What evidence is there for evolution being non-random?

[articulett]

No.



I have a master's degree in electrical engineering. I took two classes specifically on probability. One for advanced undergraduates, and one for graduate students. In studying optimal control and non-linear dynamics I had to study random processes. Professionally, I did work on signal processing.

I've never taken a course specifically on statistics, and the only way I can reconcile what you are saying with my work is that the definitions used in statistics must be different than those used in probability.

Well then, what does random mean in regards to probability in the engineering world? I took probabilities as well as statistics... You do understand how even a random number generator doesn't produce truly "random" results--more "pseudo random"--but as random as we can get via a non-random processor (a computer). Right? Did you read the fly link... about how something can be both non-random and non-predictable--we call it chaotic--chaos theory--not random... random actually means that all probabilities are equal (or at least relatively so)--not just that they are unpredictable. So what is your definition...and, once again, why use the word when you aren't even on the same page as biologists and we're talking about evolution?

I'm no slouch--and neither is Paul, Dawkins, Talk Origins, nor the New Scientist article I linked. You sure demand a lot of evidence and defining terminology for one who isn't exactly clear on who is, admittedly, unfamiliar with creationist tactics--and someone whose expertise lies outside the subject we are discussing. A little humility is order, don't you think? You've asked for a lot of examples of creationist language abuse, dismissed the information from those more familiar with the subject, suggested I may not be informed and that I may be over reaching in my conclusions when all of these issues might be better aimed at you. You're the one who thinks random is not an issue in the creationist debate while not knowing much about the wedge or anything in it or without dealing with these people regularly...you're the one claiming random means something nuanced and probabilistic without a single cohesive definition except whichever one suits whatever it is you want to claim. You asked of my training before revealing your own and after dissing Dawkins and others who are more versed on the subject (that is, they understand the intricacies of evolution) better than you--an all because you think you know what mijo means when he says random...

 

[articulett]

You had a class called "random processes"? Do you mean this:

stochastic process

In probability theory, a family of random variables indexed to some other set and having the property that for each finite subset of the index set, the collection of random variables indexed to it has a joint probability distribution. It is one of the most widely studied subjects in probability. Examples include Markov processes (in which the present value of the variable depends only upon the immediate past and not upon the whole sequence of past events), such as stock-market fluctuations, and time series (in which temperature or rainfall measurements, for example, are taken at the same time each day over several days).

Once again...random components do not a random process make. Moreover, the examples are not life forms and they don't "evolve"-- one is fluctuations not dependent on the whole event (unlike evolution) and the other is averaging weather phenomena...

Assortive mating (Mendelian Genetics --recessive and dominance etc.) is stochastic in that way--it's sorted out, in a random way...but that is involved in the randomness of mutation. In what way does selection apply to this...how is evolution even stochastic (which isn't random--it just contains random elements)...there is no joint probability distribution in evolution. And the whole sequence of what came before defines which strands of DNA make it to the next DNA elimination round.

Ugh. And all of this is why the word random shouldn't be used in relation to evolution--and why people who don't have expertise on a topic might want to cede the floor to those who do. Especially if the goal is actually understanding something rather than playing semantic games.

(Boy, do you NOT understand selection...)

 

[articulett]

No.

See this nice simple quick answering of the question, "are you a proponent of intelligent design?" Expect that people who feel as you do will answer the same way. All those who avoid the question or skirt the question or answer obliquely should give you pause. I'm not say they necessarily ARE "proponents"--I'm only saying that those on this forum who feel as you do, answer as you do.

(And I apologize for being snarky, but you did start it--you asked about my understanding of probabilities in a way that inferred that I didn't know what I was talking about...you question everyone who tells you "random" is a problem word in creationist "lingo", you verbally flog me for accusing mijo of being an "intelligent design proponent" (he is, I assure you...), and you are slow to cede a point to anyone...but, then again, engineers are known to be like that.
I think you are nice to stick up for mijo--but still wrong. I think you'd really enjoy learning exactly how selection isn't random--please look at the Sagan clip I linked. It's great. And it will make you see why applying "random" to selection is a recipe for making someone very muddy on the concept of both selection and the evolutionary process.)

 

[mijopaalmc]

(Boy, do you NOT understand selection...)

Then why not tell how selection works?

Could you tell us, for example, what is wrong with how the UC-Berkeley Understanding Evolution site describes natural selection?

That has been what I have been trying to summarize when I say that natural selection confers a probability of survival and reproduction on an individual. In fact, the site says under differential reproduction:

Since the environment can't support unlimited population growth, not all individuals get to reproduce to their full potential. In this example, green beetles tend to get eaten by birds and survive to reproduce less often than brown beetles do.

(emphasis mine)

The point is that you have repeatedly told us that we do not understand selection, but here we have a site hosted by a reputable university and funded by two prestigious grants that tells us essentially the opposite of what you have been saying. Please tell us why the probabilistic interpretation of the above statement (i.e., that individuals with low fitness are less probable to reproduce) is incorrect instead of telling us that we are wrong and don't understand.

 

[articulett]

Then why not tell how selection works?

Could you tell us, for example, what is wrong with how the UC-Berkeley Understanding Evolution site describes natural selection?

That has been what I have been trying to summarize when I say that natural selection confers a probability of survival and reproduction on an individual. In fact, the site says under differential reproduction:

(emphasis mine)

The point is that you have repeatedly told us that we do not understand selection, but here we have a site hosted by a reputable university and funded by two prestigious grants that tells us essentially the opposite of what you have been saying. Please tell us why the probabilistic interpretation of the above statement (i.e., that individuals with low fitness are less probable to reproduce) is incorrect instead of telling us that we are wrong and don't understand.

Yes, Berekely is great...and their evo source is excellent. And random means with equal probability and/or without regard to the past or future...while that can be applied to mutation or recombination--the word cannot be used to describe selection. Not all mutations are equally likely to be selected AND what is selected is wholly dependent on what came before. You will notice, that there is no mention of the word random. I absolutely agree that individual with low fitness will less probably produce...you don't seem to understand that the least fit will not. And, the most fit, by definition of the word fit as it applies to genetics--is the organism which has the most descendants which produce descendants. Varying probabilities is not a factor in randomness...unless you are aware of something I am not aware of. Even stochastic processes...which you can apply to recombination are still not "random processes"--rather they are processes that produce patterns with random variables which are connected to each other. Selection is not stochastic. Read the definition supplied or supply your own. I got my definition of random from a science dictionary...where are you getting yours?
And since the word is obviously problematic for you...and apparently for understanding just about everything... why would you insist on using it? And why would you ask for clarification about how something is non random when you seem to only be able to hear how it is? Moreover, you have some real problems understanding the basics of evolution--both selection and the fossil record. These are really important things to understand if your goal is actually to teach people or disabuse creationists of their claims. But I think we all know that that is not your goal. Your goal is to hear that evolution IS random. You're the one making the claim in complete disregard to every biologist here, on talk origins, etc.-- so why don't you give us your nuanced definition of "random" with a cite please...and a peer reviewed article saying that "evolution is a random process". Because you seem dead set on believing that is the case. It isn't. You are wrong. You do not understand selection if you think it IS a random process.

Read the link again, and get a clue as to what random means...lesser and greater probabilities are just lesser and greater probabilities--not aspects of randomness where all probabilities are equal. REALLY mijo--now you are really stretching to hear what you want to hear. Why don't you give up the word random until we are all on the same page as to what it means...

 

[mijopaalmc]

Random means pertaining to probability.

 

[Meadmaker]

Well then, what does random mean in regards to probability in the engineering world?

It means characterized by a probability density function.

The thing I can't reconcile is your comment that "random with non-uniform probabilities" was "gee...not random". What would you call a random number with a normal distribution? Or a Poisson distriution?

If a highly fit organism had a probability of 0.55 that it would pass on its genes, while a less fit organism had a probability of 0.45, would it not be "random" whether or not those organisms would pass on their genes? In my textbook that is still random, but not characterized by a uniform distribution function.

 

[Meadmaker]

You had a class called "random processes"? Do you mean this:

Random processes were elements in the classes I took. The graduate level class was titled "probability and random processes". The most common random processes I dealt with were systems of differential equations with randomly varying inputs or measurements. For example, devising an optimal control law for a system whose state variable measurements were modelled as having random measurement error.

 

[Taffer]

Just to clarify for this squishy science grad student, is the argument now only about the technical definition of "random" and "random process"?

Because I always thought "selection" was the opposite of "random".

But, y'know. Squishy.

 

[Paul C. Anagnostopoulos]

You answered your own question. If it involves any randomness, it is correct to call it random, even if it involves the largely (phrasing which allows for yet more random stuff) deterministic.

Yes T'ai, we all agree that evolution is a random process. The question is whether explaining evolution as "random" is helpful.

So how is it the "exact opposite of the truth", according to Dawkins, to call it stochastic/random then?

I suspect that when Dawkins says "evolution is not random," he is using random in the way it is commonly understood.

IMO that's just some wishy-washy English tacked on to things. Kind of a pseudo-explanation/weasel word kind of thing to pretend that stochastic suddenly does not mean random. Like referring to philosophical debates about whether everything in the universe is really deterministic or not, it is in the same genre of a pseudo answer to this scientific question.

I think that calling selection nonrandom with respect to the environment is quite helpful. Do you have a better way to say it? Or do you think that saying "selection is random" is accurate?

Should we say "mutation is random and selection is random with highly nonuniform probabilities"?

Once again...random components do not a random process make.

Actually, I think they do.

~~ Paul

 

[Paul C. Anagnostopoulos]

Let's say I had a die with the six weighted a bit. I throw it a bunch of times and come up with a disproportionate number if sixes. What would your reaction be if I said "This die is not random."

~~ Paul

 

[Dancing David]

If you think that I don't understand why natural selection is non-random, articulett, then explain it to me. I have given you the tool to find what I have read on the subject so that you can discern what I might be thinking about the topic, yet you seem not to have payed attention. Furthermore I just asked to provide a definition of "non-random" and you have yet to, but you still insist that I don't understand natural selection.

It would help if you would define what you mean by non-random as well, there have been many possibilities mentioned and discussed as one of the sub-plots of the thread.

You might seem to have not payed attention to those requests for clarification of your use of the word 'non-random'.

Dialouge requires communication from both sides in the discusion.

 

[articulett]

Let's say I had a die with the six weighted a bit. I throw it a bunch of times and come up with a disproportionate number if sixes. What would your reaction be if I said "This die is not random."

~~ Paul

I'd say your die was, indeed, not-random. It was weighted--hence not random. Random means having equal probabilities. I'd suggest that you use Vegas dice, if you wanted as close to true randomness as possible. (the die with holes means that the side with one dot, will be heaver than the side with 6 dots and tends to land side down--with one on the bottom and 6 on the top )

Random processes are actually "stochastic processes" wherein random elements are connected in a way that produces a pattern. Stochastic processes are processes, which describe a pattern or a series of action which IS oppisite of random in that random means without regard to the past or future.

To review: Random components do not a random process make. Random means: having equal probability Process: means a pattern or a series of actions... Selection (and your die) do not have equally probably outcomes--moreover, in both you'd expect to see a pattern--in the die example, it would be a preponderance of sixes; in evolution, it's not only "survival of the best survivors"--but it's 100% dependent upon that which cam before--another definition which makes selection--and thus evolution absolutely non-random.

Evolution can only be said to be random if you are using the term loosely to suggest purposeless.

But the fact that nobody agrees on this word...and that I have quoted the definition from a scientific text as well as linked a science article which describes how something can be "chaotic", yet not random should have clarified the issue as to how some of the random parts of evolution are not even random-- (mutation)...(recombination is stochastic)--much less selection. Unless someone provides a definition from some other source I think we ought to go with the common scientific meaning of the word, don't you. In which case, Dawkins is correct. Selection IS the opposite of random.
In respect to "random" mutation it's about as far away from the word "random" as you can get. Even if you are using the word random just to mean "unpredictable" we can predict with increasing certainty up to 100% that the least fit will not survive to reproduce more of themselves. And we can also predict that those that survive to pass themselves on are the "most fit" when it comes to reproduction (and by fit I just mean that they have what it takes to survive--even if luck is involved in them not being obliterated from the gene pool...) If a gene exists and can be copied it can count itself amongst the fittest. The same goes for genomes. The same goes for representative members of any genome.

But-- the fact that the word "random" has become the issue rather than the "non-random" aspects of evolution and understanding how selection is not random--I think it's far and away proven that using random to describe evolution is absolutely meaningless. I would use more precise words. It IS random only if you mean purposeless--or containing random elements--it most certainly is not random in respect to selection-- AND that is the key area where creationists have their sticking point. To them "random" means purposeless and unpredictable...from there they hop right over how selection builds complexity and cast off the scientific explanation as "how can complexity come from chance"? Every creationist argument depends on obfuscating facts about evolution...rather it's loosely defining words like "truth", "theory", "random", "adding info. (to the genome)", "discontinuity", "transitional" or something else.

The goal is to obfuscate rather than to ever provide any evidence of any competing theory.

And Paul, try to tell a dealer in Las Vegas that your loaded dice was random...
remember...we now use 5 decks with continuous reshuffles in blackjack to make card counting (which takes advantage of the non randomness in a deck of cards) much harder.

So--if anyone wants to use this word "random"-- it is upon them to tell us what definition they are using from what source and why. Otherwise--random means "having equal probability" and/or not dependent upon the past or future. The only definition I see of random in relation to probabilities is what is the likelihood (probability) that something is truly "random"... it's a way of teasing out codes and chaos from seeming randomness-- please, anyone, show me a single definition of randomness that means unequal probability descriptions or one definition that shows how processes containing "random" can be concluded to be random in themselves.

Or don't. Are we playing a semantics game? Or trying to understand how selection is not random...that is, how it builds complexity from what came before...(random and not)--?

 

[mijopaalmc]

I am definitely out of my depth when it comes to discussing the rigorous mathematical definitions of stochastic process and random variable, but there are definitely conclusions that I know than I can make without the year or so of real analysis and probability theory I would need to fully understand the aforementioned rigorous definitions.

The most important conclusion that can be made (and this should be evident to anyone who has taken a elementary statistics course) is that a variable need not be uniformly distributed to be random. In fact, two of the most commonly used random variable in biology, the (standard) normal random variable and the chi-squared random variable, are non-uniformly distributed but are nonetheless by definition (and not just nomenclature in which the "random" is often left off) random variables. Thus, the requirement that a variable must be uniformly distributed to be random unnecessarily restrictive and not mathematically rigorous.

However, such a requirement is incredibly important to creationists and those who strive to refute them, because the argument "evolution cannot happen if it is random" rests on the uniform distribution stipulation. It is in fact true that if the probability that every individual, regardless of its genetic composition, passes on their genetic information is equal, evolution cannot happen because, in such a case, there are no selective pressure. This though assumes that "random" means "uniformly distributed", which it is clear form the most rigorous mathematical and statistical definitions that it does not. In fact, as I have mentioned before it is the very fact that the probabilities of passing genes to the next generation is non-uniformly distributed that makes the probabilistic interpretation of evolution possible. Thus, as I have also already mentioned, the creationist argument that evolution cannot happen because it is random is as fundamentally flawed as the evolutionary biologists' attempts to refute it because the definition of random is not rigorously true.

 

[jimbob]

As an observer of this discussion I suspect most of the problem is with the semantics of the word "random".

In statistics, a random event is an event not affected by either previous or future events--all events are equally likely.

A random number generator generates numbers so there is no detectable pattern...but even that is only "pseudo random" in the strictest sense of the word random.

May I attempt to answer this from what I assume is also Meadmaker's POV.

In statistics, a random event is an event not affected by either previous or future events

Yes

all events are equally likely

That does not follow for any distributions other than a uniform distribution.

In the case of radioactive decay, it is equally likely that a nucleus will decay or not decay within any time period defined by its half-life. Within two half-lives, it is three times as likely to have decayed as not.

As far as I know, the only truely random processes are those involving quantum effcts, which might get amplified to real-life effects, of which the most famous "mechanism" is Schrödinger's cat.

Every nucleus of a particular isotope has an equal probability of decay within any certain time period, *but* the half-life is different for different isotopes. A billion atoms of Ploutonium 238 will decay far quicker than a billion atoms of Uranium 235.

In statistics, probability distributions describe how the probability of particular random occurances is distributed according to a particular variable. The integral of the probability density function over its full range (which might be from minus infinity to plus infinity) is one.

Some of the processes might not actually be random, but simply a complex mixture of subtle pseudo-random and unknown effects this often leads to a gaussian type of distribution; for example, spreads of machined parts (which are due to deterministic effects) still often produce a normal distribution.

Jim

 

[jimbob]

I supect that Meadmaker's understanding of evolution is similar to mine, and I also suspect this is not actually different from articulet's either.

I have not seen anything in Meadmaker's posts to suspect anything other than an evolutionary perspective.

I doubt that it is relevant, but my academic background is somewhat similar to Meadmaker.

(Trained as a physicist and work as an engineer/physicist)

However the thing about natural selection is that the basic idea doesn't realy even need an understanding of mendelian genetics.

It is a very elegant, and to me, self-evidently true ninteenth century theory, that is inherent to any self-replicating system with less than 100% fidelity)

I use the phrase "natural selection" rather than evolution, but see this as including e.g. sexual selection as well as "survival of the fittest". It is just sucessfull production of viable offspring of the the most sucessful breeders...

Again to clarify.

Meadmaker and articulett,

I assume that both of you would agree that the evolution of any particular species was not inevitable. I know that meadmaker would.

I would argue that a highly* technological intelligence (such as chimps as opposed to crows) was probably not inevitable.

Humanity was certainly not inevitable: the relative lack of genetic diversity in modern humans compared to other animals suggests a population bottleneck (i.e. at one stage Homo Sapiens was an endangered species)

*for "highly technological", I am using the idea of taking one tool and modifying it to make a better tool, for example turning a stick into a bushbaby-hunting spear.

Meadmaker, would you agree that in certain environments, certain traits are very likely to evolve?

For example:

Eyes in any environment with light.
Flight
Meat-eating in an environment without many large predators (c.f. Australia).
Flightlesness in a safe environment (c.f. New Zealand, and other isolated islands, this would be selected for fairly quickly because of the associated energy/size costs of maintaining the ability to fly)
Antibiotic resistance in bacteria exposed to antibiotics
​

Would it be fair to sum up your view as evolution being a percentage game?

Present traits that slightly increase an individual's ability to breed viable offspring will become more prevalent, and if extreme examples provide the best improvement in the breeding sucess in that environment, then these will evolve, and possibly become a separate species.

These traits can change dramatically over geological time, c.f the evolution of the mammalian inner ear

Jim

 

[articulett]

Jim, I think most of the people here agree on what evolution is. Some people want to define evolution as random when it is a process with some random variables--biologists prefer "random mutation (more or less) coupled with natural selection (which is the opposite of random more or less). For some reason people think that it's okay to call a process random, just because random variables are involved. But the word process, itself, implies non-randomness...it has to do with a series...each step dependent on the last...just like evolution.

I give up with mijo. He asked a question (the title of this thread) to supposedly address a creationists argument that mischaracterized evolution as "complexity by chance. " This is a common creationist claim which really shows a cluelessness about selection via time... the builder of complexity. But mijo keeps wanting to define evolution as a random process for some reason or to have peer reviewed papers tell him it's "non-random". Multiple people have told him that calling evolution "random" is misleading if not incorrect. A random component does not a random process make...

I teach biology and have dealt with many a creationist--the key isn't to get them to understand randomness, the key is to get them to understand selection--how it's not-random....it culls the fittest through time. All disagreements here center around why one would call evolution random. It's obfuscating at best, and only true if you use a lay man's term of the word random (purposeless or unpredictable). If artificial selection is not random (and I don't think anyone would argue that it is), then natural selection is not either for the same reasons--it just takes longer to hone the process... Humans speed up what the environment does all by itself...by ensuring that some genomes preferentially (not randomly) survive. I can't say it better than Dawkins nor Sagan. And the way mijo is saying it is clearly confusing, doesn't convey meaning, doesn't show an understanding of selection, and shows an unwillingness to depart from this notion that evolution is "random". He's the one hung up on using that word. And he's doing his best to find a way for that statement to be "true" and convey that he understand selection and how to answer his a creationists argument. I am quite certain he IS a creationist (or rather, proponent of intelligent design) because he sounds exactly like them. And I doubt he could clear up any creationist confusion as long as he has this need to call evolution "random"-- He just isn't saying anything that conveys any information of value.

I am just a bit nauseated from the way "intellilgent design" proponents latch on to some science words (like theory) and use it obfuscate in some semantic game. The stuff they say is so convoluted, and evolution really isn't that hard of a concept to grasp... unless you have a need to call it "random" and then suggest that an "intelligent designer" put this "randomness" in order for a divine purpose (or however Deepak Chopra might refer to it...). The truth just needs to be understood...not bandied about, debated, argued, spun, or conveyed through semantic flummery.

 

[articulett]

I am definitely out of my depth when it comes to discussing the rigorous mathematical definitions of stochastic process and random variable, but there are definitely conclusions that I know than I can make without the year or so of real analysis and probability theory I would need to fully understand the aforementioned rigorous definitions.

The most important conclusion that can be made (and this should be evident to anyone who has taken a elementary statistics course) is that a variable need not be uniformly distributed to be random. In fact, two of the most commonly used random variable in biology, the (standard) normal random variable and the chi-squared random variable, are non-uniformly distributed but are nonetheless by definition (and not just nomenclature in which the "random" is often left off) random variables. Thus, the requirement that a variable must be uniformly distributed to be random unnecessarily restrictive and not mathematically rigorous.

However, such a requirement is incredibly important to creationists and those who strive to refute them, because the argument "evolution cannot happen if it is random" rests on the uniform distribution stipulation. It is in fact true that if the probability that every individual, regardless of its genetic composition, passes on their genetic information is equal, evolution cannot happen because, in such a case, there are no selective pressure. This though assumes that "random" means "uniformly distributed", which it is clear form the most rigorous mathematical and statistical definitions that it does not. In fact, as I have mentioned before it is the very fact that the probabilities of passing genes to the next generation is non-uniformly distributed that makes the probabilistic interpretation of evolution possible. Thus, as I have also already mentioned, the creationist argument that evolution cannot happen because it is random is as fundamentally flawed as the evolutionary biologists' attempts to refute it because the definition of random is not rigorously true.

random variables do not a random process make... stochastic, maybe...

You do not except to see uniform probabilities in randomness--just because all choices are equally likely (such as any combination of heads and tails in a series of coin throws) does not mean that one can expect uniformity...in fact, it would be odd if coin tosses were "uniformly distributed" hthththt....

 

[Paul C. Anagnostopoulos]

random variables do not a random process make... stochastic, maybe...

I believe random process and stochastic process are synonyms. From the Penguin Dictionary of Math:

stochastic process. A random process. Common usage excludes deterministic processes, which are subject only to random errors.

There is no entry for random process.

~~ Paul

 

[Meadmaker]

Meadmaker, would you agree that in certain environments, certain traits are very likely to evolve?

Yes.

Would it be fair to sum up your view as evolution being a percentage game?

Yes.