I wondered if this blog post by Eric Holloway (@EricMH) would be of interest. Dr Holloway is a member here though he hasn’t commented since February, 2019. In his post
Dr Holloway claims to have falsified a claim about common descent on the Talk Origins website saying:
Talk Origin’s claim that the nested hierarchy of species is well attested by the data is highly questionable if not outright false, and should be retracted as evidence for evolution until such time as a much more rigorous analysis with DAG eliminating controls is established.
As TSZ is not a particularly busy site these days, I thought I would bring Dr Holloway’s potentially earth-shattering revelation to a wider audience. Also, I hope to hear from anyone with expertise in phylogenetics as to whether we need to start rewriting the text books.
Very odd. It’s not completely clear just what he did, what fake sequences he actually analyzed, or how he aligned or analyzed them. He doesn’t say. But my first point to make is that consistency index is not a measure of phylogenetic signal. I can’t say much more until I see how he actually produced his data matrices. His little figure is no help, though oddly enough his model appears to be a form of phylogeny that has some hierarchical structure to it.
Based on his brief description, he didn’t actually do a phylogenetic analysis. He just generated a tree (probably many trees, though he doesn’t say) and looked at its parsimony score. There is no suggestion that this tree is the most parsimonious, and no indication that he did any sort of test of support, which would require an actual analysis. CI, RI, etc. are not indices of hierarchical structure.
I can’t figure out, based on the description, what his taxa were or how he aligned the data set. It may be that his scheme does produce data with hierarchical structure. But that could be the result of the inheritance and branching in the model, even though the inheritance is of an odd sort, and the branching involves some reticulation.
That figure is exceptionally confusing. It looks like he’s generating characters de novo, passing them on to descendants, and then combines them with characters from other lineages while also introducing some degree of shuffling(it sorta kinda looks like what you get from sexual recombination). It would be false to say there is no genealogy going on in this process.
As DNA_Jock points out, there’s plenty of common descent going on in the scenario depicted in the figure.
What, in that figure, are the terminal taxa? How would he align the sequences, which it would seem are composed of short fragments that are either identical among taxa or unshared? How would you even get a CI from alignments of these sequences, which would seem to consist entirely of a combination of identical sites and gaps? Puzzling.
The other members in this discussion, most of whom have professional expertise in the relevant fields, are having difficulty seeing how he actually did that. Why don’t you help them out?
I’m a little confused by this as I’ve gained the impression that, as a YEC, you are absolutely convinced that Biblical truth trumps empirical experience. Which is why I wonder why Eric’s attempt at an empirical test that purports to undermine common descent matters.
I can clear up the confusion. Neither his analysis nor yours qualifies as empirical evidence. I have told you many times your computer simulations and sequencing and what not are not empirically based. All I care about is that he is able to manipulate the data against your paradigm. That makes my case that neither is empirically based.
I seem to have reached an impasse, as the data set he provided on TSZ is not a valid NEXUS format. For one thing, it has different numbers of characters for the various taxa.
This is how a phylogenetic tree comes out after obtaining an alignment from one of these DAGs(containing 30 “taxa”):
I just obtained a new DAG from the link he gives to his data generator tool.
I first sent the “taxa”(which comes in fasta format) from the DAG to a multiple sequence alignment tool (MAFFT) using default settings.
Btw the alignment consists mostly of gaps as the sequences from different taxa have wildly different lengths.
Regardless, none of this makes much sense because I’m still not sure how Holloway generates his data. It appears like his terminal taxa basically consists of anything output by his DAG generator, be they ancestors or descendants. I’m not sure.
I suspect that, to the extend there is any tree-like structure in the data, it’s because his DAG actually happened to implement a tree-like process, so if and when Holloway claims to measure any noteworthy CI, it’s because his DAG really has coincidentally implemented something like common descent.
