A computer programmer noticed that he was not able to type very much in a single day. But he mused that if there were a large number of software bots working on his code, then they might be able to proceed via totally blind trial and error. So he decided to try an experiment.
In the initial version of his experiment, he established the following process.
1. The software was reproduced by an imperfect method of replication, such that it was possible for random copying errors to sometimes occur. This was used to create new generations of the software with variations.
2. The new instances of the software were subjected to a rigorous test suite to determine which copies of the software performed the best. The worst performers were weeded out, and the process was repeated by replicating the best performers.
The initial results were dismal. The programmer noticed that changes to a working module tended to quickly impair function, since the software lost the existing function long before it gained any new function. So, the programmer added another aspect to his system — duplication.
3. Rather than have the code’s only copy of a function be jeopardized by the random changes, he made copies of the content from functional modules and added these duplicated copies to other parts of the code. In order to not immediately impair function due to the inserted new code, the programmer decided to try placing the duplicates within comments in the software. (Perhaps later, the transformed duplicates with changes might be applied to serve new purposes.)
Since the software was not depending on the duplicates for its current functioning, this made the duplicates completely free to mutate due to the random copying errors without causing the program to fail the selection process. Changes to the duplicated code could not harm the functionality of the software and thereby cause that version to be eliminated. Thus, in this revised approach with duplicates, the mutations to the duplicated code were neutral with regard to the selection process.
The programmer dubbed this version of his system N.E.C.R.O. (Neutral Errors in Copying, Randomly Occurring). He realized that even with these changes, his system would not yet fulfill his hopes. Nevertheless, he looked upon it as another step of exploration. In that respect it was worthwhile and more revealing than he had anticipated, leading the programmer to several observations as he reflected on the nature of its behavior.
Under these conditions of freedom to change without being selected out for loss or impairment of current function, what should we expect to happen to the duplicated code sequences over time and over many generations of copying?
And why?
[p.s. Sincere thanks to real computer programmer OMagain for providing the original seed of the idea for this tale, which serves as a context for the questions about Neutral Errors in Copying, Randomly Occurring.]
Thanks for posting.
My expectations would be very low. I don’t see this as a realistic version of evolution. But that’s good. This is a thread where we can discuss why.
The most obvious problem is that computer software is based on logic gates. And systems based on logic tend to be very brittle. A small error can cause catastrophic failure. By contrast, evolution is based on homeostatic biochemical processes. And because they are homeostatic, they are more robust. They are self-adaptive. You change things a little, and they adapt. That’s the homeostasis at work.
I’ll keep this short, and come back later with a comment on another problem that I see.
I see this as a huge flaw in your thought experiment. I’ll grant that Darwinism is described in a way that may seem to suggest that as a selection process. But it doesn’t really work like a rigorous test suite.
Let’s suppose that I give a test to my class. And some of them find ways to cheat. The most successful cheaters, rather than those that withstand the rigors of the test, will be the winners. Biology is all about finding some way, any way, to survive and reproduce. It isn’t about passing a test suite. It is about finding a way to cheat, about finding a loophole in the test suite that can be exploited.
Additionally, you would need your test suite to be continuously changing, thereby accidently creating new loopholes (niches) to be exploited.
Neil Rickert,
Yes. Proto life could not pass the test to ‘be a whale’.
(shrug)
Wouldn’t be a day on the web without some know-nothing Creationist making another dumb and inappropriate argument from analogy between DNA and computer code.
Sigh.
One more round from me.
You really wanted to get at duplication. Suppose genes for teeth are duplicated, and suppose that this is not a fatal mutation. Then you possibly get more teeth. The organism uses the additional teeth in parallel with the other teeth, and can perhaps use a larger range of foods as a result.
In a program, you duplicate some code. But the duplicated code is another subroutine to be used in series, rather than in parallel. So this isn’t a reasonable analogy. Instead you would need something like duplication of the processor or duplication of a memory chip, or a kind of duplication in the manufacturing process that gives you 12-bit bytes instead of 8-bit bytes.
That really illustrates the problem with the information theory arguments used in ID. When you use those arguments you are looking at it as a logic problem, rather than as a mechanical problem.
What I don’t get is why they use analogies at all.
Analogies are for teaching young learners the beginning ideas of some subject that’s a little too hard for them to grasp in its more-accurate “grown-up” version. Using an analogy in a science subject is using baby-talk or pidgin. None of us need evolution explained in creationist-pidgin.
I remember one of my kids, a toddler just learning the names for things, first saw cows standing in a field. Doggies! Big doggies! Well, we laughed, but we had to agree: they were “big doggies”. It’s an analogy, accurate enough to convey much of the important similarity: animals, 4 legs, big heads, black or tan, with spots, standing in grass – much like the dogs in the neighbor’s yard. Of course we expect children to learn the real names for things eventually. Of course we expect children to learn that the analogy only goes so far; there are important differences between dogs and cows as well as the fun similarities.
Even more, we expect them to mature and to learn a meta-lesson: analogies work when the two items have a connection based in a valid model of reality. Analogies don’t work when the two items are arbitrarily juxtaposed; then what you get is a poem or merely a joke. Poetry is great, jokes are great, but they’re not very useful tools for doing science.
I don’t understand how folks like ericB fail to learn that lesson. I blame parents and teachers; maybe they shouldn’t let ’em keep saying “Big Doggies!” all through their school years.
😀
Not completely OT:
Here’s a Spanish riddle which shows how analogies turn into jokes when the two things are unconnected in reality – with a bonus Ray Comfort prop!
¿Oro parece, plata no es?
Plátano!
[What is] gold in appearance, [and] is not silver?
You can figure out the answer yourself. 🙂
The simplest response is that typos in DNA are often neutral. Otherwise there would be no variation and everyone would be a clone.
I believe a recent study showed that the majority of point changes in a protein coding sequence made little difference.
If analogies are allowed, a better one would be English spelling, which has slowly chsnged over the last thousand years. We have the change history in pathetic detail.
Interestingly, the history of indo-european languages went through a kind of cambrian explosion, with new languages appearing all over Europe. Writing seems to have halted the creation of new languages, and printing seems to have resulted in conservation of spelling.
The difficulty is that code is not DNA and does not operate under the same constraints.
A change in DNA might lead to the expression of more or less of a protein. A change in code would on average lead to “no protein”.
I suggest a better analogy would be to have the “equations” subject to evolution rather then the entire code base.
So let’s say that our target was to sort a list.
We have a collection of components to do with sorting lists.
E.G. Select the first element
Select the last element
Move an element to position N
etc
Putting those components together in any order then passing a list into it will always result in a list coming out, by design. But how well sorted that list is when returned is a function of the evolving “black box” of the sort operation. Regardless of how broken the sort actually is, a list is always returned.
So, now it never “breaks” (or perhaps rarely if designed to fail on some arrangements) and you always get a list out if you put a list in. But the order, structure, arrangement of the components inside the back box is subject to evolution. The time taken to perform the sort is also a factor.
I might implement something like this, you’ve got me going. Who knows, perhaps a better sort is out there hidden in sort-space.
Relevant: http://panthema.net/2013/sound-of-sorting/
If ericB’s little scenario is supposed to be some kind of analogy about evolution or abiogenesis, there is nothing whatsoever in his portrayal that has any resemblance to anything in biology, chemistry, or physics.
As near as I can tell, his “program” has an objective or a targeted purpose, but he doesn’t state what that is. Evolution and abiogeneis do not have targets.
There are no charge-to-mass ratios for computer instructions. Computer instructions don’t obey the rules of quantum mechanics. Computer instructions don’t form compounds made up of computer instructions.
There are no acids or bases. There are no covalent bonds, no ionic bonds, no hydrogen bonds, and no Van der Waal potentials. There are no temperature dependent probabilities associated with bonds; one would certainly expect to see something like a Maxwell-Boltzmann probability distribution or a partition function for the bonded states of computer instructions. There are no hydrocarbon computer instructions. There is nothing about computer instructions that form amino acids and other organic compounds.
There are no primary, secondary, tertiary, and quaternary structures with computer instructions. Computer instructions don’t coil like hydrocarbon and polymer chains in the presence of water into complex shapes that further determine their properties. There are no catalytic computer instructions. How do the first and second laws of thermodynamics affect the formation of computer instruction bonds?
Computer instructions don’t behave in any way like atoms and molecules. Using computer instructions as stand-ins for the properties and behaviors of atoms and molecules is as ludicrous as using junkyard parts, or Scrabble letters, or dice, or coin flips as representations of atoms and molecules.
So the only point I see from ericB’s little scenario is that it is irrelevant to anything having to do with biology, chemistry, and physics. But we already figured that out on his previous thread. So what is ericB trying to tell us?
Perhaps ericB could be a little less coy and explain what this scenario is all about? Make the one-to-one correspondence between his computer instructions and something having to do with biology, chemistry, and physics; if, in fact, this is what his tale is all about.
Until he can do that, we have no idea what he is trying to prove. How does one respond to a complete non-sequitur?
I think you are fundamentally right. There are no analogies to chemistry.
You can exercise various GA scenarios, but you cannot anticipate uncharted locations in chemistry space.
Which is why biological evolution feels its way, moving by Braille.
I have for several years asserted that design is impossible without evolution. I think it would be interesting if someone conversant with chemistry would post an OP discussing the problem with analogies to chemistry.
To answer your question:
However, this reveals just how appallingly bad your analogy is.When a gene duplicates, the two copies are equivalent; neither one is commented out. Both functions are getting called, and (since gene expression is not computer code) there is no “ambiguous function call”. However, both copies possess a series of parameters that determine when they get called. From their identical starting values, both the “when” parameters and the underlying functions of each duplicate can diverge, creating new functionalities. Interesting things result.
Analogies can be very useful in conveying concepts. As an example, think of Einstein’s use of the analogy of traveling in an elevator, as part of his explanation of general relativity.
Let’s give ericB the benefit of the doubt, and assume that he is using an analogy to help him explore and attempt to grasp the concepts. I’d say that most of the comments in this thread have been helpful from that point of view. Whether ericB can actually grasp the concepts, I do not know. ID proponents seem to have great difficulty with that.
Analogies are also very useful in science for formulating concepts to begin with.
Whether evolution is successful or not in improving fitness depends (in models of evolution) on the fitness surface. If it is extremely rough, with a great many local peaks, evolutionary change may achieve little.
When they get mathematical, “ID theorists” and creationists always seem to invoke extremely rough fitness surfaces. Examples include the default fitness surfaces in William Dembski’s “No Free Lunch” argument, and the surfaces from which we sample in Dembski and Marks’s “Search for a Search” argument.
Are these surfaces realistic? In my article on Dembski’s arguments I’ve pointed out a property that suggests that they are very unrealistic: on those surfaces a single mutation is just as bad for you as mutating every site in the genome! (See the section on “Mutation” in that article).
In biological systems this just plain isn’t true.
The programming analogy also has this extreme-roughness property. So, as Neil Rickert said above, “a small error can cause catastrophic failure”. There is no reason to take computer programs as good models of genomes of real organisms. Nevertheless ID types and creationists keep assuming that computer programs are a great analogy to genomes. I wonder why? 😉
I really appreciate that some good discussion is getting going. However, most comments have missed the central point and focus of the post.
So far, petrushka’s comment is the only one that even mentions the word “neutral”.
The focus of my post and this thread is to consider a somewhat neglected topic, i.e. the effects, direction and tendencies of Neutral Errors in Copying, Randomly Occurring, hence the emphasis even in the title on N.E.C.R.O..
For the purposes of the discussion, the context of a software program or the test suite, etc. is window dressing or a frame to the central question of what is happening to the duplicated sequence due to random changes that are not either detrimental or advantageous to the reproductive success.
So, while people have made some legitimate points about the differences between computer code and biological systems, they are not relevant to the question at hand that this post is directed toward. I’m not asking about the brittleness of the uncommented software. I am asking about sequences undergoing random neutral changes — and especially about sequences that do not yet provide selectable reproductive advantages for the entity/organism.
Though without referencing the word “neutral”, DNA_Jock has begun to answer the actual question of the post.
I should point out that the original context for raising the subject (before making it into its own post) is not biological evolution, but rather in the area of the origin of life and prebiotic evolution or at least pre-protein evolution related to the origin of features such as the translation system and specifically its subsystem of the Triplet-Reading System.
I would have considered putting the post under the category of Origin of Life, except I was amazed to find that TSZ doesn’t even have that category yet! (Moderators, couldn’t that be added?) Someone else gave it the category of Evolution.
In the intended context, the question is about the gradual development of systems that do not yet make any advantageous contribution to reproductive success — but that might do so eventually. Hence the focus on random changes that are selectively neutral.
Even in the context of cells with genes, while it may be that duplicated genes could both function initially, one key potential evolutionary advantage of duplication over having only one copy is that one of the copies can go on performing the original function without loss, even if the other (whichever one changes first) is modified to the point where it no longer fulfills that original function. Thus it is potentially free to become something else, without detriment to the organism due to the loss of the reproductive advantage from the original function.
If the organism ceases to depend on one of the copies, then random changes to it may likely be neutral with regard to natural selection — unless and until it either becomes detrimental or advantageous to reproductive success. This thread is about considering what happens to a sequence while under the influence of random neutral changes.
I’m not certain, but I think you could have added that category yourself.
I just tried, but the software told me that you are currently editing the post.
I have created an OOL category. I’ll try switching this post’s category later.
Thanks! I’ve just changed it to use your new category.
I saw the list of categories, but I didn’t see an obvious option for me to add a new one. I’m new to this system.
This is why I don’t consider myself a Darwinist. I see the whole notion of neutral or beneficial or deleterious as flawed. Whether a mutation is neutral or beneficial or deleterious depends on the ecological niche. A mutation might be neutral, or even deleterious, in the current niche. Yet it might allow the organism to expand into a different niche where that mutation turns out to be greatly beneficial. And if it is sufficiently beneficial in that different niche, you might get an “explosive” expansion into that new niche, perhaps producing a “punctuation” in the “punctuated equilibria” picture that Gould argues.
You really cannot understand evolution without looking at the importance of changes in the environmental niche space to which organisms adapt.
Perhaps a good place to start would be the Lensky experiment.
Let me expand on that, but in the context of computer programming.
The original program IEFBR14, used on IBM mainframes, consisted of one instruction (the instruction was exit). It was later changed to consist of two instructions (set return code to zero; exit).
If you think of that change as a mutation, then it was neutral or mildly deleterious. However, the computing milieu changed so that programs giving an exit code became important. The mutation was very beneficial in the newer computing milieu.
See the Wikipedia entry for details.
I confess that I had forgotten that you were posing your analogy in the context of the evolution of a Triplet-Reading System (not abiogenesis, btw), probably because your analogy makes even less sense in that context. However, my response stands, since it does not depend in any way on the genes being protein-coding genes.
There’s an assumption here that one of the copies sits tight, keeping the homefires burning, while the other is free to go walkabout. That is only one of many possible outcomes.
Going out on a limb here, I do not consider that ‘neutral’ mutations exist. However, many mutations have such a tiny selective effect that any signal is swamped my the noise. If you duplicate a gene, many more mutations fall into this category…
It’s called “drift”.
Speaking of drift, is there some point to the OP?
The analogy to computer code fails because biological code is not brittle. What’s left?
The following are a few initial observations related to Neutral Errors in Copying, Randomly Occurring. They are numbered for convenience of reference. If you object to anything, please indicate specifically which observation(s) and the specific reason(s).
[Reminder: While the issues of random neutral change do have applications within aspects of biological evolution, the original and primary context is one in which even proteins and many other features of a cell as we know it do not yet exist. The origin of these features (e.g. of the Triplet-Reading System where no ability to read nucleotides as triplets previously existed) raises issues about sequences that do not yet contribute a reproductive advantage.]
Observation #1: In general, there are three possible results from a random copying error.
1. The copying error is disadvantageous or harmful to the reproductive success of that instance compared with other instances that do not have that copying error. To the extent this is so within the current environmental context, copies with such an error will tend to be weeded out by natural selection.
2. The copying error is advantageous or beneficial to its reproductive success of that instance compared with other instances that do not have the copying error. To the extent this is so within the current environmental context, the frequency of copies with the copying error will tend to increase due to the influence of natural selection.
3. The copying error is neutral — it has no significant advantage or disadvantage with regard to the reproductive success of the organism within the current environmental context. The selective effect, if any at all, is so negligible that any signal is swamped by the noise.
For the questions at hand, the focus is on the third category of neutral copying errors and their effects and tendencies.
Observation #2: For any sequence that is not providing a selectable reproductive advantage, random neutral errors in copying will accumulate.
By definition, since they are neutral, they are invisible to elimination by natural selection. Subsequent copying will tend to copy and retain the neutral copying errors. (Corollary: To the extent that a random error is neutral, this will also be true in sequences that are providing a selectable benefit.)
Observation #3: For any sequence that is not providing a selectable reproductive advantage, random neutral errors in copying will accumulate randomly throughout the sequence.
By definition, the location of the copying errors is random. Therefore, any part of the sequence could incur a copying error.
Observation #4: For any sequence that is not providing a selectable reproductive advantage, over time the accumulation of random neutral errors in copying throughout the sequence will randomize the content of the sequence.
So long as the changes remain neutral, it is only a matter of time and increasing probability before every location in the sequence incurs a copying error. The inherent trajectory of the process is to reduce the starting sequence, whatever it was, into one that is not effectively different from the process of generating a completely random sequence from scratch. Its nature is to move toward random content at every location.
From the OP:
We expect them to change to random strings of characters. If the duplicated code sequences are never called by the program, then the program will still work. No harm has been done, but nothing interesting has happened either.
As for the programmer’s “hopes”, ask a psychologist.
The last paragraph is nonsense. Not all substitutions are neutral, so the accumulation of neutral errors will only not affect every locus, nor will it include every possible value.
As genomes are refined, the percentage of allowable substitutions diminishes, for the simple reason that it is easier to degrade than improve. Every creationist knows that. And by observation we know that many sequences are conserved.
ericB,
Higgins: I think she’s got it. I think she’s got it.
Eliza: The rain in Spain stays mainly in the plain.
Higgins: By George, she’s got it.
Yes. Flat landscapes are traversible.
But you are assuming that, because part of Spain is a plain, the whole country is. Wander around the soggy plain for long enough, and you will happen upon a stream.
[I have always thought that hill-descending is a better analogy than hill-climbing: absent barriers, a population WILL move to higher fitness]
Of course the topography is constantly changing too. Baby steps, baby steps.
Could we say that N.E.C.R.O. is D.O.A.?
I assume Eric meant this as humor.
What IDists seem unable to fathom is the fitness has dozens, perhaps hundreds or thousands of dimension, and movement in one may not have the same value as movement in another.
That is why differential reproductive success is the only metric that addresses all dimensions simultaneously. Without differential reproductive success, design would be very difficult.
And conversely, all “designs” with differential reproductive success would be impossible to implement and have it stay implemented as intended. As soon as differential reproduction kicked in, the created “design” could not possibly- not even in a thought experiment – remain in the state in which it was created, No matter how competent the “designer” to calculate beforehand what it intended to create.
Since we know that we do indeed live in a world with differential reproductive success – even die-hard creationists are forced to admit that fact – all design hypotheses for the existing characteristics of life are moot before they even begin. Unless, perhaps, one presupposes – as christians do – that the “designer” continues to meddle and interfere with its creations from moment to moment to keep its designs in accordance with its intentions, not drifting (or evolving) too far off track.
But that’s a whole ‘nuther can of worms, and a bag full of malaria parasites, too, which their “designer” must at minimum be allowing to remain untouched while it interferes with all else in its intended creations. (That’s at best. Alternatively it must have specifically created malaria on purpose – what a delightful god they believe in!)
Aside from the fact that ericB’s “analogy” has nothing to do with biology, chemistry, and physics; he also appears to be thinking in the context of having a target. Otherwise, what does neutral or deleterious or beneficial mean if not with respect to some imagined outcome?
That isn’t how anything in biology, chemistry, or physics works. Modifications to the templates of emergent structures tumble into different niches depending on the first and second laws of thermodynamics. Some of those structures survive longer than others. There is no targeting whatsoever; just whatever is energetically available at the moment.
Trying to find scenarios against the origins of life and evolution without any knowledge of science doesn’t get anyone anywhere except into more deeply ingrained misconceptions and misrepresentations.
As near as I can tell, he hasn’t clarified anything. He speaks no science; and his “analogies” make no sense. To what do they refer?
Well gpuccio, among others, believes in continuous undetectable meddling.
I’m not sure what the point would be, since the Lensky experiment indicates that realistic sized populations can explore all the possibilities in near space without assuming any intervention.
Unless one claims that the Lenski populations were actually guided by that “undetectable meddling” in ways that look exactly as if there were no intervention. But that makes ID no different from “intelligent falling” where apples are guided by invisible angels to fall in space in ways that look exactly as if there were no intervention.
As you say, what’s the point of that?
I suspect the point is to have your cake and eat it.
The more interesting result from the Lensky experiment is that realistic populations can “hold” neutral or near-neutral variants until they become useful in an “irreducible” function.
That’s always been the theory, but it’s nice to see it happen in real time.
rhampton,
Chaitin’s model of biological evolution is one arbitrary model, and there is no reason to believe that the patterns of interaction of alleles in his model is similar to real biological systems.
Joe Felsenstein,
There is no reason to believe that the scenario ericB lays out as a challenge is similar to real biological systems, either. But that’s not the point, is it? Chaitin’s experiment is the (only?) available empirical data, and it does suggest that ericB’s incredulity is unwarranted.
Exactly so. Spot on. The tendency of the effects of the random neutral errors will be randomizing on any sequence that provides no selectable reproductive advantage.
Even the “nothing interesting has happened” result is important to recognize. Too many evolutionists assume that even in such cases they are warranted to expect a different kind of result even with regard to a sequence that provides no selectable reproductive advantage.
Reminder:
Being specific about what you might think is in error is helpful to productive discussion. The observations are numbered to make this easy to do (or you can just quote the relevant observation).
You’ve noticed, of course, that the entire thread in general and the observations in particular are based on considerations of differential reproductive success, and what happens when a sequence in the entity does not contribute toward differential reproductive success.
As a general point, you shouldn’t assume that we are talking about the refined genome of a fully developed and operational modern cell.*
In particular, we are talking about those sequences (e.g. a portion of a larger strand from which a non-coding RNA transcript might be formed) that do not provide any selectable reproductive advantage to the entity.
That being so, why do you think any position of such a sequence would be immune from accumulating a copying error? What do you think would prevent that? (Genuine question, not rhetorical.)
[Edit: *p.s. Of course, as I mentioned earlier, some of these observations will have some applications to the context of fully biological evolution. I’m just cautioning that one should not assume this is the necessary context.]
Observation #5: Even a sequence that formerly provided a selectable reproductive advantage will no longer be protected from randomizing degradation if it ceases to provide a selectable reproductive advantage.
If a sequence is no longer contributing to reproductive success, then there is no selectable reproductive contribution to be lost from the loss of that function — no downside reproductively, which means it is invisible to natural selection.
This is why, for example, fish in cave pools with no light can lose their ability to see. Since the sequences that formerly enabled sight no longer provide a selectable reproductive advantage, they are no longer protected by natural selection from degradation.
Observation #6: If a sequence that is redundant (e.g. due to duplication) that formerly provided a selectable reproductive advantage is affected by a neutral copying error such that it ceases to provide a selectable reproductive advantage, then so long as this remains true, it will no longer be protected by natural selection from randomizing degradation.
This is just a restatement or corollary to Observation #5, but it is worth mentioning explicitly, since so many assume much from the mere fact that genes can be duplicated.
Observation #7: If a gene or any functional sequence is duplicated, it actually makes it much easier for mutations to be neutral, leading potentially to the situation of Observation #6.
If there is another copy that continues to provide the reproductive benefit, then there may be no reproductive downside from a change that disables a formerly functional sequence. As DNA_Jock correctly observed…
Hmmm…
Looks to me like ericB’s scenario includes the essential characteristics of evolution—(a) reproduction/duplication, with (b) variations in the duplicates, and (c) some way for the variations in the duplicates to influence the likelihood of a variant’s continuation or deletion. At the same time, the scenario has rather a lot of aspects which are not at all like those of biological evolution.
For instance, consider point mutations (i.e., mutations in which one particular nucleotide is replaced by some other nucleotide). By definition, a point mutation can only affect 1 (one) codon… and given the codon-to-amino-acid relationships instantiated in the genetic code, approximately 1 out of every 4 point mutations do not result in the point-mutated codon yielding a different amino acid. In ericB’s scenario, what is the likelihood that a random Change X will not result in any discernable effect on the unit-of-action which carries Change X?
As well, biological evolution is a massively parallel process. If you’re looking at bacteria as the basic units upon which evolution acts, the number of acted-upon units is the number of bacteria on the planet Earth… and that number is certainly higher than 10^20, with good reason to believe that it’s at least as high as 10^30. How many copies (acted-upon units) is the programmer in ericB’s scenario dealing with?
How ‘effective’ biological evolution is, depends in large part on the nature of the relevant fitness landscape. In real ecosystems, the fitness landscape is going to be relatively smooth; being 5% better at camouflage won’t yield as large of a benefit as being 15% better at camouflage, but a 5%-better-at-camouflage critter will nevertheless gain some benefit from their 5%-better camouflage.
So: Just how smooth is the ‘fitness landscape’ in ericB’s scenario? Will a 5% improvement in a trait yield no benefit at all?
Given the points of dissimilarity between ericB’s scenario and actual biological evolution, it is unclear that any results derived from ericB’s scenario will be applicable to, let alone binding upon, actual biological evolution.
ericB,
In the case of genotypes where the genes function in parallel, not as parts of a serial program, things are different. A piece of code that is never called might evolve neutrally, a duplicated gene that is expressed does not. (Duplicated genes that are not expressed often do decay away just as the chunk of code does in your example). So “evolutionists” are often justified in regarding duplication as providing potential raw material for differentiation of gene function.
I’m at a loss as to where this is going. Critical gene function is preserved by purifying selection. It isn’t going to stray off the reservation.
Meltdown doesn’t seem to happen in bacteria, so there must be something about ratchet theory that doesn’t model reality.
What am I missing.
If there is one thing you can be sure of, it is that a “computer lesson” about evolution or the origins of life given in only one dimension tells us almost nothing about the multidimensional real world.
We haven’t been shown why shuffling computer instructions around is any different from shuffling junkyard parts or Scrabble letters around.
EricB hasn’t said what laws of chemistry and physics are operating in his scenario? More to the point, he hasn’t even acknowledged any such laws exist or are applicable.
This definitely isn’t going anywhere.
As I understand it, Eric’s argument is as follows:
A) When a gene duplicates, the original continues to perform the ‘original’ function, and the duplicate is free to ‘accumulate neutral mutations’ since it is under zero selective pressure.
B) Under these circumstances, ‘neutral’ mutations include those that render the duplicate non-functional, thus the duplicate will rapidly become space junk.
His first error is in (A). Immediately following duplication, the two copies are equivalent. They can subsequently diverge.
His second error is his assertion that the ‘duplicate’ is under zero selective pressure. No, it is under reduced selective pressure. [As an aside, his focus on a pre-protein world with fewer homeostatic mechanisms is a GSW to the foot here]
His third error is his assumption that ‘neutral’ changes will always remain neutral. This is really glaring, as his descriptions of beneficial [Obs#1.1] and harmful [Obs#1.2] mutations include the phrase “To the extent this is so within the current environmental context“, which he strangely omitted from his description of neutral changes [Obs#1.3]. Thus he is defining ‘neutral’ changes as ALWAYS neutral, and he is denying the possibility of beneficial drift by fiat.
Eric, if you insert this missing phrase into your definition of ‘neutral’ mutations, I think you will find that your problem disappears.
Alternatively, you could think about it in terms of my plain in Spain: just because the landscape is locally flat(ter), it does not follow that it is flat everywhere.
Muller’s Ratchet is unsprung if there is more than a minimal amount of recombination, if population size is large enough, or if there is enough back mutation. Advantageous mutation at a high enough rate can reverse it too. Just because the phenomenon exists doesn’t mean it is important.
Good points. The only slight qualification that I would make is in regard to, “a duplicated gene that is expressed does not [evolve neutrally].” I agree with that statement so long as the expressed genes (both original and duplicate) are contributing a selectable reproductive benefit.
One of the points I have been making (e.g. Observation #7) is that since the duplicates provide a redundant benefit, it becomes easier for random copying errors that are neutral with regard to the reproduction of the organism to change one of them (either one) such that it no longer provides the original benefit that is advantageous for reproductive success. (That is, after all, the point of “regarding duplication as providing potential raw material for differentiation of gene function.” One of them can change without losing the original benefit.)
But, even though still being expressed, the fact that it is expressed alone does not guarantee that it continues to contribute a selectable reproductive advantage after such a change. It might or it might not. If it doesn’t, then it has moved into the category I describe in Observation #6.
So I would say that being expressed is a necessary but not sufficient condition to retaining the protective benefits of natural selection. After all, the genes for vision were still being expressed (initially) for fish that moved into cave waters without light (Observation #5). Since those expressed genes no longer provided a selectable reproductive advantage, they no longer had the protection of natural selection — even thought they were initially being expressed just as when that feature had provided a selectable reproductive advantage.
Therefore, I would qualify that expression is not sufficient and that the necessary condition would have to be providing a selectable reproductive advantage. Otherwise, the copying errors are free to accumulate randomly without adversely affecting the organism’s differential reproductive success.
EricB keeps referring back to this assertion:
EricB keeps using this expression; but what does “randomizing degradation by natural selection” actually mean?
What is it about natural selection that does “randomizing degradation”? Sanford’s “genetic entropy” perhaps?
Something doesn’t smell right here; ericB seems to be jerking everybody around. I’m seeing ID/creationist misconceptions and misrepresentations at work; and ericB is refusing repeatedly to answer direct questions about his scientific knowledge.