How Bugs Led Us Astray (Abstract of Invited Talk)

摘要

When the field of formal methods began, it had broad and noble goals. But somehow, over time, these goals were eclipsed by a more reductionist view. Nowadays, quality is measured by defect counts, and eliminating bugs has become the central focus of our field. In this talk, I'll explain how I think this came about, why it's insidious, and what we can do about it. My key observation will be that bugs are not the causes of problems but are instead symptoms. To improve our software-to make it more secure, safe and usable-we need to move from symptoms to diagnosis, to determine the underlying causes of poor software and fix those. I will argue that design is essential to achieving this, and that we need to reinvigorate design as a central activity in formal methods research and practice. I will give examples of designs, good and bad, drawn from my ongoing work on conceptual design of software.