Seriously, are the ID proponents at UD ever going to wonder why Gould and Eldredge remained persuaded that common descent occurred, and that “punctuated equibrium”, although contrary the uniformly incremental pattern that Darwin envisaged, was nonetheless consistent with Darwin’s proposed adaptive mechanism of heritable variation in reproductive success?
Because Darwin was indeed wrong about uniform change. Unlike us, he didn’t have computers with which to model the predicted output of his mechanism. Indeed he didn’t even know what the vector of heritability was. We do. Here’s a sample output from Eureqa, a program that uses Darwin’s proposed mechanism to “evolve” equations to fit data:
Look at the bottom left plot. It records the best-fitting equations as they evolve. On the vertical axis is the “error” in the evolving equations – the better the fit, the smaller the error. On the horizontal axis is the complexity. The program is set up so that complexity carries a penalty in terms of reproductive success but accuracy carries a reward. The most efficient equations – those that give best accuracy for least against complexity – are shown more green, while less efficient equations are shown more orange or red.
What happens over time is that as the equations evolve, there are discontinuities in the best error rate: note the step changes on the vertical dimensions. From time to time a small change in the equation will occasionally introduce a large improvement in accuracy. However, reductions in complexity tend to be more gradual. As a result, we see “punctuated equilibrium” – step changes in accuracy followed by gradual reductions in the equation’s complexity.
And the system is entirely Darwinian. Darwin couldn’t know that this is what his theory would actually predict. Of course he was right that adaptation would be incremental – and it is – but it is more incremental at the genomic level than at the phenotypic level. A small DNA change can result in quite a large phenotypic change. Again, Darwin could not know this. But, even phenotypic changes are gradual. The key point is that the rate of change is not uniform – indeed, uniform rates of change turn out to be very unlikely under the Darwinian mechanism. If one of the evolving equations in Eureqa gets a good “idea” (as in AVIDA) then there is very rapid change for a while as the population optimises itself to this newly available resource, followed by diminishing returns as a local maximum in accuracy is approached. Until something else happens – a novel mutation along yet another dimentions allows the population to exploit a whole new resource, following which, again, stasis is approached, and is maintained as long as the resource remains, and the population is not outcompeted by another lineage.
And that’s before we even consider that small populations will tend to adapt faster than larger ones – or die.
h/t to whoever introduced me to Eureqa! It’s brilliant, but I’ve forgotten who it was!
It was me, obviously. Did you get the full product?
Look at AIC, possibly a good frontier for evolutionists and IDists alike.
Thanks again, then! I’m still playing with it, but plan to use it in anger shortly.
Also we should note that Eureqa does not presuppose ant particular shape of solution (as a linear regression might, for example) allowing us more fully explore the higher dimensions of the solution space.
You are Wrong. Darwin didn’t expect a uniform rate of change.
Doesn’t that mean that we should cease placing our faith in Darwin the substitute God, though?
That does seem to be the “reasoning,” which would be fine if it weren’t so baseless.
Glen Davidson
Hey Lizzie – my good friend Wes Elsberry points out:
“The curve Lizzie provides is the Pareto frontier, which has nothing to do with timing. The curve dealing with timing is on the “Start Search” tab in the latest version of Eureqa, and it shows the fitness on the Y axis and log-transformed number of seconds of runtime on the X axis. …. There’s a population of solutions that the Pareto frontier summarizes “locally dominant” instances out of. “
It depends on the Darwin. He learned, he changed, he evolved, he dealt with problems. Now I don’t know about “uniform rate of change,” which seems unlikely from the start (would the evolution of flight really be uniform?), but the early Darwin promoted gradualism. The sixth edition sees him pulling back from “Darwinistic gradualism,” it’s true, but it’s pretty obvious in the first edition:
First edition of the Origin, pp. 480-481
He compares his evolutionary view with Lyell’s quintessentially uniformitarian view, even if he perhaps wasn’t pushing for strict uniformitarianism. I think he became more and more aware that speciation really wasn’t well-documented in the fossil record, and moved toward more episodic and localized speciation in later editions, but I think that the gradualistic view in his early work tended to overshadow the fact that he later moved away from such a view.
Glen Davidson
My bad. Can you support this?
Yes, I knew that – did I imply otherwise? didn’t mean to! I chose the Pareto plot because it illustrates a two-dimensional fitness landscape.
I guess the time-plot might have been clearer.
I believe that passages like these are at issue:
http://www.talkorigins.org/faqs/darwinism.html
The second doesn’t actually involve “non-uniformitarianism,” but does deal with one reason why “abrupt changes” would seem to occur, of course.
The first does anticipate a kind of “punctuated equilibrium” reasonably well, in fact. But it tended to be forgotten because Darwin’s earlier view was largely Lyellian, with gradual changes occurring reasonably uniformly through time.
Glen Davidson
Or page 279, here
http://www2.hn.psu.edu/faculty/jmanis/darwin/originspecies.pdf
Reference courtesy of AtBC.
Not that Darwin was perfect, but he was aware of varying rates of change.
Still gradual, even with puncs.
An odd post from Querius:
um.
Substitute ignorant and wrong for odd. Nothing personal.
It’s just weird to me that someone can say so many things that aren’t true, without citation, as though they are.
I don’t suppose he’s lying. But why does he think that those statements are true?
Lizzie:
Could you please explain what you are testing ? What data are you using?
” Of course he was right that adaptation would be incremental – and it is – but it is more incremental at the genomic level than at the phenotypic level. A small DNA change can result in quite a large phenotypic change. ”
May you explain this please. Any example?
I don’t think this is strictly a model of punctuated equilibrium. Eldredge and Gould tied stasis and rate of change to speciation – to cladogenesis, and differentials between large established populations and small peripheral ones, with subsequent incursion across the parent range (and hence vertically above in strata). It’s a combination of pop-genetic mixing and range change aiming to explain the pattern of discontinuity in the fossil record.
If one plotted instead the actual genetic change occurring in a particular lineage, one may or may not see a close correlation of the slopes with the morphological pattern revealed in fossil strata. It certainly wouldn’t be steady, and I don’t think even Darwin thought it would be.
And it doesn’t have a great deal to do with the size of individual generational ‘steps’. Fossilisation patterns are at far too coarse a scale to reveal much about that.
Interesting discussion. One can also look at the works of George Gaylord Simpson, Sewall Wright, and Ernst Mayr, three of the major figures of the Modern Synthesis, for discussions of rates of evolution that did not assume uniform rates of change. As for the advocates of Punctuated Equilibrium, Stephen Jay Gould certainly tacked back and forth a bit as to whether he included nonuniform change by individual selection in Punctuated Equilibrium. (By the way, Punctuated Equilibrium is very much not dead among paleontologists, though not as widely popular as in the 1980s).
As to whether the behavior of the Eureqa system models evolutionary forces, the same could be said of models of long-term evolution like Avida, Gregory Chaitin is lately pushing (as “metabiology”) the analogy of mutation and selection in a one-individual population of programs to solve the n-th Busy Beaver computational problem. Add to the mix Stuart Kauffman’s NK model.
I’d say that all of these are interesting, but that their long-term behavior is strongly dependent on how their fitnesses relate to their genotype spaces. Biologically, all of them are bizarre. We see the same issue when creationist software engineers argue that mutations in computer programs will only destroy their functioning, so therefore evolutionary processes must behave likewise.
I agree, AIC is important for evolutionary biologists.
😉
Seriously, did you mean the American International College, becoming an Associate In Claims, the American Institute of Constructors, the Agricultural Issues Center, the American Iranian Council, the American Islamic Congress, or the American Institute of Chemists? I get those when I search on “AIC”.
It’s how the program works. The specifics of the data is irrelevant. Why not download the trial and see for yourself?
Hi Joe!
http://en.wikipedia.org/wiki/Akaike_information_criterion
Not to be confused with Joe Gallien’s “A CAKE” information criterion 😉
I’ll have to put this here:
“We have learned from long experience that if Eureka uses Darwinian mechanisms nothing interesting will result. If something interesting results, it came about by something other than Darwinian mechanisms.” – Eric Anderson at UD.
It must be great knowing stuff up front without studying it or doing any work. Don’t waste your powers in the Culture Wars, Eric, when so many crimes remain unsolved!
So it was the AIC I first thought. I linked to that, but as a joke, because I don’t see the relevance to this discussion. Then I looked for other AICs and found irrelevant ones.
So how is this AIC relevant to the discussion here? Basically it’s a statistical method for penalizing models for the number of their parameters in doing assessment of their fit, using likelihoods.
The efficiency frontier created by Eureqa uses AIC. It trades off descriptive power for Parsimony.
just a comment on the phrase “Darwin was wrong”. Lizzie is using it somewhat facetiously. Keep in mind that:
For most of the public, it means that Darwin was wrong about common descent, or wrong about the use the natural selection to explain why there can be so many adaptations. Those are the only issues for which they want to know whether Darwin was wrong. They really don’t care about whether Darwin was wrong about, say, the Parallel Roads of Glen Roy (which he was).
Creationists and ID advocates see evolutionary biologists as treating Darwin as figure for cultlike veneration. They endlessly tell their followers that evolutionary biologists worship Darwin and are unwilling to admit that Darwin was ever wrong about anything. This is a bizarre distortion (for example, that Darwin was wrong about pangenesis was freely admitted since, and even before, the rediscovery of Mendel’s work). When we acknowledge that Darwin was wrong about one thing or another, they regard this as a dramatic admission that our veneration has failed. Whereas in truth statements by evolutionary biologists that Darwin was wrong about this or that are fairly common.
I’m not expecting much disagreement on these points here, but they are worth someone making.
The software tries to find relation between various variables by trying out mathematical and non-mathematical operators. It is basically a type of symbolic calculations and takes ages. I don’t see how is it relevant to Natural selection.
If this Gould and Eldrege are persuaded still its a miracle eh.
In fact Sceptical zone is missing a equation here.
These two were actually early iD critics. I mean they realized things were wrong and didn’t work as evolutionists described them in textbooks.
So they had to fine tune it. PE.
ID/YEC simply insust fine tuning can not save evolution from its critics.
PE simply admited there was no intermediates and a new idea was needed.
So they said it was quick/sudden/not shown in fossils and then stabilirty.
PE is a retreat because evolutionary biology is in retreat.
Truly they were a first awareness things were not right.
ID people will be more famous in the future for going further in correction.
The more research there is on evolution the more it flops.
15 years i think is the most that remains for it.
It takes ages if you do the calculations by hand. Doing it on a computer speeds up the computation enormously. Give it a shot.
Robert Byers:
Don’t be so timid, Robert. Dembski says it will be over by 2016:
Aided no doubt by the millions of young people who have rallied behind Dembski.
I think you need to do that. It takes 53 minutes on an 8 core machine to spit out a model which on evaluation is far worse than Mathematica, and this with just 430 observations and 8 variables.
Now, coldcoffee, all you need to do is calculate FSCO/I and ID is complete!
eureqa won’t help. It will take more time than the entire available probability space to spit out a non-working equation. 🙂
Then we agree on that, FSCO/I cannot be calculated.
Can mathematica solve The double pendulum problem from experimental data alone?
http://creativemachines.cornell.edu/natural_laws
http://news.vanderbilt.edu/2011/10/robot-biologist/
Probably a combination of Syntheyes to get the data and modelling in mathematica could give the result, but I am not aware of any one who has done it yet.
Wolfram System Modeller can do most real world modelling
I don’t have enough expertise to comment on whether it can be done with some software, but certainly not with eureqa.
Short version: it can’t because of the solution class
Long version: try for yourself! I’m sure data sets are available.
Data available here
Will try.
Because it uses Natural selection to find the equations.
It’s awesome.
Well, that would certainly be assuming your conclusion!
Eureqa most certainly works using Darwinian mechanism. It’s precisely how it works.
You start with the minimum essentials – an “organism” – in this case a simple, randomly constructed equation – that reproduces by mating with another, and producing offspring with some characteristics of each, plus some random variation.
Some offspring equations predicts the dependent variable better than others, they themselves are more likely to mate and produce new offspring.
As generations go by, we get equations that are better and better at predicting the dependent variable – as accurately and simply as possible.
In other words, the population of equations adapts to thrive in an environment in which being able to predict Y accesses resources, and being able to do so simply accesses even more, just as living organism that can digest seeds will thrive, and if it can find efficiently it will thrive even more.
In other words it uses the exact method that Darwin proposed.
And it works.
eppur si muove.
Eric is simply wrong.
Perhaps then you’d care to provide pseudo-code for the calculation of FSCO/I?
You evidently think it’s possible to calculate, so…
10 FSCOIbits=0
20 IsItDesigned=Y
30 IF IsItDesigned=Y THEN FSCOIbits=a lot
40 Print “It is designed.”
The CSI version is:
Joe G, to his credit, writes to the Eureqa coders:
It would be good to know what question Joe asked, but either way, I most readily acknowledge that the Eureqa process is an analogue of the scientific process – it takes a model – the equation – and incrementally tweaks it so that it more and more closely fits the data.
ID scientists do not do this, of course, but most scientists do.
However, unlike scientists, whose “tweaks” are informed by intelligent reasoning, the eureqa “tweaks” are random, and the better the equation gets, the greater proportion of new variants will be deleterious.
So far from rebutting my point, Joe provides an opportunity for me to restate a different one I have made many times (as have others, notably petrushka) – that evolution is, in many senses, extremely “intelligent” – its process is very like the process by which scientists discover more about the world. The difference is that it does not have foresight – “tweaks” are random, not foresighted hunches. But interestingly, the results, though slower, can be as good, if not better!
The other sense, of course, in which eureqa more closely resembles the scientific process than evolution does, is that the “organisms” are equations (which most scientific models are) and what they have to do enhance their chance of breeding is fit the data well (which is what we ask scientific models to do). Eureqa, in other words, is an evolutionary method of “evolving” scientific models.
However it does so by the Darwinian strategy of random tweaks, rather than by the ID strategy of informed tweaks.
Thus, the equations evolve.
And they do so non-linearly.
Joe seems to think the Eureqa coders agreed with him. However, my read is that they sidestepped his question.
Lizzie, I realize that no one here likes Kariosfocus, but I give him credit for bringing to my attention the metaphor of islands of function.
Anyone wo writes a GA of any kind knows that you need a Goldilocks, just right sized mutation generating engine. One that doesn’t wander too far from the current tested configuration, but which wanders far enough to bridge local maxima.
You also need a functional space that is connectable.
That is why I look to Thornton et al and Lenski and others, to demonstrate that living things embody these characteristics.
If theists want to believe that chemistry is designed, it doesn’t bother me. It doesn’t affect what science is and does. The properties of matter support evolution, and it is simply stupid to deny it. Dembsky seems to have come to this conclusion in his search for a search. Without making the rather obvious inference that evolution works and is true.
Yes indeed, petrushka.
Although the point of this OP was (possibly badly) to illustrate the inherent non-linearity of the rate of adaptation.
On your main point, I agree entirely. Which is why I find it so odd when people mix “oh, but that’s microevolution” arguments with “Darwinian evolution can’t do anything” arguments.
Clearly, it can. How much is the more interesting question.
Well, he doesn’t say what his question was.
It would be more interesting if darwinist instead of explaining how an algoritm evolve explains why an RNA that replicates do not evolve but involve.
http://en.wikipedia.org/wiki/Spiegelman_Monster
Sorry, Blas, I think that lost something in translation. Will you explain what you think is the difference between “evolve” and “involve” and what you think the Monster has to do with either one or the other?