Reactions: 22 Nov. 94

Making Usability

Lewis, Gould, Boies

Ok there are three parts to this paper. On the whole I think it's a good paper (I like ITS-like systems), but they (like me) make sweeping generalizations.

Status Report

• Fair enough. Usability has not had a high profile in industry - a friend and recent graduate was met with "Huh?" whenever the subject came up in job interviews.
• However, the claim that the process works and that it leads to systems that contain the right functions, are well liked, and are safe is a bit sweeping. Second: usability is more than this list of criteria. Third: other factors are also important: program size, cost and maintainability may be considered more important that usability.
• Iterative design is seen as too risky. This is a management problem, not a technical one. Of course, enlightened management should develop software the ITS way.

Usability Metrics

• The idea of setting quantifiable metrics for usability so that trends can be measured must be treated very carefully. Any benchmarker will tell you how difficult it is to create a good metric.
• As an example consider the task "document composition" - a good metric must be repeatable and consistent. So the document being composed must be the same - better make it a set of documents to smooth things out. The problem is that five years down the line the documents you are making are very different. Five years ago few people mixed text and graphics. Now it's common. Next they embed video... Yesterday's benchmarks suit yesterdays tasks.
• The fuzzy idea of "group productivity" is being helped along by the idea of business process re-engineering. Suddenly the group and what it does is the focus of attention and measurements are being taken. The idea of "group efficiency" is becoming central.
• The authors point a lot of fingers, but say very little about how to actually do things. i.e. IBM has evolved [usability goal values] over the years. How they evolved their metrics would be useful to know.

ITS

• I am all in favor of the ITS-type of system: clear abstract layers of functionality. Indeed, my european view of thinks of ITS as a powerful UIMS with a clear separation between UI and function.
• The ITS system is what the Madsen and Aiken CISP system may one day hope it grows up to be.
• Their worry that current prototyping tools lead to starting over once the current prototype is ready. First: the point of a prototype is so that you can throw it away after implementing as a crutch for understanding. Second: Prototypes (as discussed in this paper) are really mock-ups. Mock ups are rarely used except as UI specifications. Prototypes in the engineering sense also contain functionality and are an aid to understanding the tradeoffs in a particular solution.
• The quote Iterative design - so necessary to achieve good usability seems to sweep a bit broadly. Usability might be possible without iterative design, depending on how you define usability. Second: iterative design does not lead to usability.
• Some of this sort of work is currently being done so that the World Wide Web can share logical and presentation based markup without having the two step on each other's toes.