Joseph Wilkicki's Notes on Riven Brooks' "Comparative Task Analysis: An Alternative Direction for Human-Computer Interaction Science"

Before getting to the questions I had about this article, a brief description is probably in order. Brooks' article is a discussion of what guidelines Human-Computer Interaction Science should provide to designers of software programs and what kind of descriptive language is best used in HCI to describe these guidelines. Brooks provides a sample of what a useful analysis could be. This analysis has the properties of being "based on the use of systematically recorded knowledge about human behavior....presented in a framework and at a level of analysis that is useful for evaluating the impact of the major design decisions on user behavior... [and it is] highly specific to activity F." (pp 50-51).

Brooks then describes why HCI specialists seem to avoid developing these kind of descriptions. One of these reasons is that "the advice given [in the sample response] was not based on a model of problem-solving behavior, at least not in the "technical theory" sense used by Newell and Card(1985)."(p 51). This leads to a discussion of whether descriptions should arise from a broader science to an application (such as physics concepts applied to engineering) or if descriptions arise independent of a broader science. Brooks seems to favor the latter. Brooks then moves into describing how description is currently done in HCI and the problems these methods have.

Now on to the questions I have. These occurred to me during my reading, so if they seem slightly off the mark, any comments or thoughts on corrections would be helpful.

1. First, is the analysis provided on the first page useful for the various group projects? Or is the educational goal of instruction a different scenario then the problem posed here?

2.Would suggesting guidelines for the design of our software and methodologies for evaluating how good they are be more useful, then this analysis? Equally important?

3. Are the guidelines for a good analysis useful?

4. Brooks states that two problems that HCI specialists have creating his idea of the good analysis are: the observations are not based on "a model of problem-solving behavior" and that HCI specialists are not used to guiding a design, but developing "tools and methodologies for interface design". The second problem might not be relevant here since we are all developing designs, but has anybody felt that they were not sure about what observations to make, or which weren't significant? Did you feel the need for a model of problem-solving behavior?

5. Gardner's theories of multiple intelligences might be regarded as abstractions of how people learn and understand. Do these seem to grow out of Newell and Card's ideas that technical theories arise from specialization from a broader science, or Brooks' idea that analyses are generated independently from the theoretical science?

- Is Gardner's theory useful as a guideline and what would that suggest about the Newman/Card idea and the Brooks' idea.

6. Brooks theory seems to overlap Newell and Card's when he says that "discarding irrelevant details while isolating and emphasizing those properties of artifacts and situations that are most significant for design [is necessary]"(p. 53). Compare with Newell and Card: "Purely descriptive studies...can make important contributions to HCI, if they describe something novel or describe a previously unnoticed aspect of user behavior or serve as a major confirmation of other descriptive studies." These two ideas seem similar to me, which seems to contradict Brooks assertion that Newell and Card's ideas are different and not completely (applicable, useful?) to HCI .

7. "HCI most closely resembles the class of engineering disciplines, such as civil engineering...in which most requirements for design arise out of naturally occurring entities." Are human-computer interactions natural, or, as brought up in the discussion of interfaces, are computers forcing human-computer interactions into an artificial way of doing things?

8. One problem outlined with task analysis is that even if we have a breakdown of two tasks, how can we relate them? (pg 56). Should we? Each of our projects has the goal of forwarding education, some by teaching, some by reinforcing. The systems that teach each seem to approach this goal slightly differently (games, more informative), as do the reinforcing programs. Can we compare the different tasks in each type of program or compare each type? Are there common parts between them?

9. On a side note (which might not deserve a question number) any thoughts on Brooks belief that comparisons of say, X-Windows to MacOS are only possible by line-by-line analysis, or does the idea of WIMPs cover this adequately, in terms of the article or just in general?

10. "Work on comparative description of interfaces is, in part, driven by the needs of software construction and is therefore likely to progress, although, perhaps not in the directions serviceable to HCI." (p. 57). This seems to suggest that HCI is not akin to civil engineering, for example, because interfaces are developing independently of it, just as civil engineering is developing independently of physics. Is this perhaps a contradiction of Brooks ideas, or suggesting that maybe HCI is too theoretical for the practical purpose of software engineering?