In the news today is something that really ought not to surprise, but yet qualifies as "news." Apparently scientists, human as we all are, have been fudging their results.1 Sometimes to fit their instincts, sometimes to match past work, sometimes to please those who provide the funding. Yahoo News, apparently quoting a Washington Post article, is careful to call these falsifications "minor," and to say that they do not qualify as "outright misconduct." I, however, ask the question: how often did the modifications of the experiment design or results cause the overall result to change? How often was global warming predicted because contradictory data was thrown out? How often was it predicted by an experiment that "they knew would not give accurate results?" How often do we see 10 studies "proving" something because the authors published the same data twice? These lapses are admitted to by four to fifteen percent of scientists (depending on which exact offense you are looking at), which to me makes the problem seem fairly widespread for a process that we are supposed to trust implicitly.
