a great article in which it discusses endemic problems in biomedical research. They start out by citing a pair of studies in which experts were unable to recreate the results of more than a handful of seminal cancer papers. The article goes on to blame a variety of factors that include the peer review process, the publish-or-perish culture, journal editorial policies, and poor understanding of statistics amongst others.
I have to say that, while I started the article convinced it would be baloney, I found myself agreeing with many of the problems they pointed out. There are, it seems, a whole host of structural problems with how science and engineering research are funded and vetted. Pretty sobering.
I'll be adding this to the "required reading" list for my graduate students. The article can be found here.