SP changes data. The intent is to make the data better in some way, such as more efficient to communicate and store, faster to display, or easier to interpret by a human observer. This raises the important issue of data quality and integrity: how does one validate SP algorithms for a particular application in the sense of quantitatively demonstrating the quality and utility of their output? These issues are particularly important in medicine and science where ``enhancement'' to one user might be degradation to another, or the ``noise'' removed for one user might be the cherished signal to another. It is critically important to develop acceptable and trustworthy protocols for validating the utility and quality of signals altered by SP and to develop effective means of predicting the utility and quality from easily made quantitative measurements. This can involve simulations of user practice, statistical analyses of differences in decisions or measurements made on processed versus unprocessed data, human perceptual models, subjective testing, and the development of improved quality metrics for signals that are seen and heard.