New Paper Demonstrates Superiority of Design Model

(Steve Schaffner) #68

Obvious issue with the paper. . . From the Conclusion section:

Yes, that’s an obvious objection. His response is to remove from the paper any utility for comparing common descent with ID. What he has chosen to do is compare two models, neither of which corresponds to the expectations of biological evolution. Which of the two non-relevant models fits better might of interest to, well, I can’t think off-hand of anyone, but somebody, but it is not a test of the superiority of evolution or ID for explaining biological data.

If this were an actual scientific effort, an appropriate step would have been to simulate gene family evolution under a fairly realistic model of evolution – including missing data, since these genomes are neither perfectly sequenced nor perfectly annotated – and run the same test. I think it likely that the dependency model would be a better fit for that dataset than the simple tree model he’s been using.

Another obvious sanity check would be to take some of the gene families that are missing in scattered species and (a) do a BLAST search on those species, to see if the gene family is actually present but incorrectly annotated, and (b) see if there is sequencing coverage in the appropriate regions of the appropriate genomes.

These kinds of tests are essential. A study like this, which depends critically on the absence of data in public databases assembled from many different analytical approaches, can very easily reach completely incorrect conclusions if assumptions about data completeness and uniformity are not checked.

(Matthew Pevarnik) #69

I’ve been in lectures where there are some very well supported phylogenetic trees. What is your point? Anecdotal evidence means nothing. These are legitimate challenges to molecular studies and your reaction is just to mock them?

That’s an ironic statement coming from a proponent of the design model. I actually like that there is some kind of paper to discuss here… but again it’s not actually being published in a non-ID journal! If it actually describes reality (as quantum mechanics and plate tectonics do) then fight it out in the scientific community as a whole as @BoltzmannBrain keeps saying. Engage with those that actually do similar types of studies. It is certainly a bit silly to publish a paper that was not critiqued by other experts in the field and then to come in claiming its superiority!

(Cornelius_Hunter) #70

Here, here! There are all kinds of follow on work that come to mind, and you are pointing out one of them.


The fact that you can think up additional experiments to do doesn’t mean the work is not “scientific.” Demarcation arguments have a terrible track record, and nothing more than a fallacy.

This work demonstrates a different model that demolishes common descent. One can always add mechanisms (to any model), but that is terribly costly in terms of parsimony. Evolution / CD does this in spades, but all we’re doing it making it unfalsifiable.

Evolution says that the origin of the entire biological world can be accurately described (and indeed it is an undeniable fact) as one long spontaneous process. And of course, evolutionary thought has never been limited to the biological world. It applies to cosmology, geology, etc. This has never been motivated by, or supported by, the science.

So how is the claim that it is a fact that the entire biological world arose spontaneously good science? IOW, what scientific evidence and argument could possibly justify this claim? Do you not see that this is, at least a little bit of a tall order?

(Cornelius_Hunter) #71

You were pointing out that evolution / CD has a multitude of ad hoc mechanisms to draw from. I was agreeing–point being that evolutionists are typically unaware of the enormous violation of Occam’s Razor that evolution presents. This is why model selection is important and helpful. Ewert’s paper shows how badly CD fails on this test.

Just to be clear, I’m not actually a proponent. I’m a proponent of science, giving the empirical data a chance, and so forth. One way to do that is to compare models, which this paper does. If you’re only going to believe papers that are written in evolution journals, then OK, that’s your bias.

(Matthew Pevarnik) #72

Sigh. Cornelius- these are legitimate mechanisms that make modeling difficult. Guess what, they are present even if the model is ‘supernatural design!’

Typically when I see numbers like 2^10000th power or 10^3000th power, someone has done something wrong. Terribly wrong.

Also just real quick, what is the design model that Ewert is proposing? No common descent/supernatural creation of all species including each of the yeast for example we were chatting about earlier? I’m trying to follow his figure 4 describing null, tree, and dependency. The null model is one where no species are related, the tree is meant to represent common descent (though it looks like the arrows are pointing the wrong direction as I imagine they should point from the ancestral species of circles to extant species of squares), and the dependency model is what now? From the paper:

The dependency graph model is as similar as possible to the tree model. Instead of ancestral species, the model has modules. Instead of a single ancestor, each module may have multiple modules that it depends on. These are called dependencies. Each species is a top-level module: no other modules depend on a species module. Every gene family is introduced in a single module and inherited by all modules that depend on that module. Even if a designer were to design two different genes for the same purpose, we would expect them to end up being different enough to be classified into different families. A module may lose a gene that was inherited from a dependency. Structurally, both models are the same except for common descent having a single ancestor where the dependency graph has multiple dependencies.

What kind of sorcery is this? I’m gonna go with the huge numbers are nonsense.

Well that’s good. What is your model then?

I’m not so sure about that-the paper seems to be comparing pretend versions of various ideas, creating a magical model that performs better than how he decided the common descent model was supposed to be represented.

You casually dismissed a paper I linked earlier written in Nature. I’m not sure what you mean by ‘evolution journals’ either. A nice historical example is the steady state model of cosmology and the big bang model- both were published side by side in journals until one stopped describing reality as time went on. It is very revealing to me that proponents of the design model aren’t publishing their ideas side by side in journals in the broader scientific community.

(Curtis Henderson) #73

You’ve hit in a question I’ve asked Cornelius a dozen times or more over several months. I initially tried asking for evidence for his model, but gave up on that after multiple attempts and lowered the bar to simply asking for his model. I could never get him to offer either on his blog. Perhaps a different setting will give different results?

(Cornelius_Hunter) #74

The dependency graph model has genomes organised into modules that have dependencies on other modules. This is inspired by computer software. So you can have a “module” of genes that is used in various species. You can look at Figures 5 and 6 for an example of how a set of software is fitted into a traditional phylogenetic tree, versus a dependency graph.

By the way, folks should understand that Ewert’s work not only shows that the traditional tree model is wrong, it helps to explain why the evidence can be mistaken for a tree when the dependency graph is not considered. Dependency graphs can generally be force-fit into a tree model.

(Cornelius_Hunter) #75

So you don’t like the dependency graph model Curtis?

(GJDS) #76

My impression is that the ID approach makes their argument unnecessarily complicated. By this I mean they (a) insist they have an alternative to the current paradigm in biology, and instead of simply promoting this, they also (b) insist that all of the work carried out under that paradigm must be wrong.

I think if the ID approach concentrates on (a) and simply leaves (b) to itself, they may get a greater interest from the scientific community.

(Matthew Pevarnik) #77

Not quite tracking you here and my question of ‘what kind of sorcery is this’ still stands, even after looking at figures 5 and 6. Are you saying there’s a magic box of genes that various species can have and typically, species that are more closely related have greater access to this magic box than those further related as the ‘dependency model’ was made to closely model the tree model on purpose.

We should? Just by claiming over and over again the superiority of this model in your eyes doesn’t make it true.

(Curtis Henderson) #78

I don’t know, I haven’t looked at the paper. I’ve been asking you to answer a simple question for months. My apologies, but I have no inclination to play your game until you do.

(Curtis Henderson) #79

Possibly, I would at least like to see it attempted. I am already sympathetic enough to the idea that the incredible beauty and complexity of life is suggestive of a Creator, I would like to see that effort in action.

(Cornelius_Hunter) #80

Sorry for any lack of clarity. Dependency graphs are no secret. They are used in diagramming software. Try googling Dependency graphs and you should find some good documentation.

(Cornelius_Hunter) #81

Here, here. Agreed! I’m certainly not asking anyone to believe me. The data analytics are overwheming. Let me review the results.

Ewert’s three types of data are: (i) Sample computer software, (ii) simulated species data generated from evolutionary / common descent computer algorithms, and (iii) actual, real species data.

Ewert’s three models are: (i) A null model in which entails no relationships between
any species, (ii) an evolutionary / common descent model, and (iii) a dependency graph model.


First, for the sample computer software data, not surprisingly the null model performed poorly. Computer software is highly organized, and there are relationships between different computer programs, and how they draw from foundational software libraries. But comparing the common descent and dependency graph models, the latter performs far better at modeling the software “species.” In other words, the design and development of computer software is far better described and modeled by a dependency graph than by a common descent tree.

Second, for the simulated species data generated with a common descent algorithm, it is not surprising that the common descent model was far superior to the dependency graph. That would be true by definition, and serves to validate Ewert’s approach. Common descent is the best model for the data generated by a common descent process.

Third, for the actual, real species data, the dependency graph model is astronomically superior compared to the common descent model.

So where it counted, common descent failed compared to the dependency graph model. The other data types served as useful checks, but for the data that mattered—the actual, real, biological species data—the results were unambiguous.

Ewert amassed a total of nine massive genetic databases. In every single one, without exception, the dependency graph model surpassed common descent. Darwin could never have even dreamt of a test on such a massive scale. Darwin also could never have dreamt of the sheer magnitude of the failure of his theory. The results do not reveal two competitive models with one model edging out the other.

(Matthew Pevarnik) #82

Can you explain what the model is in the context of biological species? A casual glance at Wikipedia’s article on dependency graphs was not particularly enlightening.

The dependency model intentionally began very similar to the tree model and then adds a magic gene box that various species are dependent on? And no multiple species are actually related, they just happened to get a gene from one of the magic gene boxes? Are there any predictions of which species can access which magic gene box? Presumably not as there is no mechanism for how the magic gene box actually gives genes to species it just does so the model outperforms the tree model by a factor of 10^3000th?

Or perhaps you can clarify here a bit for me.

(Cornelius_Hunter) #83

Let me explain the results in a little more detail, because we’re not talking about a close call here.

For one of the data sets (HomoloGene), the dependency graph model was superior to common descent by a factor of 10,064. The comparison of the two models yielded a preference for the dependency graph model of greater than ten thousand.

So what is this ten thousand number?

Ewert used Bayesian model selection which compares the probability of the data set given the hypothetical models. In other words, given the model (dependency graph or common descent), what is the probability of this particular data set? Bayesian model selection compares the two models by dividing these two conditional probabilities. The so-called Bayes factor is the quotient yielded by this division.

The problem is that the common descent model is so incredibly inferior to the dependency graph model that the Bayes factor cannot be typed out. In other words, the probability of the data set given the dependency graph model, is so much greater than the probability of the data set given the common descent model, that we cannot type the quotient of their division.

Instead, Ewert reports the logarithm of the number. Unbelievably, the 10,064 value is the logarithm (base value of 2) of the quotient! In other words, the probability of the data on the dependency graph model is so much greater than that given the common descent model, we need logarithms even to type it out. If you tried to type out the plain number, you would have to type a 1 followed by more than 3,000 zeros. That’s the ratio of how probable the data are on these two models.

This is found using standard methods and practices. No sorcery here.

(Cornelius_Hunter) #84

Oh, gosh yes. Let’s say you have the dependency graph for a set of species, and then you get a new genome, for a species in the graph. It will give you all kinds of predictions of the organism’s genotype. This, of course, is basically the same thing an evolutionist would have said, if asked this question about the value of phylogenies. Of course the evolutionist would have been wrong a lot. The paper does have some examples as well. I think applications of the dependency graph model will be very interesting.

(Cornelius_Hunter) #85

Now about the Bayes factor. By using a base value of 2 in the logarithm we express the Bayes factor in bits. 10,064 bits is far, far from the range in which one might actually consider the lesser model. See, for example, the Bayes factor Wikipedia page, which explains that a Bayes factor of 3.3 bits provides “substantial” evidence for a model, 5.0 bits provides “strong” evidence, and 6.6 bits provides “decisive” evidence.

So 6.6 bits is considered to provide “decisive” evidence, and when the dependency graph model case is compared to comment descent case, we get 10,064 bits.

But the Bayes factor of 10,064 bits for the HomoloGene data set is the very best case for common descent. For the other eight data sets, the Bayes factors range from 40,967 to 515,450.

In other words, while 6.6 bits would be considered to provide “decisive” evidence for the dependency graph model, the actual, real, biological data provide Bayes factors of 10,064 on up to 515,450.

We have known for a long time that common descent has failed hard. In Ewert’s new paper, we now have detailed, quantitative results demonstrating this. And Ewert provides a new model, with a far superior fit to the data.

(GJDS) #86

I am intrigued - any ideas on the effort in action?

(A.M. Wolfe) #87

Hello Cornelius,

Do you have a peer-reviewed article to cite that says the entire biological world arose spontaneously? Or are you speaking not of evolution but of evolution_ism_, that scientific positivist worldview trumpeted by antitheists like Richard Dawkins in their popular (philosophical, non-scientific) works?

Speaking of fallacies, I would submit you’re attacking a strawman here.

Science in general speaks of proximate causes, not ultimate causes. It cannot therefore speak to an ultimate sort of randomness or spontaneity. There are some great resources here on the BioLogos site to help explain different kinds of randomness if you’d like to take a look. Scientists do not speak of ultimate, ontological, theological randomness… unless they happen to be a particular strand of antitheist scientists with an axe to grind, speaking outside of scientific journals.