Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: version B 2.10.1 6/24/83; site decwrl.UUCP Path: utzoo!linus!decvax!decwrl!daemon From: daemon@decwrl.UUCP Newsgroups: net.cog-eng Subject: Re: expert-friendly: are long names a waste of time? Message-ID: <4081@decwrl.UUCP> Date: Mon, 14-Nov-83 12:09:51 EST Article-I.D.: decwrl.4081 Posted: Mon Nov 14 12:09:51 1983 Date-Received: Tue, 15-Nov-83 19:21:44 EST Sender: daemon@decwrl.UUCP Organization: DEC Western Research Lab, Los Altos, CA Lines: 75 From: Ed Featherston HL01-1/P06 225-5241 Begin Forwarded Message: ------------------------------------------- Newsgroup : net.cog-eng >From : KRYPTN::TS1::BURROWS Organization : Digital Equipment Corp. Subject: Re: expert-friendly: are long names a waste of time? ------------- From thomas@utah-gr.UUCP (Spencer W. Thomas) Wed Dec 31 19:00:00 1969 There was an article in a recent CACM which 'demonstrated' that it didn't matter *what* you called the commands in an editor - the command names could be totally meaningless - when people learned any set of commands they were equally proficient. Actually, that's not what they demonstrated, although it's not hard to see why one might get that impression. (In fact, a friend who reviewed the article warned one of the authors that it might be so construed. Ah, well...) Their actual conclusions were: 1) Novices and system designers use different language to describe text editing. 2) That people in general describe the operations differently, and inconsistantly. 3) That guessing what is "natural" or "common" rather than observing and analysing real users is unlikely work 4) That although familiar words are easy to learn as command names, there is a danger that the imprecision of a number of the most common words will cause them to be misused. 5) That a reasonable compromise might be to use words that were only moderately common, so that they are recognisable, and thus more easily remembered than totally unfamiliar words, but don't suggest so many different and possibly conflicting means as to lead to misuse. 6) That it is more important that the commands be unambiguous and precise than familiar and memorable. 7) That 'rational design of commands and command names for usability requires deeper understanding than is captured in the slogan "make the language natural."' What leads to the misinterpretation is that they point out that they have no direct evidence that the approach of taking moderately familiar words which they suggest in #5 above will work. It is merely a conjecture they have drawn. The confusing passage is: "We have no direct evidence of this conjecture. In fact, the results of Experiment 2 give little support to the belief that word choice is of much importance at all. However, as stated previously, we are reluctant to conclude that word choice would not be influential in larger command sets used for longer periods." Given that subjects in the study only learned the Unix (tm) ED editor's APPEND, DELETE, and SUBSTITUTE commands in addition to the command to start an exercise (START n), and how to abort the entry of a line (with @), and terminate input for an insert (with .) virtually *ANY EDITOR* would have a significantly larger command set! So that warning at the end is very important. Also, note that although they say their experiment gives little support to the notion, they do not say that it disproves it. It is very important in logic, math, and research to remember that there is a big difference between failing to prove something and disproving it. So, although it may not matter whether your editor's entire command set consists of DELETE, APPEND and SUBSTITUTE; or OMIT, ADD and CHANGE; or ALLEGE, CIPHER and DELIBERATE; if you use have a larger command set it may make a difference what you chose as command names. Also, they did find that command syntax could be important. Mail address : ...decvax!decwrl!rhea!kryptn!ts1!burrows ------------------------------------------- End Forwarded Message