Christensen deliberately chose a provocative and somewhat misleading title for this short paper. His point is that testing should be taught throughout the curriculum, as an integrated part of the core knowledge of computer science, rather than as an isolated topic like search algorithms.
To this end, the author has built an advanced programming course around the principles of extreme programming, and in particular around the emphasis on high quality. Projects are assigned in which the scope (namely, number of functions programmed) is under the control of the students, while high quality is a mandatory requirement. To emphasize quality throughout the development process, an automated test tool is used that automatically runs tests whenever the code is compiled. This ensures that, as new functions are added, old ones do not break. This approach contrasts with traditional courses, in which student grades are based on achieving a set number of functions (namely, fixed scope), which results in a de-emphasis on quality, given the constraints on resources and time. The author claims positive anecdotal evidence for the validity of this approach.
I am personally aware of a leading university, known worldwide for software engineering, where students grades are based on completing high functionality software under tight schedule pressure, at the expense of quality. If even this institution resorts to such pressures instead of turning out a cadre of students immersed in proper software development techniques, one wonders if there is any real hope that this flawed culture, so pervasive throughout the software industry, will ever be overcome. The author of this paper is to be highly commended for trying.