Literate Development is a proposed framework for both research into, and development of, interactive systems. It advocates explicit representation of the three key components of a 'user-centred' design:
I will then present a more simplified version, which I will immediately re-complicate with some possible extensions. The first preserves Phil Gray's idea of representing the context of development as a source of constraints on acceptable designs (and perhaps acceptable links as arguments for design decisions). The second extends literate development to cover usability evaluation (both inspection methods and testing). The third lets requirements (for those who believe in such things) be expressed as abstractions over the context of use and as constraints on the design. Lastly, links into implementation, as in Nick Graham's Clockworks system extend Literate Development into the belly of the development lifecycle.
All the extensions are handled as new forms of document within Literate Development 'Webs', but some introduce new forms of links such as simple correspondences (within implementations) and critiques (e.g., from user testing to design decisions). All of the extensions depend on Steven Clarke's use of 'parts' as an indirection mechanism for linking between different documents, records and specifications.
The proposed extensions may let Literate Development serve as a complete framework for research and development. For development, it allows complete records of the 'facts' of the context and the design to be related to each other via arguments on the appropriateness of design decisions. Design Rationales, by contrast, have not achieved such completeness of description. The linked information in a Development Web supports reflection on the use of contextual information during design, as well as providing a detailed context for later modification decisions (both within iterative development and for installed system upgrades).
For research, Literate Development has the potential to overcome many problems associated with HCI research. Development Webs support both the expression and testing of HCI theories, but force a completeness of description and precision that has been rare in HCI research to date. The talk will close with some musings on the relationship between 'good' representations and 'good' research in HCI.
For more information about this talk please contact cs0gco@isis.sunderland.ac.uk