More Recent Comments

Monday, August 13, 2012

An Example of "Directed" Mutation and an Idiotic "Gotcha"

There's nothing in modern evolutionary theory that allows for mutations that arise specifically because they will produce a future benefit. That's why we say that mutations are "random" with respect to outcome.

There's nothing in the known history of life that suggests it has a purpose or direction. In particular, there's nothing to suggest that 3.5 billion years of evolution were just advanced preparation for the appearance of Home sapiens. That's why we say that evolution appears unguided and purposeless [Is "Unguided" Part of Modern Evolutionary Theory?]. And that's why anyone who says that life shows evidence of purpose is not being scientific.

Creationists aren't happy about this so they will go to extraordinary lengths to wiggle out of the inescapable conclusion based on solid evidence. The usual excuse is to postulate that god is very sneaky. He/she/it makes a huge effort to hide his/her/its manipulations so that it only appears that evolution is unguided and purposeless. The clever creationists aren't fooled by this sneaky god; they can detect its deception, but scientists can't.

But this isn't the only form of counter-attack. Creationists also like to argue that mutations are not truly random. They point out that there are mutational hotspots in the genome and there's a bias in favor of some mutations over other (e.g. transitions are more common than transversions). In most genomes, mutations are more common at sites where C is methylated.

All this is true and the results were discovered by scientists, not creationists. It's why scientists try to avoid saying that mutations are random; instead they say that mutations are random with respect to their ultimate usefulness. Sometimes we slip up for simplicity as when I said in my previous posting that mutations are "essentially random," although I added "Let's not get into quibbling about the meaning of "random."

Jonathan Bartlett (johnny b) is a computer scientist who has written a lot about mutations on various creationist websites. His latest is from Uncommom Descent: Responding to Moran – Is “Unguided” Part of Modern Evolutionary Theory?.
I am always aghast that in the 21st century people still make the claim that mutations are unguided. This is a hold-over idea from before the discovery of DNA, simply because some mutations were found to occur independently of selection.

However, modern evidence has showed that mutations are actually in large part due to mechanisms geared for adaptive purposes, just like the rest of biology.
I haven't seen any evidence that average mutations are "geared for adaptive purposes." It certainly doesn't look that way when you compare the sequences of homologous genes in different species.

It still looks to me that the mutations underlying evolution are unguided.

Here's an example of somatic hypermutation where mutations occur much more frequently in a specific regions of the genome than they do in other regions of the genome. It occurs in the precursors of B cells.
To point to a simple example (and one that is even often used as definitive evidence of the efficacy of random mutations!) let’s look at the somatic hypermutation process in the immune system. When a new bacteria invades the body and causes an infection, the body must generate a new gene. So what does it do? It takes a close-fitting antibody gene and mutates it. Now, first of all, you should notice that the mutations only happen in the correct gene – the antibody gene. That’s 1,200 base pairs out of 3,000,000,000. But that’s not all – it also focuses mutations on the part of the gene that attaches to the antigen, not the part that signals the cell (because otherwise it wouldn’t signal the cell correctly). So, that’s roughly 600 base pairs out of 3,000,000,000. The mutation system is highly selective of the sites that it mutates, skipping over the cell signaling systems and focusing on the part that is specific to the antigen.
This is essentially correct. The target region is rich in methylated cytidine and the developing B cells produce an enzyme called activation-induced (cytidine) deaminase that deaminates cytodine to produce uridine. Uridine is recognized as DNA damage so it is repaired but the repair enzymes are just as likely to substitute an A:T base pair as the original G:C base pair. This cause a mutation.

These mutations make no contribution to evolution since these are somatic cells, not germ cells. It is, however, an example of increased numbers of mutations targeted to a specific site so we should be careful not to say that mutations are randomly distributed in the genome.



79 comments :

Enrique said...

Life has shown that cells take advantage of whatever bizarre mechanisms, even hypermutation, if it eventually works. But "taking advantage" of an existing process is different to design.

Adaptive mutation rate increases under stress is a proposed mechanism, but as far as I know not a proved one.

Stress (e.g. starvation, oxidative or osmotic stress) may increase mutation number, but several mechanisms explain that fact: stress itself may be mutagenic per se (oxidative stress, most probably high salt stress), or reduce the efectiveness of reparation mechanisms (i.e. incrementing final mutations).

I'm not aware of any active cellular response to increase mutation rate per se, in opposition to rescue mechanims, like error-prone translesion synthesis. But that's only reflects my own lack of coverage of the topic.

It is true that in many "casual" or divulgatory descriptions, on TV and newspapers, I have read sentences along the lines: "upon challenged cells mutate to adap". But that's obviously a terrible missappreciation of real evolution mechanisms. No such thing in real research.

gillt said...

These mutations make no contribution to evolution since these are somatic cells, not germ cells.

These mutational hotspots provide genetic variation within populations and you're saying they make no contribution do evolution?

How do you know?

Mike D said...

This guy's in Tulsa?!

Sigh. Why am I not surprised?

I googled "Ignite Tulsa", and found it's an 'open mic for geeks', or like a TED Talk for anyone. Not the worst concept in the world, but it also means that any idiot can get up there and make a fool of himself.

And seriously, what's with all these computer scientists lecturing people on evolution? I looked up the CVs of the DiscoTute 'Fellows', and out of 50 or so, y'know how many had any education in biology at all? NINE. Less than 20%! Maybe leave biology to, I dunno, biologists.

Robert Byers said...

YEC creationists do not see mutations as having had any relevance to origins.
no evidence they can or ever did.
Its just speculation after everything is done.
Its not been witnessed.

Where is the evidence mutations created anything new?
The only evidence would be a line of reasoning from data points after any alleged process occured!

Its unlikely and unreasonable to imagine mutations of life turned bubbles into buffalos.
Evolutionists must go a long way to demonstate such a thing could or did happen to create the glory of biology.
If more people understood evolution is about random mutations turning this into that and then into this THEN they support for evolution amongst the public, who are barely aware of these things anyways, would drop.

Creationists should hold evolutionary biology to a careful scrutiny that it is using the scientific method and demonstrate why it is not and never did!

One can't do science on these past and gone.
Evidence for mutationism in the past could only be slippery to get a hold on.

If ID creationists smell out some direction for mutations then thay are accepting the importance of mutations.
In fact everyone needs mechanisms to explain diversity.
However we need better ideas and these backed up with solid evidence by any fair analysis.

jaxkayaker said...

He just told you that the mutations are in somatic cells, not germ cells. Therefore, the mutations won't be passed along through inheritance. If the ability to generate the directed mutations is heritable, it could experience evolutionary change.

Unknown said...

The mutations produced by the somatic hypermutation mechanism are not directed. It's more like: let's shoot in every direction and hope to hit something!

The goal is to have a "good enough" hit. After each subsequent exposure to the pathogen, the affinity of the antibody is improved in an incremental fashion by a mechanism named affinity maturation.

It's actually a pretty good example of evolution in action: you start with something good enough, and then improve the efficiency only if there is a selective pressure to do so (in this case, subsequent infection by the same pathogen).

CharlesJ.

Atheistoclast said...

Good point. Mutational and recombinational hotspots is a major are of research. But don't expect Larry to keep abreast of modern science.

Benjimin said...

Put two red marbles for each blue marble in a bag; can you not reach in and randomly affect one? (And physicists still say the Copenhagen interpretation of QM involves a random process, knowing well that there is structure in the probability amplitudes.)

scientists try to avoid saying that mutations are random; instead they say that mutations are random with respect to their ultimate usefulness. Sometimes we slip up for simplicity as when I said in my previous posting that mutations are "essentially...

If we couldn't call a biased process random, how would we even describe it correctly? We couldn't really say that the non-uniformly-distributed mutations are uncorrelated to their ultimate usefulness, since we know (a priori) that those extra mutations will tend to provide extra useful diversity to the immune capabilities (as if by design).

I think it is simply a naive mistake (for creationists or others) to claim incompatibility between randomness and bias; and would defer accepting such criticism until they've done an introductory course equivalent on probability theory and stochastic processes.

I'm more concerned by how to respond to theistic-evolutionists who would view every outcome in a casino as divinely guided (since that view at least in principle seems compatible with biology).

Larry Moran said...

@Joe Bozorgmehr (Atheistoclast)

I enjoy a good fight as much as anyone but your gratuitous insults are getting quite boring.

This is your first and only warning. Keep it up and you will disappear from this blog.

Diogenes said...

The idiot is undone by his own example.

By random errors in 600 bp, a new protein binding site is formed specific for a novel antigen.

Didn't Behe say that couldn't happen without intelligent design?

Benjimin said...

I left out measure theory. If "truly random" necessitated "drawn from a uniform distribution", then would only raise the question "with respect to which metric?"

Atheistoclast said...

Fine. Then ban me, you old fart. But it won't make your arguments any more compelling or cogent as a consequence. You'll still remain the same bigoted and benighted reactionary who opposes the progress of modern research.

Allan Miller said...

I'm more concerned by how to respond to theistic-evolutionists who would view every outcome in a casino as divinely guided (since that view at least in principle seems compatible with biology).

But not compatible with that other sacred cow, free will. A mutation must not just arise, it must spread, and it would be a task for a Designer to effect that spread while allowing the bearers the luxury of 'random' assortment and general personal decision-making capacity - every choice has reverberating, amplifying consequences due to chaos.

Allan Miller said...

I guess a stress-responsive mutator could evolve, in an environment that continually weeded out descendants of less-responsive genomes. But it is hard to envisage such conditions ever arising. Cyclic stresses select for adaptations in all parts of the cycle, but this seems to call for a non-cyclic stress, yet one which is, to the genes detecting it, strangely 'familiar'. A consistent response - mutator activity - is triggered by a variety of novel stresses. If the stresses weren't novel, direct adaptation to the stress itself is much more likely than a 'panic attack'.

It's hard to see a distributed organism acting in any way like a giant 'immune system'. Scattergun-mutating germline genes is a very different response from targetted somatic mutation to one sequence whose sole 'purpose' is to be the target of this mutator. Both mutator and target sequence benefit by surviving through the germ line, undamaged by the activity. Among several other problems they face, germline mutators have to avoid damaging themselves.

Anonymous said...

Atheistoclast, are you so moronic you don't even know the difference between germ cells and somatic cells?

Of course not, what you're showing is the true character of ID. Willful ignorance.

Anonymous said...

Agreed with Benjamin. Mutational hotspots do not necessarily indicate nonrandomness, unless you view the hotspots as indistinguishable from the cold spots. You will roll 1 on a six sided die much less often than on a 20-sided die, but each toss is still random = predictable only in terms of probability. If there were a specific recurring sequence of mutations or a positive correlation with fitness, that would be non-random.

-Anaxyrus

Larry Moran said...

@Joe Bozorgmehr (Atheistoclast)wrote,

Fine. Then ban me, you old fart. But it won't make your arguments any more compelling or cogent as a consequence. You'll still remain the same bigoted and benighted reactionary who opposes the progress of modern research.

Goodbye.

Mike in Barcelona said...

"You'll still remain the same bigoted and benighted reactionary who opposes the progress of modern research..."

...says the guy ranting from his mother's basement who has never picked up a pipette in his life...

Michael M said...

I agree and sympathize with you, Benjamin.

If evolution acceptors (or however they want to refer to their intellectual relationship to evolution to avoid the analogy to creationists' belief) want to discuss evolution's unguidedness, they need to do so with the fact that they also describe evolution as "non-random" (in whatever their pet sense is at the time) in mind. If evolution acceptors want to semi-quantitatively discuss the way in which evolution is purportedly "non-random", they keep the fact that the mathematical discipline of probability theory has already devised a quantitative language of randomness in mind and that their distinctions between "random" and "non-random" may be meaningless within probability theory.

Allan Miller said...

ergo ... every immune response is intelligently designed! Gotcha!

Diogenes said...

I will miss Atheistoclast. He was useful as an example of ID dishonesty, how the IDers get caught lying and just change their story, or the subject. Personally I'd prefer it if he were not banned.

He trolls, yes, but there are worse trolls.

Diogenes said...

This is just another example of ID creationists exploiting the mathematical ignorance of their Muggle audience.

Everyone who knows statistics knows that "random" means a variable of which we only have a statistical distribution. There are, obvioulsy, all kinds of statistical distributions-- normal (gaussian), poisson, hypergeometric etc. No one of them is more or less "random" than another. All of those distributions are "random" but they produce completely different behavior-- and they can have different means and standard devs., etc.

So when we say mutation is "random" obviously Muggles may get the idea that scientists imagine mutational rates are a flat distribution, the probability of all nucleotides being equal. Obviously, that'd be ignorant to think that and everybody in mol. bio knows better.

So the IDologues come out and pull a SAFI-- "Scientists are Fucking Idiots!" SAFI arguments all consist of misrepresenting what scientists think so as to make them look dumb (non-coding DNA = junk DNA is another SAFI.) In this case the SAFI is:

1. Scientists think mutations are random!
2. In fact, the probability of getting a mutation varies at different points of the genome!
3. Therefore SAFI: Scientists are Fucking Idiots!

By making scientists look dumber, they feel like they're making themselves smarter.

Michael M said...

Innumeracy is similarly exploited by scientists who claim that evolution is not random.

Anonymous said...

@ Micheal M

Please update yourself before making a silly statement like that. There are both random (ex. genetic drift, contingency) and non-random (es. selection) components to evolution.

http://evolution.berkeley.edu/evolibrary/article/evo_14

Michael M said...

*yawn*

You must have missed the part where it was said that "biased" and "statistically independent" aren't synonyms for "non-random". It is dishonest to blame creationists' understanding "random" as "unbiased", when it is repeated ad nauseam by scientists.

Diogenes said...

@Michael M, our creationist visitor,

There is a process you may have heard of--it's called "natural selection."

When you see a store with a sign that says, "We have a great selection", do you interpret that to mean, "We have a great randomness"?

I'm wondering if you have ever heard of any word in the English language this is less random than the word "selection."

If you have, please tell us that word, and we will change the name of "natural selection."

Howzabout if we called it "totally 100% non-random non-chancey non-blind differential fecundity correlated increase in allele frequency"? Would that shut up your dishonest creationist authorities?

Michael M said...

I'm not a creationist, but you seem not to get that you are simply wrong when you insist that evolution is "not random" because it is "biased". There is no such thing as "more random" or "less random", just "random" and "not random"; and evolution is most definitely an random process. Evolution's randomness, however, does not prevent it from being the best empirically supported explanation for life's diversity on Earth.

Allan Miller said...

I have seen debates over this run on for MONTHS...

When creationists say 'mutations may not be really random', they don't mean 'they may have an uneven distribution', they mean 'someone may be causing them'.

As to whether selection is random ... I batted against this notion for yonks, but I have seen the light. It is. Biased random is still random.

You get almost-certain outcomes from random processes - as near deterministic as makes no difference. Even equiprobable ones. The most striking is the fact that, even in the total absence of selection, you still get fixation of descent from a single ancestral sequence with a certainty you could bet your house on. You don't know which one, but you can be sure one will. Selection just makes it happen quicker, and with greater likelihood of ultimate success for the beneficial allele.

But it makes no sense to say that (for example) a selection coefficient of 0.00001 gives a random result (pure drift, if population size is below a certain value) while 0.0001 (for the same population size) does not. Especially since population size and s vary randomly too. I'm with Michael M on this one.

Given an appreciable supply of positive mutations, even with small selection coefficients, large numbers of trials would be expected to lead to genomes with a bias towards adaptively-fixed traits, certainly among the phenotypically visible ones - the Law of Large Numbers at work. But for any one mutation-fixation series, 'actual' selection, as a casual effect of the allele on individual lives, may have made a trivial or nonexistent contribution. The signal is swamped by the noise, but becomes clearer the more trials you perform.

Shoku said...

Of course he doesn't understand the difference between cell types, guy has no experience with cell biology.

If he's still reading*: Those spots don't especially mutate in sperm and eggs- you've got white blood cells turning on that specific protein to come in and muck up the sequence basically at random, but you keep lots of unaltered copies around so that the white blood cells in your kids can do this same thing. If not then you'd run out of methylated bases really fast, lest you had some mechanism to randomly turn things in those 600 pairs back into CG before tacking on a methyl group.

*Who am I kidding? It's for anyone else that's not familiar with this.

Shoku said...

The easy study to refer you to is Richard Lenski's work with C. coli and citrate metabolism. The lineage of bacteria is all extremely well preserved (you can basically make twins and keep one frozen while the other keeps going through generations.)

It's like having really high resolution images of the process in action. You know right where a series of mutations brings them close to a protein that will let them digest citrate instead of having to start from the point where they actually have the gene that actually lets them do it.

There's tons more but for sanity's sake just try to wrap your head around this one first.

Allan Miller said...

bubbles into buffalos.

They love this alliterative dismissal! That's a new one. I've seen mud-to-Mozart a lot. Hmmm... adenine to aardvark, bubbles to buffaloes, crud-to-cavies, deoxyribonucleic-acid-to-dingoes ... maybe that last one needs a bit of work.

Shoku said...

Just to make this clear to a non-chemistry crowd mutations aren't random at all if you're just talking about how particles interact. If you pictured every atom and photon and such involved with all of their positions and energy levels/motions (Laplace's demon anyone?) then you'd get the same base pairs in the same order every time (generally excluding some quantum effects for simplicity's sake.)

The thing is that these are factors external to the actual sequence of DNA that a cell makes use of. As in the immune system example there are certain things you can do to the DNA that make it easier for these other factors to alter the sequence of the DNA, but even so you're relying on those other factors to actually make the mutation happen- or not happen.

The simple fact of the matter is that there is no stretch of DNA that is completely unyielding to those external factors, nor any sequence of DNA that yields completely to the slightest chemical variation. Anyone familiar with chemistry knows that you have to use statistics (because you aren't Laplace's demon so you generally only know how many specific molecules you're working with instead of all those details about their exact motions- or at least not which direction each particle is going.)

So this is really a matter of scale. If we just look at the DNA and the large proteins interacting with it, only really knowing the concentration of bases and overall temperature of the messy mixture, then we don't have enough detail to say precisely what atoms will be where at the critical instant.

The part that is usually hardest for laymen to wrap their head around is that this is the only scale worth using to talk about mutations and evolution. You don't need to know those specific smaller details, and you don't even want to. If you did manage to get them once and you started working on that scale then you would need them in every case to be able to say anything about what would happen.

So instead we work at this scale of details that we actually have access to, and we talk about probabilities where we pretend that those smaller details we've left blank can be anything that's possible (in relation to the details we DO have.)

And this works. We get the right answer to questions over and over. We're very much right to treat those smaller details as if they were random, because statistically they are random, and at this scale there's no other useful meaning for random.

The kicker though, is that if you really understand this, then you know that it's also the way everyone is trying to use the word random in the first place. Only someone out to deceive people could insist that random meant something else here.

Hopefully this sheds a lot of light on what it means for mutations to be "random in respect to their ultimate usefulness."

Denny said...

Larry said, “There's nothing in modern evolutionary theory that allows for mutations that arise specifically because they will produce a future benefit. That's why we say that mutations are "random" with respect to outcome.” – If you are correct with respect to how and why evolution works and why it works that way, why is it that virtually (if not literally) every video documentary that is explicit in using evolution as an explanation for animal origins and behavior portrays them as fighting to preserve their species? For what purpose would animals seek to preserve themselves?

Allan Miller said...

If any documentary potrayed them thus - which I doubt - it would be wrong. Organisms fight to preserve themselves, because genes that contribute to this preservation have emerged from the distribution of 'random' positive, neutral and negative mutations, due to their direct influence on their own survival.

Genes that cause their bearers to lie down and say "oh, what's the use?" are less likely to survive down the generations than those that cause their bearers to keep trying.

Anonymous said...

I've gone around on this too, and I largely agree with Diogenes, but see the other viewpoint as well. The problem is that in nature, selection always comes along with drift. I would agree that the differential fecundity we observe in nature exhibits both bias and randomness. But the randomness is drift. Selection is only the bias, the weighting of the dice, with drift being the tumble.

-Anaxyrus

Denny said...

Allan Miller said, “If any documentary portrayed them thus - which I doubt - it would be wrong. Organisms fight to preserve themselves, because genes that contribute to this preservation have emerged from the distribution of 'random' positive, neutral and negative mutations, due to their direct influence on their own survival.” - Well, Allan, I love to watch nature documentaries. In everyone I can remember, the narrator’s script says that the subject animal is (for example) simply trying to preserve their respective genes for the preservation of their species (like salmon spawning). Preserve seems to indicate something more than blind instinct. I give two other brief examples below.

“Adult males killing young of their species isn't unknown among animals, especially bears. But in most cases it happens in the spring so the male can then impregnate the sow he has left without cubs.” http://www.thestar.com/news/sciencetech/environment/article/731873--polar-bears-eating-young-due-to-shrinking-sea-ice-scientists

“That (killing ‘young’) will reset all the females into estrous, and he can maximize his reproductive success by mating with those females. That sort of scenario also happens with Colobine monkeys,’’ says Fitzpatrick. Infanticide in the wild is common and occurs for a variety of reasons, says Mark Fitzpatrick assistant professor in the biology department at the University of Toronto, and an expert in animal behaviour, mating and aggression. http://www.thestar.com/news/gta/article/1100493--when-moms-kill-their-young

The narrative of these two accounts seems to imply forethought. What evolutionary logic explains this? What I’m really pointing out here is that I think cold impersonal non-purposeful evolution is not what’s portrayed in most documentaries aimed at the public. I think those documentaries portray a kind of evolutionary romanticism where the evolutionary process itself is making sure that everything not only works out OK, but keeps getting better. That’s not what Larry said in his opening sentence – “There's nothing in modern evolutionary theory that allows for mutations that arise specifically because they will produce a future benefit.” Both of the above examples seem to imply a future (evolutionary) benefit. If, as Larry implies, evolution is bind, why animal cannibalism?

Michael M said...

Once again, someone confuses "random" with "unbiased". If you insist the "random" must be interpreted as "unbiased", you are stuck with a number of special cases, but are largely unable to describe evolution without contardiction.

Anonymous said...

Random need not be interpreted as unbiased in all instances, but with regard to sampling, that is exactly how it is interpreted. Is the fossil record a random sampling of past life on Earth? Of course not. Are future parents a random sampling of zygotes? No.

Is the mechanical behavior of a bouncing ball tossed against a wall random? Because, like evolution, it takes both random and nonrandom input.

Michael M said...

The fact that the large-scale behavior of ensembles with many components is predictable does not mean that the behavior of its components is "not random". Moreover, there is already vocabulary that allows people to discuss phenomena that are "random with respect to" without restorting to calling such phenomena "random": these phenomena are "statistically dependent from" or "uncorrelated with" whatever they are "random with respect to". These terms are actually more well-defined than "random with respect to", because "random with respect to" covers both "statistically independent from" and "uncorrelated" (as well as whatever other descriptions the speaker wants to throw in there), and they are not equivalent descriptions of phenomena.

Allan Miller said...

In everyone I can remember, the narrator’s script says that the subject animal is (for example) simply trying to preserve their respective genes for the preservation of their species (like salmon spawning).

That may be what your hearing, but I'm not sure that's what is being said. 'The Good of the Species' is an outmoded concept, though it sometimes resurfaces.

When people use terms like 'trying', 'preserve', or Dawkins's gene-level equivalents, they are speaking metaphorically. As, surely, you must have got by now?

Trees, with no nervous systems, are 'trying' to outgrow each other to reach light, bacteria are 'trying' to grow faster than their neighbours, pollen is 'trying' to get stuck on bees ... ? It all boils down to a belief that the user of such terms means something teleological by their use. That belief is mistaken. Gene survival is dependent upon being copied, and if different genes have different rates of being copied, those with the higher rates will tend to outdo those with lower, even if the differentials arise completely 'at random'.

So if there is a causal link between - say - infanticide in bears and the propagation of infant-killing genes, the surmise would be that infant-killing genes got into more offspring than infant-leaving-alone-genes. It requires no knowledge or foresight on the part of the bears. Selection is not always pretty. The alternative might be an ID who generated the behaviour by directed mutation, which would be even less pretty IMO.

Michael M said...

Is the sum of two die rolls "not random"?

One is, after all, more likely to roll a "7" than a "2".

Anonymous said...

This is not the question. Two tumbling "fair enough" dice are about as random as anything we know. All microstates are equiprobable. But if my dice are heavily loaded towards 6, then against an opponent with fair dice, I will have predictable success at Risk or Backgammon. We don't need to play the game. Random inputs won't matter.



-Anaxyrus

Denny said...

Allan, I presume ID = Intelligent Designer. But what's IMO?

Denny said...

Allan Miller said, “they are speaking metaphorically. As, surely, you must have got by now?” – Well, No. I live mostly in the world of the science layman. I watch and read what’s written and produced for curious layman in the public sector – desiring mostly to be entertained. For fun I do delve into actual science reading and sometimes watching. I hang out with people with scientific qualifications and vocations. And because evolution purports to be an argument against the existence of God, I visit Sandwalk. Therefore, much of the TV documentaries I see speak of evolution as a wondrous ever-improving process – continually making biological things better and better. They package biological evolution in virtual reverent terms. Even the cosmologists get involved, as they attempt to find a life-hospitable planet, which would, ‘demonstrate’ that life on earth is not unique, but simply another part of an accidental random process called (cosmological) evolution. Unless one catches Lawrence Kraaus or Neil Tyson or even Richard Dawkins in their more candid moments, the layman never hears that the universe is on a ‘journey to oblivion.’ I suspect that that might be a turnoff for most people, although not for you and Larry and other Sandwalk fans. The public seldom gets an informed chance to match the philosophical proposition of atheistic/evolution against the one offered by theistic/creationism (old-earth). I also suspect that if the oblivion aspect to cosmic (and consequently biological) reality were ever published widely, evolution might finally be seen as little more than an atheistic and apparently well-reasoned argument against theism. Hugh Ross has said (paraphrasing), ‘If science ever contradicted God’s existence, he’d give up God.’ If science ever contradicted atheism, would you give up evolution?

steve oberski said...

Denny, IMO means IDiots Make Me Ornery.

Michael M said...

You don't have to weight the die for the distribution of face values to be non-uniform, and that 's really the whole point. A die that has three face marked "6" has three time the probability of coming up "6", even if each face has equal probability of coming up.

Anonymous said...

@ Micheal M
Please explains how p~2sNe/N is random.

Anonymous said...

@Michael M,

A fair die is still six equiprobable microstates. If you paint 6 on both the 4 and 5 side, then you are binning microstates together. In a sample of tosses, your deviation from 1:1:1:3 will be random. But with a very large numbers of tosses, your outcome will be significantly outside of expectation from random sampling with replacement of [1,2,3,6]. It will be equivalent to random sampling with replacement of [1,2,3,6a,6b,6c].

-Anaxyrus

Anonymous said...

Apologies for the double post above.
-Anaxyrus

Shoku said...

And do you think anybody that wants to argue that an entity has been guiding mutations behind the scenes for the full duration of life on Earth to specifically produce humans is going to grasp the meaning of any of those phrases? In my experience it's hard enough to just get regular laymen to stop using random interchangeably with unpredictable.

Denny said...

Thank you, Steve. You're such a sweet guy.

Michael M said...

Uhhhh...doesn't pstand for the probability of fixation for an allele of selection coefficient s in a population with an effective size Ne and census size N?

Notice how that equation implies that neutral alleles (s=0) never get fixed?

Shoku said...

As I recall there was once a notion among scientists that animals had the good of their species in mind. This was used to explain why male deer and such would compete with less than their life on the line (ex: they don't usually take a cheap shot if the other guy slips and falls down.)

Game theory and the like showed that there are much clearer explanations for these behaviors.

-

@ Denny: I've watched a majority of those documentaries, and while they do have a couple of lines of dialogue like that (I get frustrated watching the nuances get mangled like that so they really stand out to me,) you're largely getting that impression from somewhere else.

You could argue that the general public already has these notions in their head and without more strongly working against them the media fits well enough to convince you that you do already know what they're talking about.

If science contradicted evolution I'd give up evolution and if it contradicted atheism I'd give up atheism (but probably be right pissed off at some god.) Logically it can't contradict atheism though because even with a booming voice in the sky and miracles right in front of you it always makes more sense that you're being fooled by some lesser entity. A truly omnipotent being could come up with something unimaginable to get around that problem but within the universe as we know it there doesn't seem to be any option like that.

Anonymous said...

A fair die is still six equiprobable microstates. If you paint 6 on both the 4 and 5 side, then you are binning microstates together. In a sample of tosses, your deviation from 1:1:1:3 will be random. But with a very large numbers of tosses, your outcome will be significantly outside of expectation from random sampling with replacement of [1,2,3,6]. It will be equivalent to random sampling with replacement of [1,2,3,6a,6b,6c].

-Anaxyrus (the p~ above is not mine).

Anonymous said...

Thanks for defining the variables. Now how is such a relationship random?

p.s. Notice how there is a ~, not an =. Sometimes approximations are easier to type out.


Michael M said...

If you make any of the faces indistinguishable from any of the other faces and only record the value on the face facing upward at the end of the roll, it does not matter if the die is physically fair (i.e., each face is equally likely to come up); the die is mathematically unfair (i.e., each value is not likely to come up), precisely because each "6" is indistingushiable from the other "6"'s.

Allan Miller said...

In my experience it's hard enough to just get regular laymen to stop using random interchangeably with unpredictable.

Why would you want them to? It is, essentially, what mathematicians do, though they use the fancier term 'probability distribution'. One can predict trends - and over many trials, results generally converge on expected values - but there is nothing fundamentally to mark any probability distribution as more or less 'random' than any other, unless you favour 'equiprobable'. .2/.2/.2/.2/.2 vs .1/.2/.4/.2/.1 - which is more random?

Allan Miller said...

Denny - IMO: In My Opinion.

Forgive my testiness on the metaphorical usage - but I do see a whole host of misunderstanding as to how genes can be selfish, and how organisms can maximise their reproductive output if they don't know what is going on ... I guess what seems plain to me is not always.

If science ever contradicted atheism, would you give up evolution?

Absolutely not. I don't see any essential connection between the two at all. Evidence for evolution is not evidence against God - though it probably is evidence against literal readings of Genesis.

If science found evidence against evolution, or for a better theory, then yeah of course I'd give it up. I am only interested in what is the case, not what I would like to be the case, so sticking to the wrong idea would be pretty dumb. Evolution explains patterns in nature. It does so mechanistically, but I guess the possibility of a God pulling the strings can't be entirely ruled out - as is considered by theistic evolutionists.

Michael M said...

I was just clarifying what the relation meant.

Does it mean what I said it means?

Anonymous said...

Nothing mathematically unfair here. If 3 of 6 individuals are "6", half of singleton samples are expected to be 6. In fact, if far less than half are "6", we would then suspect bias (or perhaps even nonrandom behavior). We expect all microstates to be equiprobable (equally accessible). The "6" outcome consists of three microstates. (In this example, we can definitely distinguish them, but even if we can't, it's irrelevant.) Did you ever take statistical mechanics?

Here is where it can get tricky. Can a phenomenon be random and biased? Yes, but the bias must be stochastic (statistical) not systematic. Natural selection introduces systematic bias.

-Anaxyrus

Michael M said...

I'm sorry but the 6's are numerically indistinguishable from one another. If you want to distinguish them from one another, you have fundamentally changed the mathematical structure of the model, and you are no longer discussing a model that is analogous to the standard dice model.

The whole point of statistical mechanical models is that they are invariant under the exchange of indistiguishable particles. This lowers the of permutations that you have to consider by a factor ~N! (i.e., the factorial of N, the number of particles in the system). However, you cannot exchange distinguishable particles without altering the microstate.

Since all that is being considered in both dice models is the numerical value on the upward facing face of the die (and not any other property), each 6 is indistinguishable from the others, and we have a model in which the die has three indistingushable faces. Anything done to distinguish the faces alters the model in such a way as to make it incommensurate with the previous one. It is as if you have changed the number of particles of one component of a binary mixture and claimed that such a change was identical to creating a ternary or quaternary mixture.

Anonymous said...

Yes! That is the correct equation.

Sorry - re read this series of posts and the tone is more confrontational that it is meant to be. I just want to see what you are meaning by your statements above, and this is a concrete example for me to use to clarify what you mean.

Anonymous said...

What is relevant here is that you have three of them, and each one of them has probability of 1/6 of landing "heads up". Again, you fail to see that binning states is not the same thing as giving them weight.

But, again, no two things as complex as die faces are truly identical. Further, one is opposite 1, one is opposite 2, one is opposite 3 in this example (easily distinguishable).

The only similarity in this case to that of the loaded dice is if we take it to the application of the game. Is it random luck that you beat somebody in backgammon with two of these dice? No.

Will the result of multiple tosses of this die be equal to random variation in a fair die toss? Yes, by definition.

Will the result of multiple tosses of a heavily weighted die be equal to random variation in a fair die toss? No.

-Anaxyrus

Anonymous said...

My stat. mech. inference obviously didn't help because you misinterpret the die faces as being the analog of indistinguishable particles. No, these are the microstates. It is each toss that is the indistinguishable "particle" that goes into one of the six pigeonholes. What you have done is create a class of observations equivalent to any one of three microstates. The probability of a toss landing into this class is 1/2.

-Anaxyrus

Michael M said...

So particles in a n-ary mixture are distinguishable based on their nearest neighbors?

Shoku said...

To be honest I doubt he sees our answers there as a positive thing. Probably thinks we live in some chaotic world with no real anchor :/

I'm not going to go arguing against things that haven't even been said though, so if I'm right he needs to talk about the subject a little bit first.

Shoku said...

I'm saying that the more expanded explanations are more useful to laymen. Unless you're saying that there are zero laymen in the audience of this blog I don't see a reason to continue this exchange.

Allan Miller said...

Well, fair enough if you're done ... I merely point out that it would seem incorrect to divert laymen, however strongly represented, from a usage that is entirely in accord with that within mathematics. Of course, it would take a little grounding in statistics, probabilities and population genetics to properly get it across, but I don't see any real benefit in saying 'ya see, selection is not random'. I wouldn't want to insult the layman's intelligence - among the 'lay' audience will be mathematicians, physicists, engineers, statisticians ...

I would simply argue for a consistent usage across the sciences, particularly those where a mathematical formulation is in use - a vain hope, perhaps. The word 'random' has too much baggage, but we are stuck with it.

Anonymous said...

Sides of a die are distinguishable. This is the example we are working on. The result of a standard die toss is one of six distinguishable states.

-Anaxyrus

Anonymous said...

And again, you are botching the analogy. It is each toss that is analogous to the indistinguishable particle.

-Anaxyrus

Michael M said...

No, the sides of the are not distinguishable, unless you consider the global topology of the die. All we are concerned about is the value on the upward facing face of the die.

Michael M said...

How can I botch your false analogy?

Anonymous said...

"the sides are not distinguishable, unless you consider the global topology of the die"

The cow has five legs, unless you consider that the tail is not really a leg.

You might only care about numbers on faces for some application, but we can see the die has six sides and that not one, but three of them are painted six. Even if we don't distinguish them, we expect 3/6 tosses to end up "6".

Now, if you do not SHOW me the die. If I know NOTHING about the die, except tabulated numbers that you call out, then eventually I could infer that these were not random tosses of a tetrahedral die. I could also infer that these were not random tosses of a standard die with faces numbered 1 through 6. If I do have information about the die (3 of the sides are "6"), then I might find that I do NOT reject a model of random tosses of this type of die.

Take home point: whether data are termed "random" or not depends upon an underlying model. Data that appear randomly Pareto distributed will not appear to be random Gaussian. Larry is right in this post.

I don't know enough about you to realize why you couldn't follow this particular analogy I gave, or why you think it is false. But regardless, it's not helping you, so we can drop it. Can you follow what I have done here?

-Anaxyrus

Michael M said...

Yeah, you're still trying to prop up your false analogy. You just don't seem to want to acknowledge the fact that all matters when answering the question "What is is the probability of rolling a '6'?" is the value on the upward facing face of the die. The location of the the "6" relative to the locations of the other of values on the die's other faces is completely irrelevant to the model I presented many posts ago.

Anonymous said...

Your model from many posts ago was three indistinguishable sixes on an otherwise standard six-sided die. That's logically inchorent, so all we learn from this model is that you didn't think this problem through very well.

After showing you the three sixes ARE distingusihable if we are able to look at the die, I proceeded to show you that they do not need to be distinguishable, so long as we can see that there are 3 of them and only 1 of anything else. That affects our underlying model before the die gets tossed.

Then I switched to a model where I can't see that there are 3 of them. Here you could say that your tosses of this die are random, but seeing only the numbers reported from the die toss, I might reject randomness based on my assumptions going in.

-Anaxyrus

Michael M said...

For the last time, you are adding physical parameters that aren't present in the model. All that matters is the numerical value on the upward facing face on the die.

Anonymous said...

Besides describing the die, you haven't elaborated on the model at all. So what is your model? What are we doing with this die, and what and how much information is being recorded?
"the numerical value on the upward facing face"

Are you talking about one toss and one toss only?
It's pointless to ask if a single result was random or not. Maybe it was, but there is no way to test it.


Another statement of yours from long ago is that you can produce a nonuniform distribution without weighting (by binning three sides together as 6). That is correct. But random does not equal uniform unless uniform is the assumption going in. If we know three sides are six, we change our model before we test random deviation from that model (we're not going to bother to test for random deviation from the uniform model because we don't expect it).

-Anaxyrus

Shoku said...

At the end I turned that around. Maybe I didn't emphasize it well enough but "there's no other useful meaning for random" seemed direct enough as I was writing it.

I suppose we've been talking past each other for this exchange.

Anonymous said...

Directed mutation is NOT intelligent design, nor does it imply any kind of long-term planning for a goal or for conditions that have not yet appeared. It just implies that lifeforms ghange their own genome to adapt instead of waiting for random mutations and dying if the right mutations do not come. A scientific site should not conflate such different concepts just because they go under the same name. The ENCODE project have shown that at least 80% of the human genome is functionally active. Some try explaining it away by claiming that most of the activity is insignificant, but that away-explanation ignores several factors. One is that reading DNA is a job done by energy-expensive reading molecules, making any major amount of meaningless DNA activity extremely unlikely. Another factor is the missing heredity: protein-coding genes can only explain a minor fraction of the heredity shown by twin studies. Some explain away discoveries of functions in non-coding DNA by claiming that they are a minority case and that the vast majority is still junk, theoretically "supporting" it by reference to limitations of how many mutations natural selection can purge per generation. But they are overlooking the fact that hereditary diseases that cannot be explained by protein-coding DNA outnumbers those that can 20 to 1. That is patently not a "minor border revision". Rather, it is a massive expansion of the amount of functionally important genome that makes directed mutations necessary to avoid that literally everybody dies from genetic diseases. A possible mechanism for the directing is that working groups of proteins "feel" when one particular protein in the group does not do its job properly, and sends a signal that triggers unscrambling of the responsible parts of the genome. Experiments that supposedly prove that mutations are random, are in fact victims of false generalizations. Those experiments are really about introducing diseases or poisons that kills too fast for the organisms to have time to direct mutations. John Cairns experiments supporting directed mutations in the 1980s, on the other hand, were about malnutrition (only giving the bacteria food they lacked the genes to digest, in this case giving E.coli only lactose to eat), which gives more time to direct mutations. Not only must there be time to signal the error and change the gene(s), there must also be time to get the production of the modified protein up and going and distribute it to the relevant working groups of proteins. Several molecules are known to interact directly with specific locations in the genome, molecules chemically resembling ones known to affect the risk of cancer when intaken more diffusely. That mechanism should be able to locally alter mutation probabilities. Mechanical changes of the structure of the DNA spiral, a known part of epigenetics, also almost certainly affects the vulnerability of the DNA by making it more or less exposed. And as shown in "Bacteria evolved way to safeguard crucial genetic material", there is evidence that mutation rates do differ between parts of the genome, falsifying the dogma of mutation rates only being able to change over the whole genome and not locally. Epigenetics crosses over from soma to germline, and the is no reason to assume that mutation-directing complaints is anyhow unable to cross with it. Weissman's so-called "barrier" is a false generalization, assuming that the stop must be between soma and germline when it may just as well be between macroanatomy and cell behavior instead. See Pure science Wiki, the pages "Self-organization" and "Inheritance of acquired characteristics".

Martin J Sallberg