Genome Biology is carrying an editorial by Gregory Petsko arguing that Supplementary Data is a losing game. Web versions of scientific papers often contain links to additional tables, methods, and results for which there is not enough space in the print version. In my experience this is often where the goodies are hiding, like the information that the experiment must be done only with certain reagents, or that a whole other line of experiments gave negative results and have to be interpreted cautiously.
Petsko has, uh, a different opinion:
I hate supplementary material. It's one of the worst ideas in the history of bad ideas. It's the scientific publishing equivalent of fighting a land war in Asia.
He and I actually on the same page (pun intended) about what these massive data supplements do: by eliminating the word limit, it means reviewers can demand quite a lot more out of submitting researchers; then this mass is organized such that the tidy results and diagrams go "up front" and the wet science drifts into the back pages. And I think the "land war in Asia" analogy is correct: looking through the February 10th Cell, I see papers with 1,8,4,1 and 3 supplemental figures. What that means is a huge amount of extra work has become "published" (which affects its money and communication status for the corresponding author). There is no longer an upper limit to the amount of data which could be sucked in. And finally, especially with Cell, those files are inconveniently organized in nests of links and slow to download-- so they are at a qualitatively lower accessibility than the main data, which comes in a single pdf.
However, it used to be that these details didn't go into the paper at all, especially in Nature and Science. I agree with his objection that critical methods are frequently absent from the main text. But wasn't that always the case?