You say near the end that "the process of AHP itself ... is probably often more valuable than whatever quantitative statistics found" and this cannot be stressed enough -- although there is often merit in being able to say that "the blivit's priority is 43%" there is always great benefit in the analysis that leads up to this.

Indeed one of the potential problems faced during AHP analysis is the risk of small groups of rankers skewing the ratings -- you touch on this in the "averaging" section, and advocate weighting to mitigate the distortion, but I would stress to anyone employing AHP or similar processes that it is vital that the initial ratings are as reliable as possible, since skewed ratings can quickly destroy any quantitative benefit you draw from this, and indeed incorrect quantitative results may harm the qualitative/procedural benefit drawn from undertaking the analysis by supplying misleading conclusions.

All in all though a superb article, and definately one I shall refer others to.