Okay, I really don't blog much about science here. But in response to what I had thought of as a light-hearted, amusing post (albeit one that also sums up some very frustrating things about my new workplace), ambivalent academic posted a very interesting comment.
The authors are to be commended...if science is about figuring it out (rather than making it up) as we go along then it's important to include stuff that doesn't make any sense when communicating finds...then someone else might see how it fits or changes the working model and *presto* get it figured out. Unfortunately that only works if the stuff that doesn't make sense ends up in the publication so other people can see it...but it won't get published with stuff that the authors can't explain...I think that this is a major flaw in the way we report our findings.
And I started to respond to this in the comments, but then it got so long that I just decided to throw it out as a post. It's my blog, I can do that. And it's a very interesting point.
The data that's confusing, that doesn't fit a paper's hypothesis, usually isn't published. No suprise--why would any author include data that contradicts or confuses the story she/he is trying to tell? Negative results usually also aren't published. That transgeneic mouse with no phenotype? Will probably languish unknown. But if the experiements were rigorous and carefully controlled, then even puzzling and negative data is valid data. And when that data is not communicated, it can be to the detriment of the whole scientific community, as researchers waste time and money heading down blind ends . . .
But by convention, the scientific paper isn't a "data dump." By convention, it's a place to tell a clean, coherent, succint, and hopefully compelling scientific story. As with any story, extraneous and confusing details only (well, usually only) detract.
This isn't to say that puzzling data that dosn't fit the main hypothesis/story line is never published. I have included the odd pieces myself in a paper--sometimes this is necessary, as there may be a major experiment which *must* be done, and so you must report on it even when you don't quite understand all the results. But if you do have odd, confusing results, you better damn well try to explain or at least address it in your text, instead of just throwing it in there and hoping no one notices.
Or worst, not understanding that it's confusing in the first place (as seems to have happened in the manuscript I mentioned in the original post).
It is an odd business. Nature is messy, science is messy, but we try to tie it all up in a neat package for the journals, crafting a condensed, clear storyline out of months or years of frustration, failed experiments, trial and error, and sometimes entirely serendipitous discovery. And then we reframe the whole thing to make it appear as though blind luck was really brilliant foresight all along.
I agree that there needs to be a mechanism to more effectively communicate those puzzling results and negative data that don't get published in the peer-reviewed journals. I don’t know what the answer is. I suspect (hope?) the answer will have something to do with the growth of online scientific communities and of increased sharing of raw data in online databases. . .
The science philosopher is now off to sleep. . .