Hi, John! Another quicky comment on your blog without comments :-)The drawback is obvious -- why the heck are we generating these data, if we're not going to seriously test hypotheses about mechanisms?I think you misunderstood the idea behind this trend. The Plos ONE business model, now followed by many others, is simply to publish a lot more in a lot less discriminatory manner (and charge A LOT MORE for that, too). It makes sense from a money-making point of view. A larger, more important, view of this model is that it implicitely admits and explicitly tries to correct the problem everyone else is aware of: "subjective selectionism" is what I call it (in its extreme it manifests itself in all the problems with glamor mags; I can tell quite a few anecdotes about that part). So instead of trying to [mis]judge importance of the work, they are now lowering their pretense and only try to [mis]judge its technical quality. Of course even that modest goal is not truly feasible under traditional peer review and publication formats. Witness Plos ONE which ends up publishing a lot of utter BS. I see all this as a good thing because it puts the whole business into a slippery slope. Next it will be obvious that the single most important task of journals and peer reviews--credentialing--is completely bankrupt. So once we are all done slipping down, the end result will be "anything goes". We could publish our results anywhere! And the results will be judged by peers through citations and general recognition anyway and, eventually, by what they mean in a larger context (be it an important dataset or an ingenious interpretation of thereof). It's a part of a bigger trend made possible by Internet - elimination of middle men of all kinds (cf. music, news).
More cynical than me, probably, but I think you’re close to the mark. I’m more than halfway to “anything goes” already.
I wonder about archaeological site reports, which are fundamentally uninteresting and hard to get information out of to do meta-analysis. Some kind of self-published format with standard database would be enormously more useful for citation and further work. What I wonder is what kind of change in incentive structure could get people to actually do this kind of publication.