jfarrell@sun.UUCP (Jerry Farrell) (11/29/83)
Laura has discovered the reason why [at least] I usually "n" my way through most of the net.cog-eng discussions: they are often predicated on hard- ware assumptions that are a) old b) needlessly stifling. In particular, discussions of the time it takes to get from one menu context to another are evaluated much differently when you are talking about bit-blitting a new patch to the screen & inverting the item under the mouse, etc. Curiously, however, many of the same issues are controversial in the new context; you just can't carry your old intuitions over. For instance, erox' Star took a lot of flak when first introduced because features intentionally included in the user interface to make it easy to use (for novices and casual users) got in the way of more experienced users, especially by slowing them down. Lisa seems to have suffered some of the same fate. o question of long names here, but definitely one of trading user processing for machine cycles in order to find an optimum mix for the user task. In a similar vein, long names can be a problem in fancy bit-mapped environments; but the critical resource is display acreage, not keystrokes. I have no moral to draw. I do have a request: what studiies I have seen that bear on these issues are very fragmentary and spottily applicable. or instance, I have heard that visual feedback which follows within about 10 ms of a muscle action is likely to be associated with that action, but that longer delays break the association. This has obvious implications for design criteria for mouse- or stylus- selection schemes (picking, for you CORE afficionados) -- any scheme too baroque to give feedback within a hundreth of a second isn't worth it. But that's a pretty isolated result (assuming I've even remembered it accurately). What other data are there that would help a user interface designer choose a reasonable compromise between user ease & machine respoonsiveness? What are the cognitive costs of delayed response from processors? are there thresh- holds or other non-linearities? Is there a plausible categorization of tasks according to how tolerant they are of delays? Bibliographies & other leads gratefully accepted. [Errata: several places where a fat finger got the Control key instead of Shift: Read "Xerox' Star" in the 3rd line of the 2nd par, "No question" 4 lines below that, and "For instance" in the 3rd line of the 4th par. Sorry.] jf