SER Project Notes
Collaborative Design: Revisiting Reuse and the SER Model
relevant computational substrates:
Webnet project
analyze DODEs as a computational environment for
reuse
compare different object models: Smalltalk, C++,
Java (e.g., how successful is reuse in these environments; claim:
integrating code in Java is more efficient than in C++)
relevant theoretical dimensions and background
information:
- our OOD article ---> see: G. Fischer et al.,
Beyond Object-Oriented Development: Where Current Object-Oriented
Approaches Fall Short, Human-Computer Interaction, Vol. 10, No.
1, pp. 79-119
- LCM Model ---> see: G. Fischer, S.R. Henninger,
D.F. Redmiles": "Cognitive Tools for Locating and Comprehending
Software Objects for Reuse", Thirteenth International Conference
on Software Engineering (Austin, TX), pp 318-328,1991 (for detailed
work: Ph.D. thesis from Scott Henninger, David Redmiles, and Andreas
Girgensohn)
- Behavior Exchange (Alex Repenning)
- evolutionary design of complex systems
question: what is the right level of reuse?
component level is too low
reuse whole packages (e.g., in Visual Basic)
Eric Scharff: reuse in Java, GAMELON (??)
experience factory (Basili)
SER Model
- ---> see: G. Fischer et al., Beyond Object-Oriented
Development: Where Current Object-Oriented Approaches Fall Short,
Human-Computer Interaction, Vol. 10, No. 1, pp. 79-119
- analysis of success models for SER model: Unix,
Word
- an important requirement for a seed (Winograd,
p 103): " less is more. A minimal system made it easy for
a wide population of developers to experiment, to build their
own versions, and to discover possibilities that the original
designers never envisioned."
- driving forces behind the evolution:
- breakdowns, symmetry of ignorance, and the situatedness
of real world situation (LoD/EUM Figure)
- Winograd, p 287: design theorists (such as Brown
& Duguid 1992) describe how to design for learning and innovation,
rather than striving to create idiot-proof systems that stiffle
worker initiative
- open development - the SER model in the context
of the WWW:
- from a large effort by a few to a small effort
by a large number of people
- see Winograd, p 103: --> instead of relying
on a managed development project, the community lets a set of
standards emerge through a kind of bubbling up from the grass-roots;
in general, a standard is adopted because one or more groups have
used it to make something work - it is then modified and evolved
through a decentralized process in which anyone can propose changes,
invent extensions, and build new layers on top of it.
- Although the results are often chaotic and may
later need to be reengineered for robustness (or firmness), the
vitality of the unplanned development process leads to rapid innovation,
and the pragmatic "Take what works and build on it"
ethic of the community leads to rapid evolutionary selection.
Back to Top