If evolution is not a search, why is the term “evolutionary search” not an oxymoron?
Over at Uncommon Descent Elizabeth posted the following:
“…any “search” algorithm worthy of the name of “evolutionary search” comes with its own moderately smooth fitness landscape built in.”
So evolution is a search if it comes pre-built with its own moderately smooth fitness landscape built in?
I don’t see evolution as a search. And I don’t talk about “evolutionary search.”
Good for you Neil. But why not? Do you think it’s not possible to model evolution?
Taken literally, evolutionary search is an oxymoron. But we often use analogies to simplify the understanding of a concept. Selfish gene is another example. It was never intended to imply that a gene is “selfish”, it was just a different way to look at natural selection.
Neither evolution nor genes have any goals. They just are. The “search” is provided by the environment. Not that evolution is actively searching for anything but if the natural variation within a population contains traits (adaptations) that will allow some individuals to perform better (ie., reproduce more effectively) under the new environment, it’s traits are more likely to survive and increase in frequency in subsequent generations. From the outside it gives the (false) impression that evolution “searched” out this solution. But this is simply applying our vernacular to the process. If the necessary adaptations did not exist in the original population, or if they didn’t arise through mutations or other non directed processes, the population may become extinct. All you have to do is look at the fossil record to know that extinction is a very common event.
ok. so why did you feel compelled to qualify that statement?
And the winning analogy is?
mung, are you suggesting that these types of simplified concepts, and incorrect terminologies are not used elsewhere? That they are not used in ID? The incorrect use of terms is used in every human endeavour. In my field, analytical chemistry, we use the term measurement uncertainty when what we are really talking about is confidence levels.
I don’t really understand your argument. I agreed that the term “evolutionary search” is an oxymoron. So are “jumbo shrimp”, “selfish gene” and “specified complexity”. All you are demonstrating is that the misuse and abuse of language is universal. I don’t know anybody knowledgable of evolution who interprets the term “evolutionary search” as if evolution is actively searching for a solution to a problem.
Evolution is adaptation to a changing environment. Any model would have to model that dynamically changing environment. And the way we model that is likely to be very artificial.
Acartia, if you agree with me, then stop disagreeing with me.
ok. score one for the anti-elizabeth team.
“ok. score one for the anti-elizabeth team.”
Then you obvious have no idea of the context under which she said this.
Well gee Neil, that helps alot. Thanks.
Any model that suggests that evolution can be modeled as a search is artificial.
Arcatia, if you think that Elizabeth had something interesting to to say, do tell. It’s time to put up or shut up. I don’t have to post here at TSZ, and don’t have to suffer idiots in threads that I author.
Well Mung, if you ever showed any intent to have a good faith discussion instead of merely trolling with childish semantic nitpicks you might be taken more seriously.
That hurts Adapa. That really hurts.
But is “evolutionary search” an oxymoron?
It’s a metaphor.
In the phrase “water seeks its own level” do you think that means water goes out and actively searches all the boundaries of its container?
Adapa: “Evolutionary search” is a metaphor.
ok. it’s metaphor for what?
Does Mung think that if we don’t consider evolution to be a search, that this means we can’t model evolution? A couple of comments:
1. There is a large literature of mathematical modeling of evolution. In 1981 I put together a Bibliography of Theoretical Population Genetics. It was pretty comprehensive, the only attempt at such a comprehensive bibliography. It turned out to have 7,982 papers listed. By now there must be at least several thousand more.
2. Of course it is possible that Mung doesn’t think that any of these is really a model of evolution. In that case Mung is welcome to having Mung’s own standards as to what is a model of evolution. And Mung may consider those standards to be higher than those of the theoretical population geneticists, but there is simply no reason to take those standards seriously.
Of course I am imagining what Mung’s position on that would be. So let’s hear from Mung: do you think that only models that have a “search” are really models of evolution?
Mung at Uncommon Descent:
Bingo! Mung’s “unless” wins! DEM considers as a “search” any process that ends up coming up with a point in the space (say, a genotype). It can be stochastic or deterministic. Even if it always comes up with the same outcome, that is of course a probability distribution, though a pretty uninteresting one.
It’s a metaphor within the larger metaphor of “search space”, which is the conceptualisation adopted, among others, by Ewert, Dembski and Marks, in which some configuration of bits and pieces has some properties that make it remarkable in some way (a “target”), and is a relatively rare configuration amongs all possible configurations of bits and pieces.
Any process by which such a configuration can be arrived at can be considered a “search” for one of those configurations, of which a process in which each configuration is equally probable is very inefficient, although not as inefficient as a process in which the “targets” are less probably than other configurations.
But the word “search” and “target” are metaphors based on the idea of looking (“searching”) for a “needle” (“target”) in a haystack (“search space”), and the various kinds of processes that can result in a “target” being “found” (both more and less efficient processes) are metaphorically referred to as kinds of “searches”.
Evolutionary processes are such processes. So they are sometimes referred to as “evolutionary searches”.
The point being that “search” is not a terribly good metaphor, because in plain English, a “search” implies an intentional agent doing the searching, with a target (i.e. the thing she is searching for) in mind from the beginning. In this context, there is no agent, and the “search” is simply a system that makes certain configurations with specific properties (e.g. an organism particularly well-suited to thrive in its environment) more, or less, likely to turn up than others.
I agree with Aurelio Smith when he wrote:
You might say the environment is the searcher in the process.
According to Mung at we are:
So there you have it Elizabeth. You are an ignorant ranter and we are your sycophantic fanboys.
Adapa nailed it up thread.
Genotypes are not really ‘searching’ for alternatives, they simply … uh … ‘probe’ – no, make that ‘explore’ … or maybe ‘visit’ … oh, hang on I’ll get back to that one – the nearby genotype space and … uh … ‘find’ them anyway. Ah well, language eh?
Of course at another level, if we write a program and it uses evolutionary processes to locate an optimal target given the fitness function, then it really is a search. We can use evolution to search, but that does not make evolution a search in the same sense.
sez mung: “If evolution is not a search, why is the term ‘evolutionary search’ not an oxymoron?”
If
evolutionaluminum is not asearchladder, why is the term“evolutionary search”“aluminum ladder” not an oxymoron?^^ This ^^
I think “exploration” is my favorite word for describing what biological evolution does. I tend to think of it as a hill-descending algorithm, and imagine it as analogous to the way that water ‘explores’ the (changing) terrain it flows over. Of course the analogy breaks down – water doesn’t reproduce. In NO SENSE is biological evolution a search.
We can write optimization algorithms that mimic how we think evolution may work.They are awfully useful for finding optimal solutions, and could be referred to as “evolutionary searches” . We can separately write algorithms that model how we think evolution works in order to understand biological evolution. These are awfully useful for understanding how biological evolution may or may not work.
Strangely, Mung asks “And the winning analogy is?”.
Err, that would depend on the context. For any analogy. Do you understand what an analogy is, Mung?
I also found ” I … don’t have to suffer idiots in threads that I author.” troubling. Perhaps a topic for the moderation thread.
Joe @UD
Apparently, there is such an admission buried somewhere above. Could one of our brighter members tell me where it is?
There has always been within evolutionary biology the lurking suspicion that evolution has a direction.
Simple to complex. Primitive to sophisticated. Ape to human.
Chardin made this explicit. As did michael Dentin.
I see a hint of this in the “flowing downhill” metaphor. I prefer Gould’s treatment in Full House. A random or stochastic walk.
Alan Fox,\
Reality would disagree with you.
Bacteria actively mutate in the face of antibiotics. They in a real and true sense ‘search’ their genomes for a configuration that works. Since we know bacterial colonies act as one organism, the speedy replication and mutation process is the bacterial colony directing its own survival.
Any passivity on the part of bacteria would be suicide.
So unguided evolution is the real oxymoron. If organisms do not direct their genomes to mutate, then evolution is useless as a concept.
This is what DEM allude to in Ewert’s reply to Felsenstein and English.
Without a design starter-kit, evolution can do diddly squat.
In other words, unguided evolution is required to co-opt a slew of design tools in order for it to have any semblance of efficacy as an explanation for life. Think about it, the first cell, designed; multi-cellularity, designed; reproduction, designed,; endo-symbiosis, designed.
But those are just footnotes. The real heavy lifting happens after…er the heavy lifting has already been done. Got that? Unguided evolution draws a hefty paycheck for doing a jury job- telling all the designers what to do, then telling them they are lucky to be alive because they in fact actually dont exist, haha!!!
so its a real curiosity that they won’t drop the evolution charade, and just keep the design tools.
Do they mutate when antibiotics are not present?
Steve,
Numerous experiments readily demonstrate that mutations are already present prior to the application of the stressor. So unless they can see into the future …
I thought it was the designer that causes specific mutations?
He also wrote:
For any given discussion, there needs to be an agreed definition of terms. I am happy to stipulate, for the purpose of a given discussion, that “information” quantified as Shannon Entropy. But as we all know, Shannon entropy does not tell you how informative that “information” is. It simply tells you the capacity of the channel in question to carry information, and is quantified as -sum of p*log (p), where p is the normalised frequency of each pattern in the available set of patterns, and which, if the base of the log is 2, will give you a quantity in bits.
So I am not, of course, laying down any law as to what the term “information” must always mean, which would be ridiculous. But if people want to discuss information, they need to be clear what they mean by the term, as it has many meanings, and can be measured in many ways.
Well, there’s a sense in which this is true and a sense in which it’s not anything to boast about.
If you start simple, and diversify, there are more ways to be different that are more complex than the simple beginning than there are to be equally simple. So the average degree of complexity over time will tend to increase.
But also, it tends to be easier to add stuff while leaving redundant stuff in place, maybe duplicating the task, or doing something more specialised, than it is to think, hey if we did B, we could dispense with A. Which you see all the time in code (at least in my neck of the woods) – once you’ve finalised code that works, it’s more complex than it started, but also more complex than it would be if started again from scratch.
And evolution, unlike an intelligent designer, doesn’t get to start again from scratch.
However, what evolution does do, sometimes, is penalise complexity. So if you don’t need eyes, things with less resource-consuming redundant eye-stuff will tend to do better. So far from eyeless fish being evidence that evolution can only break things, not make things, they are evidence that evolution can not only make things more complex when complex is advantageous, but also more simple, where complex is a disadvantage.
Can you say what you mean by “active” and “passive” in this context? I agree that bacterial evolution is a different in a number of respects from the evolution of multi-cellular organisms, and that colonies are best viewed as “one organism”.
Yes, “explore” is a far better term for what biological evolution is doing.
From a mathematician’s perspective, “explore” fits into geometry, and “search” fits into logic. They are very different.
I don’t see that as “search”. The term “search” should amount to looking for what already exists. But the bacteria are re-inventing themselves, rather than searching for what is already present.
I see it as more like perception. Biology is carving the world into niches (a bit like categories).
Mung endlessly asks questions, but when Mung is asked a question there is no answer. Crickets.
“Time to put up or shut up”?
Deeply ironic.
On the (admittedly off-topic) question of whether bacteria mutate when antibiotics are not present:
I have heard of these guys who have an experiment that they are doing, that will test whether bacteria have mutations to antibiotic resistance when antibiotic is not around.
Their names are Luria and Delbrück.
From where I sit, the biggest problem with DEM is in the assumptions about the characteristics of landscapes.
If there is always, or nearly always, a viable stone within one step of where you are now, then the Darwinian/Malthusian strategy of over-propagation accompanied by culling will eventually spread away from the point of origin. As Wallace put it in his famous paper, “…The Tendency of Varieties to Depart Indefinitely from the Original Type .”
I’ve seen threads at UD that Praise Wallace at the expense of Darwin, but they never seem to notice this key concept: “Depart Indefinitely.”
Until DEM address what actually happens in biology and address the ability of chemistry to support indefinite variation, they are simply cranks A number of people at Pandas Thumb and elsewhere have argued that there is nothing mysterious about the property of chemistry that allows multiple sequences to have equivalent functionality. If IDist fail to address this directly, they lose the argument before they start.
It would be clearer what Reality would be disagreeing about had you cited it
Nope. In a changed environment (from absence of antibiotic to presence of antibiotic) any bacterium carrying a mutation that improves its survival in the new environment will proliferate. No actions required by the bacteria. The environment decides.
Nope. If anything is searching, it is the environment.
There’s no plan. Bacteria survive where they can. Clear a patch of fertile ground, leave it and watch it being colonized by plants. Is there a plan?
?
Nonesense. The environment is the guide and the designer. Organisms just survive where they can.
More on that, later.
ToE is an explanation for the diversity of life, not its origin.
I find this a bit garbled, Steve. Can you clarify?
There is no good explanation for the origin of terrestrial life supported by evidence, yet. And I’ll agree that most of the heavy lifting has to have been done prior to the emergence of eukaryotes.
Got what?
You’ve drifted into unintelligibility here.
Who is they?
I would argue there’s a pretty good analogy between producing mutations and spreading seeds. Natural selection for seeds is even Biblical. i’m surprised — with all the bad and too narrow analogies being tossed about — that the seed analogy is not widely discussed.
There’s no plan to the propagation and spread of flowering plants. You just produce a lot of seeds and spread them as widely as possible.
Something similar could be said about roots “searching” for water.
“Feeling for…” is a better analogy, really. And, specifically, “feeling for” only proximal needs, nothing beyond the next move.
It’s really the answer to why evolutionary processes are both better, and worse, than intelligent design: evolution is better at finding solutions at the end of unpromising alleys; intelligent designers (i.e. humans) are better at taking leaps and shortcuts, guided by far-sight.
The first takes more time, of course, but with modern computing, that is less of a problem and we can now avail ourselves of the advantages of evolved designs.
Although humans are also better at avoiding the need for kludgy retrofits (giraffe laryngeal nerves, for instance).
Everything is a search.
According to Winston Ewert, in a new post at UD, “When we say search we simply mean a process that can be modeled as a probability distribution.”
So there you have it. Everything is a search. That seems to make the word “search” rather useless.
Joe, you expect an instantaneous response?
You responded on April 30 and here I am on May 1.
Too long for you?
Crickets? Really? Where I am, it’s frogs.
Joe Felsenstein:
You can model evolution any way you want. Use tinkertoys or an erector set for alI I care. Lincoln logs were always kind of cool. Model evolution as a search if you wish.
So no.
The work of Dembski, Ewert and Marks addresses these so-called models of evolution.
My interest is with models that use search and how claims that models that use search demonstrate that evolution is true can be reconciled with claims that evolution is not a search.
Do you know of a computational model of evolution that doesn’t use search?
So again, no.
AFIAK, no one knows how to model evolution. It just happened, that’s all, is not amenable to scientific methods.
Neil Rickert:
Not really. Lizzie loves operational definitions. You should too Neil.
You can’t think of a single thing that cannot be modeled as a probability distribution? Not even evolution?
Joe F:
I’m looking Joe, for that reason why I should retract and admit I was wrong. You’re not helping. Yet.
What space? Which space?
The process you mean. Yes, you’ve got it! The alternatives? You failed to mention the alternatives. Is that because there are no alternatives?
You mean a deterministic process. I’m not sure I agree with you that a deterministic process always results in the same outcome, but hey. So what about the other end of the spectrum? What about the alternatives that lie in-between?
Is evolution deterministic? You lose.
Is evolution stochastic? You lose.
So yes, that means I was right, and you were wrong. But will you admit it?
If search implies “search for”, then evolution is not a search.
The important difference between evolution and design is that evolution is a walk without preferred direction. One can build evolution into software and give preference to a direction, but that would not be a good emulation of biology.
Joe Felsenstein.
Did you and TS review the book by Gregory Chaitin? What did you have to say about it?
Of course it is possible that Joe thinks all of these really are a model of evolution.
Having read them all, Joe can tell us what they all have in common. Having read them all, Joe can tell us what the elements of a model of evolution consist of.
After all, isn’t that what we all really want to know?
Allen Miller:
Such humility Allen 😉
It is very common to use probability in population models of evolution. But it isn’t being modeled as what is ordinarily considered a search.