Who to trust? That is the question when you don't know very much (all of us). Trust is precious, and to some extent sacred. That's why I can flip out when I realize after the fact that someone more informed than me in field X sampled biased their argument in a way they knew was shady to support a proposition they were forwarding. What's the point of that? Who cares if you win at a particular bull-session? You're burning through cultural capital. And not that most of my interlocutors care, but I'm likely to never trust them again on anything. In any case, this came to mind when I ran across a James Fallows' post at The Atlantic. Here's a screenshot of the appropriate section, with my underlines:
The PNAS link is wrong. The correspondent is actually linking to an article in Quaternary International. And they do point out that there are possible problems with draft quality sequences due to contamination. But I didn't find the paper too persuasive. There are two issues. First, the Denisova genome is very good quality. So you can be more confident about those results. And those results themselves should increase your probability for the validity of Neandertal admixture; Denisova admixture was much more unexpected and surprising. In contrast, Neandertal admixture is positively banal. Additionally, one of the citations that they list prominently as supporting problems with poor quality and D-statistics
actually is from a group which strongly supports Neandertal admixture.
This doesn't mean that the group has absolute control over how their papers are cited/interpreted, but I think it's rather misleading to cite a paper in this way, because it gives the sense that the authors of the cited paper are skeptical of admixture, when they are not. Second, I'm rather sure that Fallows' correspondent was confusing the above paper with the one from PNAS a few months ago which Imentioned in large part because friends and acquaintances were telling me that it wasn't a very good paper. Since the correspondent doesn't mention the substance of this paper I suspect they weren't familiar with the details. Rather, they simply interpolated the name of a more widely known journal, PNAS, for the real journal, Quaternary International, while recalling the content of the latter's paper. The moral here is troubling, but it is worth pondering for all of us. James Fallows had no idea about this field, so he simply relayed a very confused correspondent. Second, the correspondent, an admitted archaeologist, probably has no way to evaluate the plausibility of papers in human genomics. So they forgot the content of the more prestigious article in PNAS from a more well known group, for a paper in Quaternary International, from Fudan University in Shanghai. I assume this is not a well known group. The archaeologist who contacted Fallows probably meant well, but they shouldn't be using their academic status as a scholar in archaeology to pass along muddled opinions about a field which they're not competent to evaluate. As for Fallows, I really do hope he knows that the 5% value he's quoting is a genetic inference, and not an archaeological one, so that an archaeologist shouldn't sway him too much either way (I think he should remain skeptical, though if he got the results from Geno 2.0, I know that that SNP-chip has a lot of Neandertal markers).