Dealing with reviewer comments

This is a recurring topic for most if not all actively publishing scientists, I wager. For all it's demonstrable benefits, the peer review system can feel like a crap shoot sometimes, wherein the outcome of the review process depends more on the identitis of the reviewers than on the quality - in the sense of impact factor - of the journal. Bluntly put, reviewer comments can be really annoying and unreasonable at times. Which is not to say that my reviewer comments don't come across as vexing and preposterous to the recipient authors, of course. Still; I have compiled a mental note of what I personally feel constitutes purely annoying and unreasonable from my own experiences, and I try to make sure that I never include these types of comments or demands whenever I review a manuscript, and the list is ever-growing.

Whenever your results contradict earlier findings of other authors, or alternatively if your proposed mechanism deviates from what is accepted, by which I mean published in peer-reviewed journals, the probability of encountering a variation on the following Catch-22 reviewer comment is high: "The data/mechanism in section XY of the manuscript contradicts earlier findings on similar systems reported by D. Ouchebag et al. Please provide either direct experimental evidence or supporting literature for these data/this claim."

Already published data has by definition more credibility because the work in question has passed through the rigors of peer review. Your manuscript currently under reviewer scrutiny has not, and thus it's particularly difficult to get something published if it conflicts with earlier reports, despite the fact that a) earlier reports might be erroneous due to substandard equipment or analytical techniques, b) although the vast majority of said published work could be top-notch, the particular section or even paragraph contradicting your findings could be substandard, and c) new data/mechanisms/etc. replacing old hypotheses is how science is supposed to progress, dammit. Of course, the aforementioned discussion holds true with the caveat that your research is valid, your data reproducible and your discussions/conclusions sound. Without this trifecta, you're just another windbag Don Quixote.

The problem with this particular reviewer comment then becomes obvious within the context that many molecular mechanisms and such can only be addressed indirectly, and obtaining direct evidence could be somewhere between problematic and impossible, often closer to the latter. If so, you're left with the conclusion that if something is on the interweb, it's true. Which is somewhat suspect.

So what does one do? Include more random paragraphs in manuscripts for purposes of later citing, thus beating the system but violating ethical codes? Make an entry in Wikipedia for your obscure topic and refer to that? Or simply live with the fact that publising results which are not in complete agreement with other work is much harder than it perhaps ought to be.

On a related note, I've noticed that on the occasions I get poor marks for language or "clarity of presentation", the reviewer writes his comments in the most epic broken English imaginable for someone with a PhD who supposedly writes and publishes peer-reviewed research.