This paper studies the informative content of controversial scientific arguments depending on the state of the debate. Researchers are assumed to differ in their degree of opportunism. The arguments considered are manipulable, may stem from sequential private experimentation and can be revealed selectively. Arguments of opportunistic researchers tend to be informative if there is a lack of consensus in the debate and uninformative if it is more settled. Arguments of more sincere researchers may be uninformative if there is a lack of consensus in the debate and informative if it is more settled. The effect of institutional incentives on the informativeness of arguments should depend on the state of the debate.