If software quality is a perception, how do we measure it?

For over twenty years, metrics have been being invented to measure software quality. And yet quantifying quality presupposes agreement on what constitutes quality. Quality has been portrayed as an absolute quantity, subject to objective measurements. We believe this effort has been misguided. We argue that quality, like beauty, is in the eye of the beholder - that is, that quality is not absolute, but depends on the perspective taken by the evaluator. As such, any direct measure of quality must necessarily be subjective, summarizing the impressions of some particular class of people who interact with the product. Indirect measures of quality are less objective than they may appear to be - beyond the arbitrariness of the choice of measure, and any difficulty in its interpretation, there is always the tenuous link of the metric to the perception of quality by any specific group. The need for this novel point of view is especially clearly illustrated by mathematical software.