chris@grkermit.UUCP (Chris Hibbert) (01/04/84)
I attended the CHI '83 Conference in Boston during the week of December 12. The following is the first part of the trip report I'm writing for my boss. As I finish the other parts I will submit them to the net. Feel free to forward this message to other bulletin boards that might be interested. Monday the conference offered a selection of half- and full-day tutorials on design and evaluation methodologies for taking account of the Human Factors aspects of new computer systems. I attended a full-day session entitled "How to Design User-Computer Interfaces." This seminar was taught by James Foley of George Washington University, and consisted of a condensed version of a full-semester graduate level course he has been teaching for a few years. "How to Design User-Computer Interfaces" Seminar by James Foley James Foley's talk contained sections titled "A Methodology for Designing User Interfaces," and "Guidelines for Designing User Interfaces," as well as a "sampler of interaction techniques and devices" and a set of guidelines on "the visual presentation of information." His message could be summarized most simply as "consider alternative designs," which was the text of a slide he showed three or four times during his presentation. Foley's system is a formal methodology for moving from an analysis of the task to be automated, definition of the goals of the design process, through four levels of design of the language of interaction, all culminating in a design review. He stressed that the methodology should be considered an iterative process, allowing any level to be repeated if a succeeding level or design review showed that some of the goals weren't being met. The methodology was presented as a series of 7 steps which form the design process, to be followed by an unelaborated implementation step. The steps were 1) Task Analysis, 2) Conceptual Design, 3) Semantic Design, 4) Syntactic Design, 5) Lexical Design, 6) User Environment Design, and 7) Design Review. After doing a complete analysis of the job to be automated in the first step, the system is designed in increasing levels of detail in steps 2 through 5. The environment designed in step 6 includes the reference and user manuals, any "cheat sheets", and the physical surroundings of the user. The design review at the end of all this is the correct time to find weaknesses requiring iterations of some of the steps. When a satisfactory design review is completed, the implementation can begin. An interesting part of Foley's method is the insistence on separation of the design for the conceptual, semantic, syntactic, and lexical levels. He said that this separation had allowed them on one occasion to redesign a particular system from one physical implementation (keyboard specifications of commands, as I recall) to another (touch screen menus, I think) when the environment and available hardware changed. The fact that there were common concepts and semantics underlying both systems made it much simpler for the users to move from one system to the other. James Foley gave a small set of guidelines intended to help the designer of a user interface. As usual, these guidelines came with several caveats. The biggest one is that the guidelines are neither complete nor orthogonal. The most useful caveat suggests that guidelines are most useful if the designer takes the time to think about why the guideline was formulated, and then proceeds with an understanding of it, rather than attempting to rigorously follow something that was intended to be a somewhat hazy pointer to a good idea. Foley's guidelines were 1) Provide feedback, 2) Help user learn system, 3) Provide error correction and prevention, 4) Control response time, 5) Design for consistency, 6) Avoid memorization, and 7) avoid unnecessary spatial-linguistic conversions. Users of computer systems require feedback in order to be sure that the computer is "listening." Foley mentioned the analogy of human conversation: even if one person is talking all the time, the other person will maintain eye contact, and mutter "uh hunh" or "mmm" whenever an important point is made, or the talker will reiterate to make sure the point is understood. Feedback needs to be provided at several levels. At the Lexical level, the user needs to be reassured that individual keystrokes, mouse movements, etcetera, are noticed, or he or she will wonder if the computer is alive at all. At the Syntactic level, the user should be told that words and commands are well-formed, and if not, what part the computer didn't understand. At the semantic level, the computer systems should say that the command was understood, is being processed, and then it should respond with an indication that the command has been completed. Foley pointed out that a constant response time to similar commands is sometimes more important to user-comfort than a response which is usually fast, but occasionally delayed. Consistency is also important at all three levels. Lexical and syntactic consistency can lead to the building of good habits and "muscle memory" while semantic consistency makes a system easier to teach, learn, and understand. Some of the things Foley recommended to help minimize the amount of memorization (especially short-term) forced by a system, included accessing recognition memory rather than recall memory. (i.e. use menus or make defaults evident in your prompts.) Foley's point with respect to "spatial-linguistic conversions" was that it's much harder for people to convert from a graph to coordinates than merely to locate a position on a touch screen or with a mouse. Occasionally, the user is more likely to know the values desired, and in situations where that is likely to occur, the designer should provide the capability to specify positions numerically. The final part of Foley's talk was a video-tape and slide presentation of various techniques of interaction and ways of providing feedback to the user. This part of the presentation doesn't translate well to paper or videotext, so the following summary will have to do. Newswhole from the University of Toronto is a system for the markup of pages of newstext. It interactively facilitates the placement of ads, pictures, and text onto full-size mock-ups of newspaper pages. Spatial Data Management System from Computer Corporation of America is a spin off from MIT's Architecture machine group, and provides a spatial metaphor for a database system. It allows the user to peruse the entire database looking at individual entries without formulating textual queries. An excerpt from the Architecture Machine Group's "Put That There" video was shown. This video (shown in full elsewhere at CHI '83) shows someone 6-10 feet from a wall-sized map of the caribean, pointing (with his outstretched arm) and saying things like "put a green sailboat there" and "put a red freighter east of Cuba." The system responds to his voice and gesture commands by drawing an icon in an appropriate location. [Foley said he had heard rumors that the device that decoded gestures (it reports X-Y-Z position as well as pitch, roll and yaw) was to be used in a video game to be out sometime in early '84.]