A recent post by Richard Sproat (here) provides an illustration (thx to David Pesetsky for bringing it to my attention). What Sproat describes is a typical case of protecting your brand. BS gets published and nothing that calls it for what it is gets a hearing. Why not? It will tarnish the Science brand. What would the journal be worth if it became clear that a good chunk of what it published was of little intellectual value? This is a rhetorical question, btw.
There has been a lot of hand wringing over fraud in science. I've talked about this some and evinced my skepticism about how serious this is, especially in cases where it appears that "fraudulent" results replicate (as in Hauser's case, to name an important recent "problem"). Read the Sproat piece and consider which is worse: the systematic suppression of criticism or scientific fraud? Which is systematized? Which more completely pollutes the data stream?
Not everything in our leading journals is BS.
I would bet that most is not. However, enough just might be to undermine our confidence in the seriousness of our leading publications. BS is very hard to combat, especially when our journals see themselves as part of the entertainment industry. Are Science and Nature becoming Variety with formulae? At least when discussing language issues, it looks like it might be.