the basis for the practice of statistics is somehow not important to how statistics are used implies a lack of knowledge about the subject. Simply stating this idea is not an ad hominem because it goes directly to the ability from true, valid, and sound argument about the the topic. In fact the questioning of knowledge and understanding is a tactic that has frequently been used against creationists on this board.

What makes it a valid strategy against creationism?

Whats really frustrating, is that I ask a question nicely, and you refuse to answer, I wait for a while, you ignore it. You won't just answer a question, I knew that if I added that little post afterwards stating the issue was closed that I could get you to come back and answer. You really are quite predictable.

I was actually checking to see if I could find evidence that the statement I made could be corroborated by the scholarly literature for all kinds of statistics (both parametric and nonparamentric), and indeed it can for a wide variety of statistics, for example:

1. the Anderson-Darling test
2. the Friedman two-way analysis of variance
3. Kendall's τ
4. the Mann-Whitney U test
5. the Wald-Wolfowitz runs test

So thanks for finally getting around to answering my question. Let me summarize your argument. Evolution is modeled using random variables. The term 'random variable' has the word random in its name. Thus evolutionary biologist are wrong to call evolution non-random. I think its pretty easy to see the flaw in this argument. The technique of using random variables to model our imperfect knowledge of a process doesn't say anything about the process itself. For example in the Monty Hall problem we use random variables to describe the probability of winning, but the actual process is predetermined. Nothing changes because of the way we model it. It seems obvious to me that there is a difference between a random variable and a random process. Evolution as a science uses the former, evolution is not the latter. We're done.

The above is a mischaracterization of my argument. Evolution by natural selection is random because it is possible, as Bennett and Lenski showed, to start with several nearly identical populations and expose them to identical environmental conditions and have them each experience a different change in fitness and, more importantly, a different trade-off in fitness.

Your analogy to the Monty Hall problem seems porrly formed, because the the probabilities in the Monty Hall problem are known to arise directly and solely from the contestant's lack of knowledge of the initial arrangement, while it is far from certain that the perceived randomness in evolution by natural selection arise solely from our lack of complete knowledge of the initial conditions.

 

[zosima]

{nitpick}
The eye is only conserved where there is a selective advantage. Loss of sight in lightless environments is also inevitible.

I also don't like the phrase "So after evolving they may be modified by evolution to suit purpose" Although I do sometimes use similar phrases. To me this implies that there is a "guided direction" to evolution that I would dispute. Evolutionary Algorithms in engineering are to a purpose, but evolution itself isn't.

I think it can hide a bit of the story.

Here is an example of what I mean:

Devil's Advocate: ""mammals evolved fur from modified scales to regulate temperature" or "fur evolved because modified scales with fur-like features gave a selective advantage of improved temperature regulation"..."



{/nitpick}

Fair enough, Its easy to use language that implies design, but really stands in for a much more complicated phrase. I don't really like having to add the caveat every time I talk about evolution.

I am also saying that there are also positive feedback loops: the physical/biochemical trait facilitates certain behavioural traits or lifestyles, which in turn modify the selective pressure. An animal with rudimentary sight might begin to use such sight for hunting/scavenging, one without this wont. There are a whole set of niches opened up by the presence of the light-sensitive cells.

That adaption will occur is inevitable, what adaptions, and how; indeed, which niches are created and occupied are not.

If this wasn't the case then ecosystems would tend to look very similar once re-stabilizing after each extinction event. Now you could argue that this is the case to a certain extent, rhinos and triceretops are/were both large herbivores, but there are also significant differences. Only one could possibly have eaten grass. Indeed the emergence of grasses must have had a disruptive effect on the whole ecosystem. Ditto flowering plants.

Life got along without flowering plants for hundreds of millions of years, so I would consider that there must be a window of tens of millions of years when such plants could have emerged.

That adaptation opens new niche's is unpredictable and we might even say that it is 'fortunate',I agree, but I don't see how you can call that random. Ecological systems reach very stable equilibriums. When they are disrupted by things like extinction events, amongst other things, these equilibriums shift and evolution encourages adaptation.

For sure there are feedback loops, particularly on short time scales. But this too doesn't imply that anything is random.

The fact that they end up not being identical after an extinction event doesn't imply that this is due to a random component. It reflects the fact that different creatures with different histories and different conserved traits are adapting to same pressures. Really I've addressed it previously several times. If you took the same creatures and placed them into those niches twice, maybe there would be some minor differences, but overall they would look very similar. Actually the experimental evolution paper that was addressed earlier in the thread shows this.

I would agree evolution is unpredictable, chaotic, historically contingent, and complex. That doesn't mean random. If two systems follow the same rules to go from different starting points to different stopping places there is nothing random about it.

I would disagree about the "rudimentary" spider sight:

Different versions seem very well adapted:

especially jumping spiders:

There are several other types of vision system within arthropods.

And at least two different types of system within molluscs. For example the scallop has a "solution" that resembles a compound eye in function if not in form.

If you are more interested in the differences, then I guess you are more prone to talk about the random aspect, if you are more interested in the similarities, then the inevitible aspect is more interesting.

I wouldn't argue any species is il-adapted to its niche, thats almost impossible if you believe in evolution. But spiders rely on other senses do to what they do. Your article says "Spiders usually have eight eyes (some have 6 or fewer), but few have good eyesight. " But the fact that there are different eyes doesn't support any sort of random conclusion, unless you ignore historical contingency, trait conservation, and think that there are no trade-offs between different types of eyes.

Completely denying one or the other is wrong. Being an engineer with a physics background, and dealing with highly probable but still random events also pushes me towards emphasising the random aspect.

I think what most people on this thread, including myself, have been arguing that they aren't equally true. One way of thinking about it is a much better characterization than the alternative. That evolution, too, is a system of highly probable events. Where its not impossible that an alternative outcome occurs, but because we are talking about such long timescales so many individuals, so many mutations, that the probability that it works another way is vanishingly small. Thinking that it is somehow less reliable than these other systems, really doesn't do credit to the work that the scientists in the field are doing.

I'm not going to say that there haven't been events where the path of a species is determined by the outcome of some probabilistic event, just that its not bloody likely.

 

[zosima]

I have made my point clearly and you have ignored it. The practice of statistics is based on making inference about the relation between populations and sample or two or more different samples, which requires understanding that quantities calculated from a specific data set almost certainly will differ from quantities calculated from another data set even if it is taken from the same population. This therefore necessitates comprehending that making a statement about the data can only be done by considering how the sample statistic is distributed in a large number of similar experiments. That you seem not to think that the basis for the practice of statistics is somehow not important to how statistics are used implies a lack of knowledge about the subject. Simply stating this idea is not an ad hominem because it goes directly to the ability from true, valid, and sound argument about the the topic. In fact the questioning of knowledge and understanding is a tactic that has frequently been used against creationists on this board.

What makes it a valid strategy against creationism?

Your most recent ad-hominem was when you got baffled that I couldn't read your mind. No it isn't a valid strategy against creationism, though you are correct it is often applied. Lots of people make bad arguments on this forum that doesn't justify yours.

I was actually checking to see if I could find evidence that the statement I made could be corroborated by the scholarly literature for all kinds of statistics (both parametric and nonparamentric), and indeed it can for a wide variety of statistics, for example:

1. the Anderson-Darling test
2. the Friedman two-way analysis of variance
3. Kendall's τ
4. the Mann-Whitney U test
5. the Wald-Wolfowitz runs test

You can link to statistics papers like the best of em. None of these papers show that there is anything inconsistent with modeling a system statistically and calling that system non-random. That is the claim you tried to make.
They certainly don't make the specific claim that you make about evolution. So just linking a bunch of esoteric papers on statistics proves nothing. Well I take that back, it does seem to suggest that your claim that you were searching the literature is less than credible. If you had been searching the literature I would expect that you would either find some evidence that actually supports your claim. I guess it could also suggest that no one supports your claims.

Either way, you need to explain how the paper:
"On a Test of Whether one of Two Random Variables is Stochastically Larger than the Other" supports anything that you are saying. Remember, Mijo, we can't read your mind. If I could do that, I'd have won Randi's prize.

The above is a mischaracterization of my argument. Evolution by natural selection is random because it is possible, as Bennett and Lenski showed, to start with several nearly identical populations and expose them to identical environmental conditions and have them each experience a different change in fitness and, more importantly, a different trade-off in fitness.

Its a mis-characterization of your original argument, but not of the claim that
"it is therefore inconsistent to state that evolution by natural selection is non-random while making the assumptions of randomness necessary to perform any number of statistical hypothesis tests" It rebuts that claim directly.
What you are doing by switching back to the Bennett and Lenski paper is called shifting your argument. You lose one, then say you were actually talking about something else. However, since I already skewered you on that topic, I'll just provide a link to that post, and you can re-read why neither, Bennett, Lenski, or myself agree with you.

This is why the Bennett and Lenski disagree:
http://www.internationalskeptics.com/forums/showpost.php?p=3604464&postcount=448
This link is why your reanalysis of their paper is wrong:
http://www.internationalskeptics.com/forums/showpost.php?p=3605728&postcount=471

If you're going to shift your argument, then please shift it to something we haven't talked about before, otherwise you're jumping out of the frying pan and into the fire.

Your analogy to the Monty Hall problem seems porrly formed, because the the probabilities in the Monty Hall problem are known to arise directly and solely from the contestant's lack of knowledge of the initial arrangement, while it is far from certain that the perceived randomness in evolution by natural selection arise solely from our lack of complete knowledge of the initial conditions.

That demonstrates my point exactly. The random variables model a system in the Monty Hall problem yet the system is still deterministic. When statistics are used in empirical sciences we have no a priori knowledge about whether the statistics are modeling our lack of complete knowledge or some fundamental randomness. Thus it is not inconsistent, to model a system using random variables but still claim the system is non-random.

As to whether "...it is far from certain that the perceived randomness in evolution by natural selection arise solely from our lack of complete knowledge of the initial conditions." That is exactly the subject we've been discussing and your retreat on every other front suggests that its outcome is more certain than you'd like to believe.

My example supports my point, I'm sorry it doesn't support your point, but the day I start using examples that support other peoples' claims is probably the day I need to take a break from JREF.

 

[articulett]

Zosima, both have insulted me too. I have taken advanced courses and both statistics and probabilities. No one but themselves imagines they have expertise on this topic at all. Those who understand the topic understand what the actual experts are saying and why the imagined experts are full of crap. Those who need to believe something or other may be fooled temporarily by the self-appointed pedants. The self-appointed pedants, however, cannot or will not cede a point. They must see themselves as smarter and righter than you and Dawkins and Gould and Coyne and Sol Invictus, and Cyborg and anyone and everyone who dares to point out their failure at communicating intelligibly.

In their heads they must be geniuses... the fact that no one else seems to follow them or agree matters not one wit. In their heads, all of us cannot follow them because we don't have enough education or something. Interesting cognitive dissonance. It appears that the only people they are able to communicate their understanding with is creationists, but hey-- who needs to communicate with scientists and others when you imagine you are smarter than them all-- you know--"divine knowledge" and all that.

Both of them switch tenses mid definition in order to define selection as "random"... they say that since you can't "predict" with 100 percent accuracy who will live or not... it's "random" or "probabalistic"... which means "random" to mijo or them.

I was a genetic counselor. They are full of crap. Nobody but creationists are as obfuscating on the topic as they are. I'm not saying they ARE creationists... but they ought to consider employment at the Discovery Institute given their talents at failing to convey understanding of natural selection and their endless need to tell everyone that it somehow makes sense to call evolution "random".

 

[mijopaalmc]

zosima-

You're main problem is that you seem to see convergent behavior as evidence of of non-randomness. Unfortunately, such an approach would make everything non-random, because even systems that we can all agree are random (e.g., radioactive decay) can have orderly results (e.g., the functioning of an ionization smoke detector). The idea with evolution by natural selection is that, as demonstrated in the Bennett and Lenski paper (which you misinterpreted and continue to misinterpret), the constituent processes of evolution by natural selection, mutation and natural selection, is random but the results of the process, the adaptive optimization of an organism to its environment, are not.

I will deal with your other points within the next three hours, but thought that it was necessary to explain the fundamental objections I have to your approach immediately.

 

[Walter Wayne]

That adaptation opens new niche's is unpredictable and we might even say that it is 'fortunate',I agree, but I don't see how you can call that random. Ecological systems reach very stable equilibriums. When they are disrupted by things like extinction events, amongst other things, these equilibriums shift and evolution encourages adaptation.

I've addressed it previously several times. If you took the same creatures and placed them into those niches twice, maybe there would be some minor differences, but overall they would look very similar. Actually the experimental evolution paper that was addressed earlier in the thread shows this.

Well, I think the main difference you and I have, is the scale to view evolution on.

Those minor difference you mention may seem negligible, but they are base material used when the next shift in environment happens. As a result differences accumulate through subsequent events.

It seems odd to me to look at the short term evolution, when the principle point of attack is so called "macro"-evolution. That is, how it acts over the long term.

I would agree evolution is unpredictable, chaotic, historically contingent, and complex. That doesn't mean random. If two systems follow the same rules to go from different starting points to different stopping places there is nothing random about it.

Well, the question is what happens from the same starting point if we care principly about the technical randomness. However, for practical purposes if the system is so complex that evolution is unpredictable and chaotic then even if it were technically determistic, it would be random in the practical sense.

When classical physics "ruled" the universe, scientist thought coin flipping and die rolls were ultimately determistic, but it didn't stop them being used as practical devices for generating "random" results in a determistic universe.

In the end though, it looks like we agree on most things except the scale we are talking about.

Walt

 

[Walter Wayne]

zosima-

You're main problem is that you seem to see convergent behavior as evidence of of non-randomness. Unfortunately, such an approach would make everything non-random, because even systems that we can all agree are random (e.g., radioactive decay) can have orderly results (e.g., the functioning of an ionization smoke detector).

I have to disagree with you that it makes all systems we ce agree on as random, non-random.

A photon travelling through space or some sort of waveguide will yield an unpredictable result at the opposite end. That system is both technically random, and random in the practical sense. If I send a bazillion photons through the same setup, the results will match the prediction of classical physics. Sure there is a noise term, but for any practical purpose 10 Watts +/- a few attoWatts is predictable. The system is technically random, but in a practical sense it is non-random, just like your fire detector.

What I do agree with you on is that convergence is not evidence of non-randomness. Once you have heredity in a system, it is possible for it to converge, but when repeated see it converge to a different point.

Walt

 

[zosima]

@Articulett
True 'dat. I was trying to see if we could come to some agreement and was just curious about some things that were said, but apparently asking why he claimed evolutionary biologists don't know how to use statistics is evidence of my ignorance. It really is a pain to try to remain polite with someone like Mijo.

zosima-

You're main problem is that you seem to see convergent behavior as evidence of of non-randomness. Unfortunately, such an approach would make everything non-random, because even systems that we can all agree are random (e.g., radioactive decay) can have orderly results (e.g., the functioning of an ionization smoke detector). The idea with evolution by natural selection is that, as demonstrated in the Bennett and Lenski paper (which you misinterpreted and continue to misinterpret), the constituent processes of evolution by natural selection, mutation and natural selection, is random but the results of the process, the adaptive optimization of an organism to its environment, are not.

I will deal with your other points within the next three hours, but thought that it was necessary to explain the fundamental objections I have to your approach immediately.

Thanks for telling me what "my main problem" is. But thats whole point. No one agrees that a smoke detector is a random system because if it were, we wouldn't use it save our lives. It is a non-random device that has some random components.

As to Bennett and Lenski, you can assert that I misinterpret their results, but Bennet, Lenski, and I disagree. You never addressed my refutation of your arguments in the first place and you still haven't. So until you look at those, there is no reason why anyone should believe your blind assertions. But one thing I will say, convergent behavior only strengthens the conclusion that evolution is non-random, but it would still be non-random in its absence. It is superlative evidence, not necessary evidence.

I'm not sure how you come to the conclusion that my definition of well...my definition of something makes everything non-random. Other than maybe the fact that you've defined all events as random, so it seemed like a good strategy to invert the statement.

Also, don't bother getting to my other arguments. I wanted to understand why you claimed that evolutionary biologists are not understanding the assumptions of their statistics or how you could claim that they're being inconsistent. You've shifted to something else, something that is already plenty well discussed. The Bennett Lenski argument is long since over. You lost. And you don't get a redo. So rather than talk about this topic again. And make the same arguments again. Lets just leave it be.

@Walter Wayne,

When I said minor, it probably would have been better to say inconsequential. They're minor deviations in a path but it goes to the same place either way. For example, a mutation might show up in individual 4 instead of individual 7, different individuals, but if the mutation is beneficial it spreads exponentially. Maybe the same amino acid gets coded using a different equivalent sequence. Those changes don't aggregate, they cancel.

I would agree that there is a lot of practical randomness. We use probability all the time to describe our limited knowledge, but its often that people conflate that with true randomness. The major point people are arguing for here is that the more scientists study evolution the more regular its processes appear. The more evolution looks like it follows a set of laws, rules, and regulations, just like other sciences, and that amongst other things suggesting otherwise tends to do their work a discredit.

I'm totally willing to agree to disagree on the level at which things become non-random. I really just wanted to figure out what Mijo meant about the inconsistency and not blow this whole thing up again.

 

[mijopaalmc]

Thanks for telling me what "my main problem" is. But thats whole point. No one agrees that a smoke detector is a random system because if it were, we wouldn't use it save our lives. It is a non-random device that has some random components.

You are missing the point entirely. The processes of evolution, mutation and natural selection, are random while the result of evolution, adaptive optimization to the environment, is not. Similarly, the process that make an ionization smoke detector work, radioactive decay, is random, the result, the actual functioning of the detector is not.

As to Bennett and Lenski, you can assert that I misinterpret their results, but Bennet, Lenski, and I disagree. You never addressed my refutation of your arguments in the first place and you still haven't. So until you look at those, there is no reason why anyone should believe your blind assertions. But one thing I will say, convergent behavior only strengthens the conclusion that evolution is non-random, but it would still be non-random in its absence. It is superlative evidence, not necessary evidence.

No, convergence does not strengthen your case that evolution by natural selection is non-random. The observation convergence happens whether or not the underlying processes are random; otherwise the functioning of an ionizing smoke detector would prove that radioactive decay was non-random, which contradicts most of quantum mechanics.

I'm not sure how you come to the conclusion that my definition of well...my definition of something makes everything non-random. Other than maybe the fact that you've defined all events as random, so it seemed like a good strategy to invert the statement.

The insistence that convergence make a system non-random is what I was referring to, and, as mentioned above convergence does not imply that the underlying mechanism is non-random.

 

[jimbob]

zosima,

Even though extinction events are rare, they have been instrumental in determing the course of evolution on this planet. I am also not sure about the long-term stability of ecosytems either. We are only about 10,000 years from the end of the last Ice Age, where environments were significantly different.

Earlier than that, it seems that the Toba eruption 70,000 years ago reduced the human population dramatically (figures of between 1,000 and 10,000 breeding pairs). A quick google search suggests that this level of poplulation would qualify for "endangered species" status. Obviously humanity survived that event, but with population levels as low as that, chance begins to become more significant.

Two significant events within 100,000 years seems like quite a high rate to me, but again it depends on whether you are interested in the similarities or the differences.

If you are talking about the evolution of humanity, which is what many people are most interested in, then random (selective) factors have been very important. This would also be the case for virtually any other particular species.

Once the hominids had begun to emerge, there might have been an inevitibility about the emergence of something akin to humanity. However I would contend that this emergence and the "opening" of the niche was a result of the environment interacting with appropriate precursor animals at the right time, and with the right traits.

Something was going to evolve, and whatever did evolve, it would have shown amazing levels of adaptation to the environment, and modulated the slective environment for other organisms. Just because humanity evolved, doesn't mean that anything special happened.

 

[Belz...]

The above is a mischaracterization of my argument. Evolution by natural selection is random because it is possible, as Bennett and Lenski showed, to start with several nearly identical populations and expose them to identical environmental conditions and have them each experience a different change in fitness and, more importantly, a different trade-off in fitness.

"Nearly" identical ?

You are missing the point entirely. The processes of evolution, mutation and natural selection, are random while the result of evolution, adaptive optimization to the environment, is not.

That's a very nice way to weasel out of the admission that you were wrong. How is "adaptive optimisation" NOT part of the process ?

 

[Belz...]

No, convergence does not strengthen your case that evolution by natural selection is non-random.

How in the blue hell can you say that ? What's causing the convergence, then ?