REC
Posts: 638
Joined: Sep. 2006
|
Theobald's second reply is worth fully quoting, and preserving here.
Hi George,
Thanks for the admission of error and the promised retraction -- that's admirable.
There is indeed (extensive) supplementary material for the paper, and it is linked to in several places, most notably at the bottom of the full text (http://www.nature.com/nature/journal/v465/n7295/full/nature09014.html#/).
Here is the direct link:
http://www.nature.com/nature....-s1.pdf
You wrote:
"I didn't mean every criteria used in the paper hinges on sequence alignments. I merely meant that without them you'd have no paper. Your statement that 'Here I report tests of the theory of UCA using model selection theory, without assuming that sequence similarity indicates a genealogical relationship' would likely lead an objective reader to think that your analysis somehow confirms CD without using sequence alignments as evidence for genealogical relationship, which would be a false impression. I'm not sure I could think of a more misleading way to put it."
There's a basic confusion here about my methodology. I obviously do use sequence alignments, but the analysis does not assume that there is a single sequence alignment for all life. Similarly, I use phylogenetic methods, but my analysis does not assume that there is a single phylogeny for all life. What do you have against sequence alignments anyway? Even the most radical YEC accepts them, for instance, for aligning homologous proteins like human polymorphisms, or, say, for aligning the homologous cytochrome c proteins of bacteria. The key point to understand is that my multiple ancestry models used a *different* sequence alignment for each set of taxa hypothesized to share common ancestry. Then the model selection tests decide which hypothesis is best. It is possible, for instance, that the model selection criteria could have chosen the hypothesis in which human proteins were in their own alignment, separate from the rest of life. But they didn't.
You then wrote:
"You [Douglas] wrote in the paper:
'by statistical convention a score difference of 5 or greater is viewed as very strong empirical evidence for the hypothesis with the better score (in this work higher scores are better).'
That is, of course, false. One hypothesis comparing well against others does not translate into 'very strong empirical evidence for the hypothesis.'"
Here I actually have no idea what you mean. It is quite true that, by statistical convention, a score difference of 5 or greater is viewed as very strong empirical evidence for the hypothesis with the better score. This is a statement I firmly stand by, and no scientist or statistician will find it controversial. Now perhaps you just don't like statistics. If so, then here the argument stops, because of course my analysis is fundamentally based on modern, cutting edge statistical methods. Pretty much all of science, however, believes that statistics is a necessary component of the scientific method -- it's really the only way to deal with uncertainty and noise in data.
You then write:
"This is yet another example of the rationalism that underwrites evolutionary thought and has infected science."
I'ts hard to imagine what you could mean by this, but you appear to think that science is wrong for being rational. But science is clearly fundamentally, and unapologetically, based on using reason. Again, it appears the argument ends here, as they say: you can't reason a man out of a position he didn't reason himself into.
Cheers,
Douglas
|
