A question that naturally arises in life testing is the following: “During the conduct of the test, what would you rather observe, a failure or a survival?” Most people answer this question by saying failure, because intuitively, failures are presumed to provide more information about the parameters of a failure model than survivals. The aim of this article is to point out that such intuition could be misleading and that the answer depends on the particular parameterization that is chosen. The argument is made through the use of Shannon's measure of information in an experiment, with the exponential as an example.