An interesting point brought up by the ever-insightful becca in one of Drugmonkey's recent posts made me think. Why do we write discussion sections for our papers? No--I mean, why do we really write them? I'd imagine that a good number of you agree with becca, that if we're prioritizing reading sections of a paper, the discussion is almost always last or never. Nobody cares what you have to say about your own stuff; let ME decide the significance of your work.
But when it comes to writing and reviewing manuscripts, I find that there's quite a bit of emphasis on the discussion. "The authors did not discuss the implications of their findings in light of the work of Fancypants et al 1997, 2000, and 2010a & b," "The authors overstate the clinical relevance of the results from Experiment 2," "The authors' explanation for their findings in experiment 3 are unfounded." I myself have been that reviewer--the one who demands that the authors dedicate at least 150 words to speculation (with citations, of course) on why they failed to replicate Generally Accepted Phenomenon. But what do I care? Am I on the authors' thesis committee or something? Do I reject the data if the authors can't come up with an acceptable reason for why things turned out the way they did?
It seems to me that in general, the discussion section IS like a thesis defense: are you enough of a scholar about your own work to warrant publication of the work itself? Do you know the literature, have you thought about alternate interpretations, can you see where your work is taking the field? If the answer is "no," then the paper must be revised and resubmitted until it's deemed suitably "discussed." But honestly, does anyone besides the people who review your manuscript give a shit about these things? In other words, if a poor interpretation of one's own work is made, but nobody reads it because it was made in the discussion, does it make a sound?