Joystick Versus Mouse

Pointing DeviceCommunication

By Vic KleyKA Design Group

Oakland, CA

For more than 20 years designershave puzzled over the problem ofhow best to enter information intothe computer. They have concludedthat a device using the controlledmotion of forearm, wrist, thumb andindex finger to point to graphic pat-terns on the display screen couldsubstantially facilitate the controland instruction of the computer.Many believe that pointing devicesprovide the most powerful non-key-board way to communicate with acomputer, and even the computingpublic has recently shown anemerging willingness to acceptpointing devices as an improvedmode of computer interaction. Al-though the current popular belief isthat the "mouse" is the only appro-priate pointing device for all appli-cations, nothing could be furtherfrom the truth.

In 1967, a classic study estab-lished that the digitizing tablet wasthe best overall pointing device. Butthis much-quoted study—by Eng-lish, Engelbart and Berman—un-derstandably did not concern itselfwith the prohibitively high cost ofthe tablet, since mainframe com-puters were the order of the day.Even today, the digitizing tablet re-mains the best general pointing de-vice in situations where money is noobject.

But, with the advent of inexpen-sive microcomputers, studies haveshifted their attention to low-costpointing devices. In a 1978 study byCard, English and Burr, the joystick,mouse, arrow keys and text keyswere compared for text selectionand editing performance. The Cardstudy (also known as the Xeroxstudy) concluded that the mousewas "found to be fastest on allcounts and also to have the lowesterror rates." Although proponentsof the mouse have relied heavily onthis conclusion, the crucial point is

what kind of joystick is comparedwith the mouse.

Until recently, there have beentwo kinds of joysticks available: theabsolute and the rate. With an ab-solute, the travel of the joystick cor-responds directly to the scope ofthe display. Moving the absolutejoystick to its topmost positionplaces the cursor at the top of thedisplay. Moving it to its lower leftposition places the cursor at thelower left corner of the display, andso on.

With the rate joystick, the travelof the joystick imparts a direction ofmotion to the cursor. Push the ratejoystick to the left, and the cursorwill move from wherever its presentposition is. Push it to the lower left,and the cursor will move diagonallyin that direction. Let the stick returnto center, and the cursor comes to ahalt. Furthermore, in typical de-signs, by varying how far in any di-rection the rate joystick is moved,one controls the speed with whichthe cursor will move. When the ratejoystick is moved up slightly, thecursor rises slowly. Push it to thetop of its travel, and the cursormoves upward at maximum speed.

While the Card study did findmarginal advantage in positioningspeed for the mouse (Figure 1), thiswas only in comparison to the ratejoystick. An absolute joystick wasnot included in this study. In the ear-lier English study, the absolute joy-stick was found to meet or exceedthe positioning speed of the mouse.It should be noted that, in applica-tions where it is desirable to inte-grate a pointing device into a key-board, the mouse is ruled out. Insuch cases, an arrow or text key de-sign could be used, but a joysticksolution would be far superior.

A New Kind of JoystickRecently preliminary tests were

69

conducted using a NorthStar Ad-vantage computer and an abso-lute/rate joystick, a hybrid thatcombines the advantages of bothtraditional designs. A switch on theside of the joystick handle under thethumb (adjustable for left- or right-hand operation) is depressed to se-lect absolute mode. In this position,a motion to the upper left corner ofthe joystick corresponds to the im-mediate movement of the graphicscursor to the upper left corner ofthe computer display. Once the tar-get area is reached, the thumb-in-dex finger pressure is relaxed. Theswitch is thus released, and thejoystick reverts to rate operation.This hybrid combines the large-mo-tion positioning speed of the abso-lute joystick with the rate joystick'ssmall movements and fine control.As a result, in applications wherecoarse positioning speed predomi-nates, it is expected that the abso-lute joystick mode (which is lock-able on software request or bydouble-clicking the thumb-actuatedstate switch) will easily meet or ex-ceed the positioning speed of themouse.

For small cursor movements,even the Card study found the ratejoystick and mouse essentially iden-tical (only a 1% to 3% difference) formovements of less than an inch(Figure 1). Furthermore, in the ratemode, a 15-degree motion from thecenter of the joystick produces anincremental change in the output ofthe joystick of one part in 4,096—

\.,the digitizingtablet remains

the best generalpointing device

in situationswhere money is

no object"

ten times the precision of the bestmouse available.

The Need for More ResearchA pointing device must be compati-ble with the requirements of bothman and machine. Yet far too littleattention has been paid to the jointsand muscles that will have to oper-ate the device—often for hours at atime. A thorough examinationshould not only focus on the re-quired positioning precision for thespecific task, but also on the dura-tion of the task for any musclegroup and its particular fatigue fac-tor. Such a study would be an im-portant guide to both the engineerand the system integrator. In addi-tion, studying muscle groups wouldestablish evaluation criteria for fu-ture pointing device development.

The two main muscle groups in-volved in pointing devices are fore-arm/wrist and thumb/finger/wrist.

The general relationships betweenthese two groups and the variouspointing devices are, of course, in-herent in the basic designs. Move-ment of the forearm/wrist groupgenerally requires more energy, andis more fatiguing, than movement ofthe thumb/finger/wrist.

The digitizing tablet, in makinguse of forearm and wrist motions,closely approximates the activity ofwriting and drawing. And, becauseof the very high lineal resolution ofthe tablet, thumb/index finger mo-tions can be captured as well. Thetablet is the only device discussedhere that can represent both the

Effect of Target Distanceon Positioning Time

ArrowKeys /

Text Keys

2 4 6 8Distance in Centimeters

15

Figure 1

Decision Factors in Pointing Device Selection

u

70

PointingDevice

DigitizingTablet

AbsoluteJoystick

RateJoystick

Mouse

Absolute/RateJoystick

Arrow orStep Keys

Light Pen

Figure 2

ResolutionCapability

High

High

High

'"High '

High

N/A

Low

PositioningSpeed

High

High

Medium

"High"

High

Low

High

Cost

High

Low

Low

Med/High

Med/Low

Low

Medium

FatigueFactor

Medium

Low

Low

Medium

Low

Medium

High

Portable?

No

Yes

Yes

Yes

Yes

Yes

Yes

Compact?

No

- ...Yes

Yes

No

Yes

Yes

No

IO

Io5

thumb/finger/wrist motion andforearm/wrist motion.

The optical or mechanical mouseuses forearm/wrist motion exclu-sively (except for switch activation),while the joystick uses the motion ofthumb/finger/wrist motion exclu-sively.

Operation of arrow or text keysrequires only the repeated "on/off"striking motion of the finger andwrist. Since the key is either up ordown, this device does not take ad-vantage of the natural human abilityto control and modulate move-ments.

Text Editing RecommendationsAlthough none of these devices isappropriate for every situation, ten-tative guidelines can be suggestedeven without the benefit of detailedstudies. The most common use ofpointing devices today is for text ortext-related editing in which a key-board is a necessity. A keyboard-mounted pointing device is there-fore advantageous, because italways leaves the wrist and forearmin a known relationship to the key-board. Thus, a return hand motionis minimal, direct, and repeatable.By contrast, the return hand motionwith either the tablet or mouse is in-determinate and not automaticallyrepeatable. Hence, the editing pro-cess is interrupted and delayed.

There is also reason to believethat, for text editing, the absolute/rate joystick may actually positionfaster and with fewer errors than themouse. Once the joystick in abso-lute mode has placed the cursor inthe region of a text line or charactercolumn, the rate mode can be usedto move it easily in a horizontal orvertical direction. This allows therapid selection of a character withina line, or a line within a paragraph.And, when moving the cursor alonga line of text in the rate mode, an in-advertent upward deviation thatdoes not exceed 15 degrees will notpush the cursor off the text line ofinterest. Thus, the cursor move-ments tend to adhere to the line-and-column grid of written text.With a mouse or tablet, however, aninadvertent motion such as thiswould actually displace the cursorone or more text lines, resulting in apositioning error.

General ApplicationE Recommendations

The following guidelines are sug-72

"...forcomparable

applications, ajoystick is

typically 60 to 80percent less

expensive thana mouse"

gested for other applications, andare based primarily on the resolu-tion and positioning speed required,as well as on cost sensitivity. How-ever, factors such as space require-ments, fatigue, compactness, andportability are also taken into ac-count (Figure 2).

The cost of a pointing device willdepend, in large measure, on thedegree of resolution required. How-ever, for comparable applications, ajoystick is typically 60 to 80 percentless expensive than a mouse.

A portable device can be incor-porated into a portable computer'scarrying package, while a "com-pact" device can be integrated intothe computer keyboard. "High"resolution means the operator'sability to resolve over one part intwo thousand (2048 x 2048 pictureelements). "Medium" resolutioncorresponds to a 512 x 512 display,and "low" resolution to 256 x 256pixels.High Resolution: For high-resolutionapplications where cost is not a pri-mary consideration, the digitizingtablet is outstanding. The tablet of-fers natural, absolute mapping ofthe user's motions onto the displayspace, with resolution as high as500 to 1,000 points per inch. Deskarea must be provided for the tab-let, and the fatigue factor is moder-ate. This device is neither portablenor compact.

For applications that are cost-sensitive, however, either a mouseor absolute/rate joystick is recom-mended. The mouse offers highspeed and high resolution (100-300points per inch), but does not offernatural, absolute mapping. Somedesk area must be provided, andthe fatigue factor is moderate. Thisdevice is portable, but not com-pact.

The absolute/rate joystick offershigh speed and high resolution(100-400 points per inch in absolutemode; 1,000-10,000 points per inchin rate mode. The joystick providesnatural, absolute mapping in theabsolute mode, and no extra deskarea is required. The fatigue factoris low, and the device is both porta-ble and compact. For high-resolu-tion applications that are extremelycost-sensitive, the absolute/ratejoystick is recommended.Medium Resolution: Either a mouseor an absolute/rate joystick is rec-ommended for medium-resolution,cost-sensitive applications. For thisdegree of resolution, the joystick'srate mode would be adjusted to op-erate more coarsely.Low Resolution: For low-resolution,cost-insensitive applications, eithera mouse or an absolute/rate joy-stick would be appropriate. For ap-plications that are cost-sensitive,the absolute joystick is recom-mended, since at this resolution therate mode is dispensable. The ab-solute joystick offers high speed,provides natural, absolute mapping,and no extra desk area is required.The fatigue factor is low, and thedevice is both portable and com-pact.

For extremely cost-sensitive ap-plications, step keys are recom-mended. Step keys, which functionmuch like a simple rate joystick,provide an ultra low-cost pointingdevice. No extra desk space is re-quired, the fatigue factor is moder-ate, and the device is both portableand compact. •

ReferencesCard, Stuart. K., William K. English, and

Betty J. Burr, 1979 "Evaluation ofMouse, Rate-Controlled Isometric Joy-stick, Step Keys, and Text Keys for TextSelection on a CRT," Ergonomics, Vol.21, No. 8, 601-613.

English, W. K., D. C. Engelbart, and M. L.Berman, 1967, "Display-SelectionTechniques for Text Manipulation,"IEEE Transactions on Human Factors inElectronics, HFE-8, 1, 21-31.

Vic Kley is president of KA DesignGroup, and has been involved incomputer graphics and input devicedesign since 1977. He holds six pat-ents in input device technology andhas two more pending.