First let may say that this whole discussion sparked by the Beck/Cooper
debate has been *really* fascinating.
What I thought was ironic about the discussion was that in some ways, Kent
came off more as thinking like a designer than Cooper did. Specifically,
central to XP is that it's an iterative process. This is a large part of
what I learned in Architecture school; develop a design by evolving it in
iterative passes. This is in contrast to the traditional engineering
approach of "plan everything first, then build" (someone referred to the
Waterfall development model, which this fairly represents).
In general, it seemed to me that Kent came off as understanding interaction
design better than Cooper seemed to grasp XP.
And it seems to me that well-run XP could be a real asset to interaction
design. Sure, it would presumably be necessary in any event to have the
large blocks carved out early in the process, but perhaps there could be a
synergy in meta-level interaction design and object modeling. Agile
development would certainly make it easier to incorporate feedback from
user/usability testing. With competent interaction design involved at the
beginning and throughout the process, this could be a big step towards the
fabled Better Mousetrap.
In the end, I think the disagreement could be boiled down to this
Cooper - Interaction design needs to be done in advance of coding because
coding is clunky and inflexible.
Beck - No; coding doesn't need to be clunky and inflexible.
And despite the fact that Cooper claims that interaction design accepts XP
but XP doesn't accept interaction design, I came away with the feeling that
quite the opposite is true (at least in Cooper's version of IxD and Beck's
vision of XP).