Pleasant@Rutgers.arpa (01/24/83)
HUMAN-NETS Digest Monday, 24 Jan 1983 Volume 6 : Issue 5 Today's Topics: Computers and People - Computer Sex & User Interfaces (2 msgs), Programming - Unix (3 msgs) ---------------------------------------------------------------------- Date: 10 January 1983 11:10-EST From: Stephen C. Hill <STEVEH @ MIT-MC> Subject: computer sex The question that arises about your computer-controlled sex-doll is whether it would be listed in the new products directory under hardware, software or firmware! ------------------------------ Date: 6 January 1983 23:10-EST From: Robert Elton Maas <REM @ MIT-MC> Subject: Use of keyboard characters and EFFICIENT user interfaces Lockheed's DIALOG (document citation-retrieval system plus a little database-fact retrieval) used a curious and possibly good command structure. They use the top row shifted , i.e. !"#$%&'(), for commands. This could be done because they had only a few basic commands (login, logout, select-database, form-sets-by-keywords, boolean operations on sets, type on terminal, list offline and send by US postal service). They supplied a stick-on list of commands that could be placed just above the top row for convenient reference. This method combined standard terminals (you don't have to buy a special terminal just for DIALOG, you don't have to reach your fingers far away from the home position, and you can even switch the stick-on if you use more than one system and need to change the command legends when you switch systems) with special-function-key terminals (the keys are labeled as to actual function via the stick-on rather than being control-funny-letter that you have to remember). HP calculators have another way of combining the advantages of function keys and control-mumble, although they have too many little teensy keys to start with for touch typing. They have three functions on each key, printed on the face, top-edge, and bottom-edge (or is it four?, one on the bottom face-edge?). Anyway, you first press a control key to get the second and third (etc.) functions for a given key. The control keys and the extra functions they enable are color-coded to make it easy to remember which control key you have to press for a given function you see on the legend. Perhaps on directly-wired keyboards (where control-# etc. is meaningful, which isn't true on 7-bit communication terminals) the most often used functions could be tied to ctrl-shift-number and/or ctrl-bottomrow-letter. Then a legend could be taped next to the keyboard for new users, but commands would still be short (for all users). For example, using orthogonal methods, ctrl-shift-number could select mode (character, word, line, sentence, paragraph, s-expression, page, etc.), which would be sticky over multiple depressions of a lower-row command, and the lower row could activate the command (delete-forward, delete-backward, skip-forward, skip-backward, transpose; I think that's all, providing all normal text was auto-insert like EMACS/MINCE, which I think is a big win). Or for easier touch typing (numbers and lower row letters are hard even for expert typists when done to excess), the home row and upper-letter row could be used, with the stick-on legend a little further from they keys so referring to the legend is a bit slower but still a direct follow-the-line action that any novice user can learn in 5 seconds. This would have the advantage that communication terminals could use this method too, since ctrl-QWERTYUIOPASDFGHJKL are all transmittable. With a sticky ctrl- prefix (no terminal mod needed, do it in the host, just like ESC is the meta- prefix and ctrl-uparrow is the ctrl- prefix in EMACS, except that it'd be sticky until undone under my proposal. The mode line would of course tell the user whether sticky ctrl- is in effect at any moment. ------------------------------ Date: 5 Jan 1983 2139-PST From: GRANGER.RS%UCI@UDEL-RELAY Subject: A General Approach to User-System Interactivity The following is excerpts from a memo I wrote in consultation with the people running the Times-Mirror Videotex Test in 1981 (this was a 200-family test of videotex using the Telidon software by Infomart of Toronto). Some of the things Ray Clark is saying are so important for system-design configurations I was prompted to dig through the files and pull this out. I offer it for your perusal and response. This memo deals with some considerations which probably ought to be on all of our minds with respect to the way users -- be they information-needers or information-providers or even our own staff -- will relate to our system (as represented, most concretely, by their keyboards and video-screens). There are 2 types of users: "inexperienced" users, who are highly prone to certain negative emotional states commonly known as "anxiety," "fear," "panic," or "low self-image;" and "experienced" users, who are prone to negative emotional states called "boredom," "anger," "frustration," and "disgust." In the former case, the emotional states named are commonly brought about by a system which (a) overestimates the user's information base, and/or (b) overestimates his/her understanding of automated electronic information media or interactive teledactylography, and/or (c) does not include adequate, appropriate, timely or intelligible documentation (either on- or off-line), and/or (d) presents itself with too much of a mystique, too much jargon, or too much other stuff not within the user's ordinary realm of familiarity. In the latter case, the emotional states named are typically triggered by a system which (a) underestimates the user's information base and/or general understanding of the medium, and/or (b) over-documents by forcing the user to wade through too much display verbiage explanatory of stuff he/she already knows, thus delaying his/her arrival at the desired display objective, and/or (c) presents itself stupidly, i.e., does unnecessary things and/or forces the user to do unnecessary things in order to get what he/she wants, and/or (d) is slow in responding to the user's input. One objective of any system ought to be to avoid producing negative emotional states in its users. In order to meet this objective, we will have to tailor the interactive aspects of the system to the needs of BOTH types of user described above. This, of course, is a kind of a tightrope-walk; but, lest we forget, that's what computers are for: they are the best logical tightrope-walkers thus far known to man, precisely because of their virtually unlimited flexibility. Note, by the way, that any given user may at any given time fall into EITHER the experienced or inexperienced category. I, for example have now gained a rather extensive experience in the use of Compuserve's electronic mail and messaging system, and so I get very bored and start to drum my fingers and mumble nasty things as I am continually forced to plod through 6 levels of menuing just to effect some near-trivial function like altering an address field. On the other hand, I am just now getting familiar with Compuserve's file-handling and editing subsystem, FILGE, and so I feel comforted by the very tight hand-holding I am getting the first few times I try to deal with it: unfortunately, I know there will be no way to turn it off when I no longer need it! Further, users tend to move very rapidly and abruptly between the two categories, and bidirectionally. The problem reduces to one of coming up with an interactive design which makes all meta-information -- information about information -- on the system highly accessible to, but not unwantedly forced upon, the user. There are, of course, clear software solutions to this problem. We just have to not forget to implement them! Editor's note: This is why BOTH ctrl/h and delete should be recognized as backspace!!!! Always and forever! It should be a universal standard! Happy New Year to everyone... Rob Segelbaum (GRANGER.RS%UCI@RAND-RELAY) P..S.: Anyone in need of knowing what Times-Mirror, videotex, and/or teledactylography are, just buzz me. It is not really necessary that you know what they are to get the drift of the argument here, but if anyone is does not know and is curious, just ask! ------------------------------ Date: 20 Jan 1983 1547-EST From: Tony <Li@RUTGERS> Subject: TOPS-20 vs VMS vs UNIX To continue the discussion about different OSes. OSes are like cars. There is no perfect OS. Each has its own particular problems and advantages. You can't ask 'What is the best OS?' or 'Is VMS better than Unix?' What do you mean by 'best' or 'better'? To put it mathematically, what is your metric? Asking for the 'better' OS is similar to asking if a Porsche is better than a Cadillac. If you want to rank OSes, you have to consider the environment around the computer. From working all over the place, there seem to be essentially three different types of environments: 1) Hackers - These are the people who are willing to put in the wee hours to customize the system to their liking. This is typical of CS departments and students. 2) Competent Users - This group is willing to write small utilities and read the manuals, but they aren't interested in making BIG changes in the way their system works. These typically are scientific labs and the like. 3) Lusers - For this group, ignorance is bliss. The less they know about the system, the happier they are. They put numbers in, and they get numbers back. Typically business users. Granted, these are large generalizations, but they seem to catch most of the systems that I've seen around. Now, for each of these categories, we can talk about the 'best' system. 1) Hackers - A definite need for modifiability. Unix is the only reasonable choice for a small group. Over several hundred users requires a 20. 2) Competent Users - In this group, there is a definite eye towards expansion. As more people discover the computer, usage will grow until everyone has their own terminal. For a small group I would recommend VMS, but for more than about 25 people, I'd think a 20 would be more appropriate. 3) Lusers - Need to think here about canned software and good batch performance. Again, if the situation is small enough, I'd prefer VMS. Tops-20 does have the advantage of better hand-holding, but most lusers aren't going to be in touch with the computer at all. They'll hire you or me to do it for them. Of course, there are mixtures and other factors to consider. For example, I consider EDT under VMS to be superior to EMACS on a 20. So for an application that was heavily into word processing, I'd probably think twice about VMS. But then again, what do I know? Halon at the ready, Tony <Li @ Rutgers> ------------------------------ Date: 21 January 1983 17:58 EST From: Ken Harrenstien <KLH @ MIT-MC> Subject: Minor quibble ITS is one PDP-10 operating system that has a "pipe" IPC facility (called the "core link device"). It also allows any specific or wildcard filename to be "translated" into any other device/filename which is how redirection is accomplished. Other ITS features include filesystem links which are better than UNIX links, and the capability to create arbitrary new devices by writing an arbitrary user-mode program; for example, this is how a filename such as AR12:KLH;FILE NAME can refer to the entry FILE NAME in the archive-file KLH;AR12 X (on UNIX you would have to extract the entry from the archive file before giving it to a program). Alan Snyder (late of MIT-DMS) wrote a "shell" for ITS that used the various ITS capabilities to emulate UNIX features, along with a few utilities and stuff, but I don't know of anybody else who really adopted this as their standard top-level process. Whether that says anything about the viability of a UNIX emulator on TOPS-20 I don't know. I have hacked UNIX extensively and there are some things it is nice for, but I notice that I rarely use it when not actually grubbing around in the software to fix things. Maybe just got imprinted on PDP-10s. ------------------------------ Date: 20 Jan 83 18:27:41 EST (Thu) From: Speaker-To-Animals <speaker.umcp-cs@UDel-Relay> Subject: VMESS...the Operating System for the 60s. Date: 11 January 1983 12:10-EST From: Gail Zacharias <GZ @ MIT-MC> Subject: VAX VMS vs UNIX As far as I am concerned, the major difference between Unix and VMS is that there are Unix simulators for VMS, and no VMS simulator for Unix. So if you get VMS, you can take advantage of the features of either system, and run programs developed for either one. Of course there are no simulators for VMESS on Unix...who in there right mind would WANT to simulate VMESS when they have Unix? If you've ever used Unix or the Unix shells, then the advantages of Unix over VMESS make themselves self-evident. There are a reasons for simulating Unix on top of VMESS though, and it isn't because VMESS holds any inherent advantages over Unix. A number of firms are buying Unix simulators so that they can take advantage of the elegant Unix interface, and at the same time run their old VMESS software without all of the expensive conversions to Unix. Remember, companies have a lot of money tied up in software and people. The cost to switch to new software and retrain people to use an entirely different system can be enormous. Sometimes the company's contractors require software that runs only on VMESS. Unix is rapidly gaining greater acceptance in industry though, so if you're looking for a good software development tool with an elegant human interface...then Unix is probably for you! Unix forever! - Speaker ------------------------------ End of HUMAN-NETS Digest ************************