I share Don Norman's view of tests. As a result I have consistently used performance assessments to determine how well students have met the objectives of my courses. (Unfortunately, I was compelled to give grades rather than P/NP and there were no opportunities for unlimited repetitions for those who did not do well.) Nevertheless, judging by course evaluations and later conversations with alumni, the efforts were effective, engaging, and efficient from the students' point of view.
Viability is another matter. In addition to the work involved in creating this type of learning situation, the effort required to provide comprehensive feedback on complex assignments for which there is no single correct outcome is daunting. My experience is limited to graduate courses in a business school and a medical school where class sizes never exceeded 30. With larger classes, I might have reverted to the traditional testing process as a matter of survival.
A related issue not addressed so far is the countervailing pressure on faculty to maintain an aggressive agenda of research and publication. Often, these activities yield more recognition and rewards than teaching accomplishments.