Can design faults be tolerated?

The short answer to the question posed by the title is “Yes”. A more cautious, and less simplistic, response would be that in certain circumstances, with appropriate provision of redundancy and allied supporting mechanisms, it is certainly possible to provide a measure of tolerance to faults of design. However, although this question may serve as an appropriate title, and starting point for discussion, it does not adequately address the significant issues concerning the application of fault tolerance techniques to deficiencies of design. As is usually the case, the first, and perhaps most important, step is to ask the right questions. In this paper, I propose to substitute five further questions in place of my title and, in answering those questions, will argue the case for the use of design fault tolerance in the development of reliable computing systems. In so doing I hope to justify the short and cautious answers already given above.