peterr@utcsrgv.UUCP (Peter Rowley) (08/24/83)
Thanks to Bob Cunningham for his review of the direct manipulation paper by Shneiderman in the August '83 IEEE Computer magazine. I looked it up today and found it enjoyable reading. It contains a persuasive argument for the more widespread use of direct (non-linguistic, essentially) user interfaces. The problems with direct manipulation are covered (p.64), though perhaps with not enough depth, as some concluding examples of direct manipulation systems seem to be poorly suited to their tasks. On page 67, Shneiderman states: "...a basic system [for bibliographic searching] could be built by first showing the user a wall of labeled catalog index drawers. A cursor in the shape of a human hand might be moved over to the section labeled ``Author Index'' and to the drawer labeled ``F-L.'' Depressing the button on the joystick or mouse would cause the drawer to open up and reveal an array of index cards with tabs offering a finer index. Moving the cursor finger and depressing the selection button would cause the actual index cards to appear. Depressing the button while holding a card would cause copying of the card into the user's notebook also represented on the screen. ..." Compare this with the (much simpler to program) interface which asks for a function (say, search by author) and a parameter (author's name). It might be fun to watch the graphics of the former system for the first few inquiries, but the novelty would quickly wear off and one would be left with the arduous task of locating the correct drawers and cards virtually manually. This appears to be a case of a concrete metaphor taken to an extreme. The latter system has a convenient concrete metaphor also, that of a helpful and efficient librarian. I think we should always be willing to consider the "intelligent agent" metaphor as opposed to a metaphor which involves inanimate objects. Admittedly, one could not ask such a librarian for an opinion on the book, but there are restrictions to all metaphors. Later on the same page, "Why not make airline reservations by showing the user a map and prompting for cursor motion to the departing and arriving cities? Then use a calendar to select the date, a clock to indicate the time, and the plane's seating plan (with diagonal lines across already reserved seats) to select a seat." I would have a hard time locating some of the cities I want to fly to on a map of the US, though I know their names. Often, I DON'T NEED to know their geographic locations; just that I have a meeting there next Thursday. Consider also the practical problem of differentiating between closely spaced cities or, worse, the multiple airports in NYC. And flipping through a graphic calendar to select a date? Moving hands on a clock to select the time? Again, consider the intelligent agent metaphor, this time a travel agent. The agent is given the names of the cities involved, possible dates, and possible times, and comes up with possible flights. A linguistic interface (such as one would use with a real travel agent) allows one to express this necessary uncertainty. But, again, the paper is well-worth reading and includes a lot of insight as to why videogames are popular and how their interfaces can be adapted to more mundane tasks. Thanks again to Mr. Cunningham. peter rowley, University of Toronto Department of C.S., Ontario Canada M5S 1A4 {cornell,watmath,ihnp4,floyd,allegra,ubc-vision,uw-beaver}!utcsrgv!peterr {cwruecmp,duke,linus,decvax,research}!utzoo!utcsrgv!peterr