A discussion of the role of empiricism and some recent results on the relevance of metrics in predicting defect (largely useless) and some new results on the power-law distributions and the apparent inevitability of the x log x nature of defect growth in a component.

Finally, defect clustering is also discussed along with some recent measurements.

TAIC2008-29-08-2008.pdf (1MB)

reference

Keynote at TAIC 2008, Windsor