I've begun reading Tracing Genres through Organizations by Clay Spinuzzi.
I bought it because I think genre theory is potentially the
most-important-yet-least-appreciated topic in information architecture.
Clay approaches the issue from his background in rhetoric, and the
practice of technical communication. Still, he spends his first chapter
laying out a cogent and fairly persuasive critique of user-centered
design practice. The gist of it is this: the writings promoting
user-centered design theory and practice overwhelmingly cast the user
as a victim, subjected to the evils of a system over which they have no
control. By studying these victims, the heroic user-centered designer
can provide a far superior system that takes into account the actual
work practices of the users. Clay recognizes that: a) it's
condescending to treat users as victims unable to influence their work
situation, and b) UCD simply replaces one form of centralized control
Though I find elements of his arguments flawed, I think calling into
question the gospel of user-centered design is a necessary tonic.
The most interesting insights the chapter offers are:
a) an acknowledgment that users are often quite innovative in how they
overcome challenges in their local work environments, and are often
b) that UCD doesn't typically address the fundamental problem, which is
the monolithic nature of any designed system. Yes, it sucks when
systems are developed without any insight into user behavior, but
having a monolithic system designed according to the principles of UCD
sucks only marginally less, because such approaches inevitably don't
take into account the immense variety of small local innovations that
people develop to get their work done. There's an assumption within UCD
that one-size-fits-all; the methods (particularly the modeling) lead to
singular solutions that attempt to collapse variegated field research
into a simple set of requirements from which to build.
Another name for that approach is “lowest common denominator.”
This relates to the problem of cluster analysis
in how its output enforces a single view of content organization,
though there's plenty of evidence to suggest that different folks
utilize different approaches.