Larry Constantine summarizes a panel from CHI 2003 that I attended, in which the discussion revolved around just how much usability testing is enough; the summary is excellent, as was the discussion.
CHI 2003 Feature: Testing... 1 2 3 4 5 ... Testing... (Usability News) -- "The most impassioned advocacy came from... [WebWord]
By the end of the panel, as I recall, it was apparent that the panelists largely agreed: more testing, iteratively, is better.
Additional linkage on the same:
Constantine on the Magic Number 5 panel at CHI. At Usability News Larry Constantine gives a great rundown of the Magic Number 5 panel from CHI. The panel tackled the long accepted discount usability notion that 5 users will uncover 80% of the defects.
Usability testing seems to be the perceived gold standard for sites - one colleague called it the 'holy grail'. But as the panel showed, 5 users and the discount approach have some serious drawbacks.
I also find it pretty amusing that usability diehard Rolf Molich is suggesting a potential end for usability testing, while Cooper (who has long dismissed usability testing) now offers training and courses in same. [ia/ - information architecture news]