Modern software development processes openly acknowledge what software
designers have always whispered: software design is not done as a
monolithic phase between analysis and implementation. Rather, the
process of designing software is incremental, experimental, often
concurrent, and interleaved with other activities. It is a game, in
the sense that it involves strategising, exploring options, balancing
objectives - and in the sense that it can be fun to do in groups!

At the same time the market for software design tools has expanded and
become more competitive, largely because of the emergence of the
Unified Modelling Language as dominant design notation. Yet these two
thematic changes in the community's view of software design, embodied
in methodologies and tools respectively, do not yet seem to be
harmonious. The current generation of software design tools supports
the recording and verification of design better than it supports
design itself. The tools are neither players in the design game, nor
reliable referees.

Is this a problem, and does it have anything to do with coordination?
I will argue that the answers to both questions are Yes, and will
support what I say with particular reference to designing frameworks
and product-line architectures. The talk reflects work in progress.