info-mac@uw-beaver (info-mac) (11/30/84)
From: Christopher A Kent <cak@Purdue.ARPA> I hadn't seen this ad till today, and it is indeed intriguing. Particularly the claim of "reduc[ing] the time it takes to edit a document by 40%". I'd be very curious to know how this was measured. I've always found it more difficult to position trackballs, and certainly it's harder to move along a single axis. Unfortunately, Card, Moran and Newell didn't ever publish measurements of trackballs when they were comparing input devices. I wonder if they did any work with them, or just considered them inferior... chris ----------
info-mac@uw-beaver (info-mac) (11/30/84)
From: Nick <NNicoll.ES@XEROX.ARPA> In a further discussion of the trackball last night we agreed that the trackball might be superior for text editing but what about MacPaint? Most Xeroids that I know can sign their name with a mouse and we wondered if that would even be possible after long experience with a trackball. Has anyone investigated attaching a Xerox optical mouse to the Mac? I am getting tired of cleaning the AppleMouse which, even when it is freshly cleansed, has nowhere near the performance of my (DLion) work mouse. \\ Nick
davet@oakhill.UUCP (Dave Trissel) (12/04/84)
In article <2359@uw-beaver> info-mac@uw-beaver (info-mac) writes: >From: Christopher A Kent <cak@Purdue.ARPA> >I hadn't seen this ad till today, and it is indeed intriguing. >Particularly the claim of "reduc[ing] the time it takes to edit a >document by 40%". I'd be very curious to know how this was measured. >I've always found it more difficult to position trackballs, and >certainly it's harder to move along a single axis. > >Unfortunately, Card, Moran and Newell didn't ever publish measurements >of trackballs when they were comparing input devices. I wonder if they >did any work with them, or just considered them inferior... > >chris >---------- I visited a customer in the Midwest which was trying to do Macintosh-like things on a trackball. The Boss was forcing the developers (hardware and software alike) to go the trackball route because previous machines already implemented them. The results was a total catastrophy (according to them.) However, it would be interesting if thru software smarts there were several levels of speed depending on precise intervals of rotation of the ball. The major complaint voiced was that it took too many rolls to move the cursor to the other end of the screen. (Note the Mac has an option in the control panel to skip every other pixel if the speed of the mouse goes above a certain limit.) It would be interesting to hear of anyone elses experiences. Motorola Semiconductor Dave Trissel Austin, Texas 32-Bit Applications Engineer
info-mac@uw-beaver (12/07/84)
From: decvax!decwrl!amdcad!fortune!rhino!redwood!rpw3@uw-beaver.arpa +--------------- | However, it would be interesting if thru software smarts there were several | levels of speed depending on precise intervals of rotation of the ball. | The major complaint voiced was that it took too many rolls to move the cursor | to the other end of the screen. (Note the Mac has an option in the control | panel to skip every other pixel if the speed of the mouse goes above a | certain limit.) | | It would be interesting to hear of anyone elses experiences. | ... Dave Trissel +--------------- A couple of years ago at a COMDEX show, I had a long talk with a manufacturer of (mostly military) trackballs about the great mouse-vs-trackball debate. He agreed that uncompensated trackballs are certainly worse than mice for office automation uses, BUT... that if you properly compensated in software they were the same as (or maybe even a bit better than) mice. The military had apparently funded some ergonomics studies (since they tend to use trackballs in target locators in very rugged conditions) on various flavors of mice, trackballs, joysticks, "torque"sticks (force-sensitive joysticks that don't move). Particularly cute were some thumb-operated torquesticks that fit on the end of a fighter plane's jotstick. Anyway, the key seemed to be to design the entire system as a non-linear control system, INCLUDING the human element, and massage the human input with an appropriate inverse transform to linearize the portion of the system from the human's intention to the cursor's position. One of the reasons an uncompensated mouse is so easy to use is that our hand's kinesthetic reflexes (internal feedback, both position and velocity) are already compensated (by evolution and childhood training) to a position-linear target acquisition, and the mouse just passes that along linearly to the application program, which (usually) converts it linearly to screen position. Likewise, trackballs are difficult because the fingers aren't "programmed" to acquire large step changes in angular position (via a ball) quickly. (Note that fine control of local position is as easy or easier on a trackball as a mouse, but the trackball loses when trying to zip across the whole page and land on a single pixel.) The various XXXsticks have similar ills. However, if you analyze the eye/hand/cursor system using standard system analysis tools employed in "servo" design (such as impulse response, large signal step response, white noise response, dead-band and dither, etc.), this guy claimed that you could "linearize" the system by putting an equalization filter between the trackball output and the cursor screen position (in software). The equalizer would model the inverse transform of whatever the eye/hand/trackball response was, so that the net system of eye/hand/trackball/equalizer was linear (or nearly so). (It's actually a little more complicated than this, since the eye/hand system has non-linear elements in it such as slew-rate-limiting, reaction time, and variable inertia. Still, there are servo systems that manage to give smooth response with only "bang/bang" output, and the human fine control is already pretty good, anyway.) The claim was that with such proper compensation, ANY of the common input devices (trackball, joystick, torquestick) could give target acquisition speeds as good as a mouse to untrained or lightly trained users. The idea of using the user's actual uncompensated response as a way to design the compensation intrigues me. Has anyone else heard anything about this? Does anyone know of a company that has TRIED using non-mice in office applications with ergonomically designed compensation (other than simple two-step scale changing as was mentioned above for the Mac)? Such compensation could, I suspect, MARKEDLY improve even stupid cursor keys found on most dumb terminals. I know that we made a stab in that direction back at Fortune Systems, in that the auto-repeat on the keyboard has two "gears", and shifts into "high" if you hold the key down long enough (like some other terminals), but from personal experience using "vi" it just isn't good enough. The irony is that it feels like it MIGHT be much better, if the timing were more dynamic and took into account keystrokes that followed one another, instead of just acting when you held a key down. Suppose your editor actually timed the intervals between keystrokes of the cursor keys and used some sort of CVSD algorithm to translate that into cursor motion? Has anyone done THAT? Rob Warnock Systems Architecture Consultant UUCP: {ihnp4,ucbvax!amd}!fortune!redwood!rpw3 DDD: (415)572-2607 Envoy: rob.warnock/kingfisher USPS: 510 Trinidad Ln, Foster City, CA 94404