What evidence is there for evolution being non-random?

[mijopaalmc]

Put as simply as possible: you can't decide fitness first. Nature defines what is fit, not you.

But this is the thing that you don't seem to understand: we're not trying to determine "fitness" by what we consider "fit". A relatively easy way of defining "fitness" is if the number of individuals of a given phenotype in the subpopulation that copies itself is more in a statistically significant way from the one that would be selected if each individual had an equally likely chance of being chosen, the given phenotype is "selected for"; however, if the number of individuals of a given phenotype in the subpopulation that copies itself is fewer in a statistically significant way from the one that would be selected if each individual had an equally likely chance of being chosen, the given phenotype is "selected against". This may not be the way that the writing aimed at a popular audience defines "fitness" or "natural selection", but the scholarly literature that addresses the stochastic or statistical aspects of evolution by natural selection is much more subtle is the way it addresses those aspects that the popular writings. That is why it is necessary to read both the original and current scholarly literature to get a grasp of what the researchers mean.

 

[steenkh]

Because there is a range of phenotypes where organisms can compete for resources to varying degrees, it is not certain that an two organisms that have identical sets of genes that make it easier for them to compete for resource will survive. That, by definition, is why evolution is a random process.

Excuse me for butting in again, but this is really not the correct way of looking at it. It is not the positive side (that some organisms sometimes survive) that is important, but the negative side (that some organisms invariably do not survive) that is important. A set of genes that make it more difficult to compete never make it. That, by definition, is why evolution is a non-random process.

 

[articulett]

So, articulett, if you think "randomness" has already been "worked into the equation" of natural selection, why do you resist calling natural selection random?

For the exact same reason we don't call Algebra a "random process" nor do we say the nozzles evolved randomly.

Random components do not a random process make.

If you don't understand this; you can't understand this.

 

[steenkh]

In physics there is something called stochastic dampening where you combine stochastic effects with a selection function. An example is putting excited atoms into a container with an open lid. Some of the atoms are energetic enough that they jump out of the container, and others stay. The result is invariably that after some time only the less energetic atoms remain - stochastic dampening. Even though a strongly random component is present, the result is predictable and non-random.

 

[portlandatheist]

I've only had time to skim through this thread. What I would suggest is to search google's videos for "The Blind Watchmaker" and watch the 48 minute video. Its totally amazing to see genetic algorithms do their magic and actually replicate things we see in nature all the time. Selection is anything but random and selection pressures change from generation to generation so what may be "fit" at some point may not be at other times.

 

[articulett]

In physics there is something called stochastic dampening where you combine stochastic effects with a selection function. An example is putting excited atoms into a container with an open lid. Some of the atoms are energetic enough that they jump out of the container, and others stay. The result is invariably that after some time only the less energetic atoms remain - stochastic dampening. Even though a strongly random component is present, the result is predictable and non-random.

Yes, the less energetic atoms are "selected" to remain in the jar. You wouldn't say they sorted themseleves "randomly".

I was listening to a podcast on that tonight...there are a lot of things that aren't random, but not completely determined either...or rather it depends on what your perspective is...like "free will"--clearly not everybody has the same will, propensities, temptations, etc. I understand those who say that nothing is truly random as well, as everything is determined in some manner by the environment it is in--so I think it's probably best to avoid the terms unless you have agreed upon definitions and perspective. At what point does something go from random to determined--?? The inevitability of any single person becomes very improbable as you go back in time, but once the sperm is just about to fertilize the egg, the resulting zygote has been determined. That sperm isn't necessarily the fittest in genetic info. by any human standard, but it is by virtue of the fact that it and only it got it's DNA copied into a vector while all the other contenders failed to pass the elimination rounds.

I don't mind saying that mutations or recombinations are relatively random--because no matter what definition you use, the idea is communicated--but natural selection isn't random in that way at all... A random smudge on a CD probably won't affect the data--and if it did, it would make the whole program fail... But you wouldn't include such randomness when you were trying to explain how the programs was honed...just as you wouldn't mention all the "random" changes the engineers didn't think of while making changes to nozzles or all the sperm that never fertilized an egg or any other random noise which detracts from the basic principle you are trying to convey--the fact that something IS selected--something passes through the elimination round...and to the extent that something in the information in the DNA (or blueprint in the case of nozzles) aided in that selection, it too gets passed on. Often multiple times...have tons of copies (lots of offspring) increases the odds that one of those vectors will carry the DNA into the future with new and improved mutations/ recombinations.

The wastage is built into the process. Not every acorn is meant to be an oak tree even if they are all equal in genetic terms. Life evolved to be "wasted" on a grand scale... but other living organisms from microbes to fungi to insects, etc. make use of these "unselected" by consuming these organisms furthering their own reproductive ability.

 

[articulett]

I've only had time to skim through this thread. What I would suggest is to search google's videos for "The Blind Watchmaker" and watch the 48 minute video. Its totally amazing to see genetic algorithms do their magic and actually replicate things we see in nature all the time. Selection is anything but random and selection pressures change from generation to generation so what may be "fit" at some point may not be at other times.

I agree. I saw a great video, and I think it was called the "blind watchmaker" where they used an algorithm to actually make watches...it was really cool...is that the one you are talking about?....I think the one that I saw was less than 10 minutes. I think it's uninformative and misleading at best to call selection random. But there are some people who insist on defining it that way for some reason or other.

 

[mijopaalmc]

Why do you insist on ignoring the fact that 85 years of research has essentially said the opposite of what you have said?

Why do you insist that Wright, Fisher, Moran, Haldane, Maynard-Smith, and Kimura are no longer relevant when performing cited article searches on their works, which claim that the stochastic interpretation of evolution by natural selection is accurate, demonstrate that they are still widely cited even in recent research?

It seems that you are putting undue emphasis on the works of Dawkins just because they support your opinion.

 

[steenkh]

Why do you insist on ignoring the fact that 85 years of research has essentially said the opposite of what you have said?

Why do you insist that Wright, Fisher, Moran, Haldane, Maynard-Smith, and Kimura are no longer relevant when performing cited article searches on their works, which claim that the stochastic interpretation of evolution by natural selection is accurate, demonstrate that they are still widely cited even in recent research?

It seems that you are putting undue emphasis on the works of Dawkins just because they support your opinion.

I think these researchers - if you are interpreting them correct - are using the term 'random' in another sense than what is used normally. Even if they show in their research that "unfit" genes in a few rare circumstances survive over hundred thousands of years, and that the selection is thus 'random', I still think that it is wrong to pin the 'random'-label on evolution as such.

This is simply not what normal people understand about 'random'. Evolution, like the dice, is loaded to the detriment of the unfit, and that is not what we ordinary non-mathematics types would call 'random'.

 

[Walter Wayne]

In physics there is something called stochastic dampening where you combine stochastic effects with a selection function. An example is putting excited atoms into a container with an open lid. Some of the atoms are energetic enough that they jump out of the container, and others stay. The result is invariably that after some time only the less energetic atoms remain - stochastic dampening. Even though a strongly random component is present, the result is predictable and non-random.

All the common examples used here where random events combine to create a predictable outcome have common traits. As an example, your atoms in an open container, what happens if in the first few moments fewer than expected atoms exit the box? Not much in the long term, as the more energetic atoms will still eventually leave. In other words, the "noise" of the random process is minimal considered to the "signal" that is the predictable outcome.

The same is true in the example of the pressure excerted on a box by the gas molecules inside. Given the large number of molecues, even a few hundred hitting a small area at once just show up as a small bump in the otherwise predictable outcome. If one had an instrument sensitive enough to detect the difference of a few more or less atoms hitting the sides than variation would still just register as an insignificant blip on the overall average.

Compare this to natural selection, where generations are linked by heredity. Now at some times it seems to be similar in nature to the above processes. When a population is large and robust things like a disease mutating and jumping to the population become "noise". The population loses some individuals, but they are replaced and the species continues with little change in make up, and many of living individuals having a resistance to the disease. However, in a small population random events can drastically affect the makeup of the species, because there are so few "trials" that the resulting variation can be significant. In these small populations the species could even be pushed over an edge to extinction.

This is where the difference in the two systems come in. A few atoms in the box not doing the "expected" at a specific time is over-shadowed in the long term. A species dying out cuts off all future possible branches from that species, and affects the selection of competing species and the species that would have competed with its future progeny. In other words, in evolution these random events can affect the long term results significantly. How many bottlenecks in population have occured in the history of life on this planet? I think it is reasonable to say many.

It is important to understand what qualities of a stochastic system lead to it have predictable results. Atoms in an open container do not exhibit heredity, and that is one of the qualities of the system that lead to predictable results.

Walt

 

[Walter Wayne]

I've only had time to skim through this thread. What I would suggest is to search google's videos for "The Blind Watchmaker" and watch the 48 minute video. Its totally amazing to see genetic algorithms do their magic and actually replicate things we see in nature all the time.

A comment on genetic algorithms: the people who generated these algorithms worked hard to make sure that they find the best solution or that it finds all the solutions. In creating computational algorithms to find mathematical solutions there is the problem of converging to local mimima or not finding all solutions.

People who run genetic algorithms have and do run into this problem, and they have carefully tuned the algorithm to make it as predictable as it is now. The first genetic algorithms did not work nearly as well as the current ones. So why assume that the current more predictable algorithms more closely mirror natural selection?

 

[Meadmaker]

Excuse me for butting in again, but this is really not the correct way of looking at it. It is not the positive side (that some organisms sometimes survive) that is important, but the negative side (that some organisms invariably do not survive) that is important. A set of genes that make it more difficult to compete never make it. That, by definition, is why evolution is a non-random process.

The process is random. The result is not.


It is perfectly possible for a random process to have predictable results. If that were not the case, then I wasted a lot of time and money studying random processes in graduate school.

 

[steenkh]

When a population is large and robust things like a disease mutating and jumping to the population become "noise". The population loses some individuals, but they are replaced and the species continues with little change in make up, and many of living individuals having a resistance to the disease. However, in a small population random events can drastically affect the makeup of the species, because there are so few "trials" that the resulting variation can be significant. In these small populations the species could even be pushed over an edge to extinction.

In general I agree with your comments. In the case of small populations or other kinds of bottlenecks, the random element takes over and makes the process inpredictable.

This is where the difference in the two systems come in. A few atoms in the box not doing the "expected" at a specific time is over-shadowed in the long term. A species dying out cuts off all future possible branches from that species, and affects the selection of competing species and the species that would have competed with its future progeny. In other words, in evolution these random events can affect the long term results significantly. How many bottlenecks in population have occured in the history of life on this planet? I think it is reasonable to say many.

The random extinguishing of some species of course lead to new possibilities of other species, or alters other species' 'fitness' in some way. The situation is so complex that we usually only determine 'fitness' afterwards: those that survived were 'fit', or rather those that were extinguished were 'unfit'. I do not think that such random variations in fitness warrant calling the entire evolution a random process.

 

[Walter Wayne]

The process is random. The result is not.

The result is still random. Even if there are some possibilities that won't emerge, or won't continue if they do that doesn't make the result not random. In stochastic terms, a few nulls in the final probability distribution of the result doesn't make it not random.

Walt

 

[steenkh]

The process is random. The result is not.

Exactly.

It is perfectly possible for a random process to have predictable results. If that were not the case, then I wasted a lot of time and money studying random processes in graduate school.

I think that this whole thread is based on semanticism and not any real disagreement. If we look at evolution as the predictable result of some processes that are random to varying degrees, we could perhaps agree that evolution is not random?

Evolution is of course a process in its own right, but it is the result of the action of processes like mutation and selection. And the result of evolution is only predictable to the degree that biological species will evolve to avoid 'unfitness', the actual solution how they will gain fitness is still unpredictable, and hence in layman's terms, random.

 

[mijopaalmc]

I think these researchers - if you are interpreting them correct - are using the term 'random' in another sense than what is used normally. Even if they show in their research that "unfit" genes in a few rare circumstances survive over hundred thousands of years, and that the selection is thus 'random', I still think that it is wrong to pin the 'random'-label on evolution as such.

This is simply not what normal people understand about 'random'. Evolution, like the dice, is loaded to the detriment of the unfit, and that is not what we ordinary non-mathematics types would call 'random'.

You know, it is a little disconcerting when I have gone to great lengths to explain myself and people appear not to bother to read what I have written. I have been completely and explicitly clear that I am using "random" only in its sense "[o]f or relating to a type of circumstance or event that is described by a probability distribution", and the researchers I cited do describe evolution by natural selection using probability distributions.

Now, like articulett, you make think that such a definition is meaningless, but such an approach ignores, as I have said before, 85 years of research. Even the examples she uses do not make sense. Genetic counseling is based almost completely on joint probability distributions of random variables. For instance, given, inter alia, the respective ages of the parents at conception and the number of previous pregnancies the mother has had, it is possible to calculate the probability that the baby has any number of genetic abnormalities such as Down Syndrome. It is not necessary, or possibly even not helpful, to describe the method of calculation, which may itself be unknown to the counselor (and judging from the way articulett describes her experience as a genetic counselor, it seems to be); however, the utility of knowing and understanding the calculation methods to the parents and the counselor is irrelevant to the fact that they are based on the joint distributions of random variables. What bothers me most about the tack that the "non-randomites" take, other than articulett's steady stream of insults toward anyone who persists in disagreeing with her, is its lack of context. While I would only describe evolution as being "stochastic" or "probabilistic", which are synonyms for "random" in so far as they share only the definition mentioned above, if I had time to sit down and throughly explain what I meant and offer examples, I see no reason for the categorical denial that is so characteristic of the "non-randomites'" argument especially in light of the vast amount of research on the stochastic modeling of evolution by natural. Thus, I think that it is perfectly appropriate and true to describe evolution by natural selection as "stochastic" or "probabilistic" if the listener is capable and willing to understand.

 

[steenkh]

You know, it is a little disconcerting when I have gone to great lengths to explain myself and people appear not to bother to read what I have written. I have been completely and explicitly clear that I am using "random" only in its sense "[o]f or relating to a type of circumstance or event that is described by a probability distribution", and the researchers I cited do describe evolution by natural selection using probability distributions.

Did I claim anything else? The point that articulett has tried to get through is that using this definition which may make lots of sense to statisticians or mathematicians, simply is not how the rest of the world uses the term 'random'. And misunderstandings come from that. You may continue using the term in its strict mathematical sense, but if you do it in discussions with lay people, you are conceding the creationists a point, or at least making your task much harder.

Now, like articulett, you make think that such a definition is meaningless, but such an approach ignores, as I have said before, 85 years of research. Even the examples she uses do not make sense. Genetic counseling is based almost completely on joint probability distributions of random variables. For instance, given, inter alia, the respective ages of the parents at conception and the number of previous pregnancies the mother has had, it is possible to calculate the probability that the baby has any number of genetic abnormalities such as Down Syndrome. It is not necessary, or possibly even not helpful, to describe the method of calculation, which may itself be unknown to the counselor (and judging from the way articulett describes her experience as a genetic counselor, it seems to be); however, the utility of knowing and understanding the calculation methods to the parents and the counselor is irrelevant to the fact that they are based on the joint distributions of random variables.

Not being a specialist myself, I can accept this completely.

Thus, I think that it is perfectly appropriate and true to describe evolution by natural selection as "stochastic" or "probabilistic" if the listener is capable and willing to understand.

But are we not talking about an audience who do not share this understanding, and thus are not capable to understand you unless you invest a lot of time in explaining?

This is not a question of denigrating 80 years of research but of using the right words to convey the research to the public.

 

[mijopaalmc]

Did I claim anything else? The point that articulett has tried to get through is that using this definition which may make lots of sense to statisticians or mathematicians, simply is not how the rest of the world uses the term 'random'. And misunderstandings come from that. You may continue using the term in its strict mathematical sense, but if you do it in discussions with lay people, you are conceding the creationists a point, or at least making your task much harder.

I think you are missing that articulett has rejected every alternative word I have offered. The description "random mutation with natural selection" is also inaccurate and misleading because it doesn't explain how fitter genes are selected. In fact, articulett has been using a circular definition of "fit" (i.e., what's fit is copied more and what's copied more is fit) when she is not trying to avoid using the concept, which is recognized through out the scholarly literature as having validity. In essence, articulett doesn't seem to want to admit that the stochastic aspects of evolution are central to its understanding.

But are we not talking about an audience who do not share this understanding, and thus are not capable to understand you unless you invest a lot of time in explaining?

This is not a question of denigrating 80 years of research but of using the right words to convey the research to the public.

Actually, it is about the research because, regardless of how evolution is portrayed to the public (which isn't really the subject of the thread, though articulett would like it to be), the stochastic models still exist and still describe evolution remarkably well for models that are based on the supposedly flawed assumption that evolution is a stochastic process.

 

[Meadmaker]

The result is still random. Even if there are some possibilities that won't emerge, or won't continue if they do that doesn't make the result not random. In stochastic terms, a few nulls in the final probability distribution of the result doesn't make it not random.

Walt

My mistake. Let me revise.

Many outcomes have probability 0. Other outcomes have some probability less than 1. The outcomes in the latter category all share certain characteristics (which we lump together as "fitness"). So, the probability that the final outcome will involve something "fit" is 1.

 

[steenkh]

In essence, articulett doesn't seem to want to admit that the stochastic aspects of evolution are central to its understanding.

Most of the discussion has been over my head, but this is not how I have understood articulett.

... regardless of how evolution is portrayed to the public (which isn't really the subject of the thread, though articulett would like it to be) ...

This is exactly how I have understood the thread right from when it was branched out from another thread. The 'Tornado and junkyard" example has been quoted early, and this is about how to counter creationist arguments. I think this is what articulett has tried to get through, but I will leave it to herself to say so ...