> I am not experienced in either agile or Interaction Design. In this
> article Neilsen has pointed out that Agile software development in
> small increments could risk the overall user experience.

> A similar statement has been made by Larry Constantine in his report
> "The Usability Challenge" published recently in which he said "the
> absence of any comprehensive overview of the entire architecture" is a
> shortcoming of nearly all techniques that are based on iterative
> expansion and refinement in small increments.

I'm a web interaction designer working on a team with front- and
back-end software developers, two business owners and a scrum master.

The essential approach (for us anyway) is to be able to continually
deliver incremental value for our customers and visitors -- and, by
extension, our business. We work in two-week sprints, and at the end
of each we do a code release.

Short cycles don't mean throwing usability tenets or user testing out
the window, nor does it mean not having "any comprehensive overview of
the entire architecture". On the contrary, we designers are always
working several sprints ahead on the big stuff, at least one on the
"test and learn" enhancements, *and* we conduct bi-weekly user tests
on our latest and best interaction design candidates. In this way
we're sort of parallel tracking with developers, though we're always
at least a little bit behind them (which is a good thing).

If I'm to take Consantine's remarks cited above at face value, I'd
respond that most of our work is not designing brand new software
every time, but incrementally improving -- in every aspect -- what we
_do_ have. And when we are designing an absolutely brand-new area, we
still gather the necessary requirements up front, sketch out possible
solutions, and iterate through comps until a likely candidate emerges.

FWIW, I've never once worked on a project where every single one of
the usability / interaction problems have been solved -- or, in some
cases, even recognized -- by launch time. We're going to miss stuff,
we're going to misinterpret our user tests, we're going to make
sub-optimal choices. But with our current methods, we're continually
re-evaluating and only have to live with the truly sub-optimal for two
weeks at most.

I will say that one *huge* key to making everything work -- and I know
that this may sound hackneyed -- is just having a good working
relationship with one's teammates. If the division between software
developers and interaction designers as presented in the article by
Neilsen is the rule rather than the exception, then I understand why
people are skeptical. Trick is to get over it and work as a team. It's
amazing what can get accomplished once you get past the artificial "us
v them" mentality.

Peace,

Kurt Morris
Senior Interaction Designer
The Motley Fool