Bernard J.

Jansen

The Graphical User Interface

An Inlroduclion
Introduction trashcan until the trash person comes. These early users sat down in front of a
The first trash can icon did not allow this computer and faced a blank screen, with
There are a variety of uni- retrieval. This contextual incongruity only a prompt. The computer gave the
versity-level human-com- caused users many problems. As another user no indication what the user was to
puter interaction (HCI) example, the Windows 95 G U I is the do next. GUIs are an attempt to solve this
programs. Although a few most modern o f all GUIs. One would blank screen problem. At a conceptual
offer breath and diversity, expect it to be fairly well developed and level, a computer human interface is a
many students graduate relatively error free. However, o f the "means by which people and computers
from universities that offer approximately 90 complaints with Win- communicate with each other" (Norman
only one or two C H I dows 95, none are performance com- 1988). One can make an analogy
courses. As such, most stu- plaints. They are all human factors type between a computer system's G U I and a
4,..,a dents have a limited back- complaints, such as how to copy a file car's steering wheel. The wheel directly
ground in the various C H I and how to get rid o f annoying icons binds the driver to the operation and
areas. This article offers a (Anonymous 1995). Finally, people have functionality o f the vehicle. W h e n driv-
general overview in one so many complaints about the X-Win- ing, a driver should not have to concen-
area, graphical user inter- dowing System, the third major G U I trate on the steering wheel. In the same
faces (GUI). A G U I allows standard, that there is whole book about way, the G U I binds the user o f the com-
a computer user to move what is wrong with it (Hopkins 1996). puter system to the operation and poten-
from application to appli- This paper will survey the c o m m o n defi- tial o f the computer system (Bonsiepe
cation. A good G U I nitions of what a G U I is and review the 1993). A good G U I design removes the
makes an application easy, three c o m m o n G U I standards in the impediment o f communication with the
practical, and efficient to market today. It will then review three of computer system and allows the user to
use, and the marketplace the many human factor concepts under- work direcdy on the problem at hand
success o f today's software lying good G U I design, which are visual (Norman 1988). In computer science
programs depends on acuity, limits to absolute memory, and terms, the G U I is a visual operating dis-
good G U I design. Con- the principle o f grouping. The paper will play that the monitor presents on the
sider the Macintosh and then present the effect of these factors on monitor to the computer operator (Har-
the IBM-PC. Computer users view three G U I design areas, the amount of ding 1989). More specifically, a G U I is a
Apple's Macintosh computers as having presented information, the grouping o f specification for the look and feel o f the
the best GUI. Correspondingly, their information, and the placement o f this computer system (Bonsiepe 1993). GUIs
positive view o f the Macintosh system is information on the screen. Following this usually have c o m m o n characteristic such
almost double that o f the Windows section, the ramifications o f bad versus as windows, icons, menus, and pointers
users. Correspondingly, brand loyalty good G U I design will be addressed. Areas (WIMP). Collectively, W I M P s are pic-
among Macintosh users is almost 20% for research and likely directions o f future tures that bring forth a certain action or
higher than that for Windows users. The G U I design conclude the paper. an action space. The user issues com-
development o f new software is mands via the G U I to computer applica-
extremely expense. With success or fail- GUls tions. GUIs usually have three major
ure o f a product and maybe the entire components. These three components
Although there are numerous GUIs in
company dependent on the application's are (Hayes and Barab 1989): a window-
the market today, the exact definition o f
G U I reception in the marketplace, a ing system, an imaging model, and an
a G U I is still fuzzy. This may be due to
good G U I design is extremely important. application program interface (API). The
the fact that GUIs are relatively new.
Unfortunately, it is not easy to define if windowing system builds the windows,
There are three deJhctaGUI-standards
an application's G U I is easy, practical, or menus, and dialog boxes that appear on
that are the basis for all GUIs. This sec-
efficient. These are attributes that do not the screen. The imaging model defines
tion reviews the c o m m o n definition of
lend themselves to counting. The mar- the fonts and graphics that appear on the
GUIs, the history o f G U I development,
ketplace does attempt to access these screen. W I M P s are products o f both the
and G U I standards in the marketplace.
attributes, however (Bonsiepe 1993), but windowing system and imaging model.
even after over 10 years o f G U I develop- Definition Finally, the API is the means in which the
ment, there are still questions concerning user specifies how and what windows and
what is a good G U I design. For example, A G U I is a type of computer human graphics appear on the screen. The his-
the early Macintosh Apple used the Trash interface on a computer. It solves the torical development o f the G U I still
can icon as a metaphor for deleting files. blank screen problem that confronted impacts the three major G U I paradigms
However, one can pull items out o f a early computer users (Norman 1988). in the market today. Historically, all

22 April 1998 Volume 30, Number 2 SIGCHI Bulletin

modern GUIs are offshoots of the Apple puter system that limited the users to The quality of the design is the overrid-
Macintosh. This has lead to a great deal contextually correct answers. For exam- ing issue for all interfaces (Benbasat and
of standardization and consistency ple, once the user made a selection via a Todd 1993; Betts 1995). There are sev-
among G U I design criteria. Therefore, menu, the menu limited the user's subse- eral screen design guidelines. On the
most applications' GUIs adhere to one of quent actions. The user could no longer other hand, there is shortage of empirical
the three major G U I paradigms, the choose something meaningless. The studies substantiating these guidelines.
Apple Macintosh, the IBM Systems Macintosh's G U I has all three major This lack of empirical research is espe-
Application Architecture (SAA), or the components o f a GUI, which are the cially apparent for modern G U I designs,
X-Windowing System. While none of windowing system, an imaging model, such as Windows 95, Quicken 7.0, and
these G U I designs are perfect, the overall and an API. Dbase 5. In a narrower sense, there are
design concepts are good enough to empirical studies that have identified
make radical departures counterproduc- IBM SAA basic psychological factors that one
tive (Norman 1988), unless there are sig- should consider in the design of good
nificant performance enhancements. Unlike the Apple Macintosh, the IBM- GUIs. This paper will narrow the discus-
SAA is more than just a GUI. It is a sion to three primary contributing
History whole system of interfaces that can span human factors, which are:
machines from personal to mainframe
Researchers at the Xerox Palo Alto computers. As such, it includes many • The physical limits of visual acuity.
Research Center designed the first appli- functions that most GUIs do not, includ- • The limits of absolute memory.
cation with a GUI, the Xerox Star, in ing a suite of networking and database • The Gestalt Principle.
1977. The Xerox Star was unique tools. The SAA'S G U I portion has all
because the researchers carefully designed three G U I components. Another unique VisualAcuity
the computer human interface before item of the SAA is that the user does not Visual acuity is the ability of the eye to
they began designing the internal work- need a mouse to interact with the appli- resolve detail. The retina of eye can only
ings of the application. Unfortunately, cation. All actions can be executed from focus on a very small portion of a com-
the Xerox Star was too slow, and it was the keyboard, functionality not available puter screen, or anything for that matter,
not commercially successful. However, in the Macintosh GUI. The most com- at any one time (Wickens 1992). This is
Steve Jobs visited the Palo Alto Research m o n SAA-type GUIs are Windows 3.11 because, at a distance greater than 2.5
Center and saw Xerox Star. He returned for D O S and the Program Manger for degrees from the point of fixation, visual
to Apple Computer and subsequently OS/2. acuity decreases by half. Therefore, a cir-
hired several of the original designers of cle of radius 2.5 degrees around the point
Xerox Star. They first produced the MIT X - W i n d o w s System of fixation is what the user can see clearly.
Apple Lisa. Like the Xerox Star, the In the G U I world, this is the Rule of 1.7
Apple Lisa was not commercially success- Although a separate G U I standard, many (Sarna 1994). At a normal viewing dis-
ful. In 1984, they developed the com- X-Window based GUIs, such as Motif tance of 19 inches, 5 degrees translates
mercially successful Apple Macintosh. In and T C L / T K , have copied the look and into about 1.7 inches. Assuming a stan-
the broadest terms, the Macintosh's G U I feel of the IBM SAA. X-Windows is still dard screen format, 1.7 inches is an area
defined the look and feel of all GUIs the underlying library for these GUIs. about 14 characters wide and about 7
today. The X-Windowing System is the most lines high (Helander 1988). This is the
popular G U I for U N I X systems. This is amount of information that a user can
De Facto Standards because any X-Windows software can use take in at any one time, and it limits the
the X-Windows library, which gives it effective size of icons, menus, dialogs
The Apple Macintosh, the IBM SAA, great portability and standardization boxes, etc. If users must constantly move
and X-Windowing System are the para- across platforms. their eyes across the screen to clearly
digms for all modern GUIs. Because of focus, the G U I design is causing a lot of
their influence in the standardization of X-Windows also works directly with net- unnecessary and tiring eye movement.
today's GUI design, a brief description of works, which allows the G U I display to
the major features of each standard is be on one computer and the application InformationLimits
necessary. that the user needs on another computer. Once the user has a desired fixation
It does not matter if the two computers point, there is a limit to the amount of
Apple Macintosh are in different rooms or on different information that the person can process
continents. It addition to the three com- at one time. A G U I design rule of thumb
Apple introduced the Macintosh as a mon G U I components, X-Windows has is that the range of options or choices
computer "for rest of us." The G U I was a a collect of application tools and utilities should never be more than five or six
major part of the overall goal of the Mac- as a built in X-Library. (Miller 1956; Sarna 1994). Seminal work
intosh. All graphical applications copied by Miller is the basis for this rule. Miller
the Macintosh in its design and usage. Theoretical Background (Miller 1956) showed that absolute iden-
The Macintosh introduced the first tification using one-dimensional criteria
menu, icons, and point-and-click, mouse Although GUIs are an integral part of an was aboutseven items, plus or minus
driven processing. With these menus and application, GUIs are not inherently eas- two. He showed that this limitation also
icons, the Macintosh was the first c o m - ier to use than command line interfaces. held for memory span. Miller introduced

SIGCHI Bdletin Volume 30, Number 2 April 1998 23

the concept ofrecoding as a method that ciple, one can group like items together fact, the U N I X system is a good exam-
people use to store information. Miller using factors like color to add more infor- ple of what not to do.
also pointed out that by expanding the mational dimensions. Too m a n y colors, 2. Avoid unnecessary detail: For example,
identification criteria from o n e t o more however, destroy the global visual group- use whole numbers if one does not
dimensions people could handle more ing o f the items. The user then begins to need decimals. Keep the window and
choices and remember more. Later concentrates on the G U I . Any primary icon designs clear and simple. Even
researchers expanded on Miller recoding cognitive task attention devoted to the when users prefer more complex icons,
to develop the concept that people chuck interface may interfere with the primary elaborate icons add nothing to perfor-
information together in order to remem- task (Norman 1988). One can derive mance. Studies show that when icon
ber more information (Baddeley 1994; basis G U I standards from basic h u m a n designs are too complex, time to com-
Shiffrin 1994).This research has direct factors, however. These standards are the plete a task actually increases (Benbasat
impact on G U I design, especially con- presentation of information, the group- and Todd 1993). In studies with 3-D
cerning the number o f menu items and ing of information, and information and 2-D graphical displays, users pre-
icons. sequencing. ferred the 3-D displays. There were no
differences in performance between
Gestalt Principle Amountof InformationPresented the two graphical displays, however
T h e Gestalt Principle states that people The amount o f information to present is (Hubona 1995).
use a top-down approach to organizing the most basic o f G U I design consider-
3. Use concise wording: Screens have
data (Helander 1988; Wickens 1992). ations. H.E. Dunsmore (Helander 1988; limited space. Screen designers should
This principle can influence how one Reiterer 1993) showed that making
avoid the tendency to place additional
should organize graphical information screens less crowded improves screen
data on the screen just because the data
on the screen. T h e Gestalt school of G U I clarity and readability. As such, G U I is available. More objective limits of
designers has attempted to identify crite- designers usually follow the guidance
screen density vary from thresholds of
ria that cause people to group certain that the interface should display only
25% to 80% (Helander 1988). There
items together in a display. Proper group- what the user needs to perform the cur-
is no empirical research that substanti-
ing results in a necessary redundancy o f rent operation. Empirical researchers ates any performance enhancement
selection information that aids the user. show that limiting the information to with any specific threshold.
For example, if users know where one that necessary for the user reduces errors
4. Use familiar data formats: With more
item in a group is on a screen, they will and time to perform tasks. Errors and
familiar formats, the user will need less
expect other like items to be there also. I f performance time increase as the G U I
information to complete the task. An
one groups the items in line with this presents more information. O f course, it
example for data entry is the standard
expectation, it allows for accurate locat- requires a thorough analysis of the tasks
USA address format of street, city,
ing and better transfer o f information to that the user must perform in order to
state, and zip code. In additional to
the user. The top-down approach also display only the necessary amount o f
requiring less instruction, the user will
allows for the development o f emergent information. Compared to a randomly
placed screen, a well-designed screen can perform the operation faster than if the
features. An emergent feature is a global
format is unfamiliar.
property of a set that is not evident when reduce time needed to perform a task by
one views each item locally. Since global as much as 4 0 % (Helander 1988; Lin 5. Use tabular formats with column
processing tends to be automatic, one and Daly 1994). Ways to conserve screen headings: Tabular formats allow for
can argue that an emerged feature space are: efficient labeling o f related data. It is
reduces the attention demand as a user especially preferable for data location
1. Appropriate use of abbreviations: tasks. Simply splitting items on one
operates a multi-element display. For this
performance enhancement, one must use M a n y design documents recommend long line into a two-line result in pro-
the Gestalt Principle in the initial place- using complete words whenever possi- ductivity improvements o f 20%
ment, and the resulting organization ble. Due to screen sizing constraints, it (Sarna 1994). Also, LaLomia and
must be compatible with the user's cogni- is not always possible to use complete Coovert's research (Heiander 1988)
tive view of the task (Wickens 1992). words. W h e n complete words are not showed that locating a data value was
possible, abbreviations should be con- quicker in tabular form then in a ran-
GUI Design Considerations textual and consistent. A good contex- d o m or graph format. For trend analy-
tual example is "h," which is usually a sis, a line graph is quicker than raw
Considering the above psychological fac- good abbreviation to use for help. T h e data (Helander 1988).
tors, one could come to the conclusion number of abbreviations should not
that one could easily extrapolate these only be contextual but also be keep to Groupingof Information
factors to the design of a good G U I . a minimum. As a poor example, in the Given a set of information to display,
Empirical studies o f G U I s show that this U N I X system, the "ls" c o m m a n d list there are m a n y ways one can display the
intuition this is not always the case. The files in a directory. T h e "Is" c o m m a n d information. Proper grouping improves
Rule o f 1.7 directly leads to the conclu- has t 7 different one-letter abbrevia- the information's readability and can
sion that a good G U I would use a lot of tions that change the output options of highlight relationships between the
icons. Unfortunately, too many ran- the "Is" command. T h e one-letter information (Helander 1988). Tullis'
domly placed icons violate the limits of abbreviations have little contextual (Helander 1988)(Sarna 1994)experi-
absolute memory. Using the Gestalt Prin- link to the options they represent. In ments in the mid-1980s showed that the

24 April 1998 Volume30, Number 2 SIGCHI Bulletin

best predictors of search time were the faces (Sears 1993). Conversely though, The o p t i m u m sequence for screen pre-
number o f and size o f the groups. There- empirical studies have shown that, sentations is a collection of various fac-
fore, one should structure displays with counter intuitively, icons do not lead tors, including:
the limits o f visual acuity in mind. The to greater increases in performance.
.user needs to be able to take in the differ- 3. Highlighting: Besides color, there are 1. Sequence o f use: O n e needs to present
ent chunks of information at one glance several other methods of highlighting the user the information in the order
to improve readability. Overall, the best including reverse video, brightness, that the user will probably utilize it.
predictors of ease of use were density and underlining, and flashing. T h e most 2. Conventional Usage: I f a c o m m o n
item alignment. Empirical research c o m m o n use of highlighting is reverse convention is in general usage, the
shows that search time increases as the video to indicate an item that is cur- G U I design should continue using it.
size of the grouping exceeds 5 degrees of rently selected. GUIs usually use For example, in the standard window
arc and the number of groupings brightness to show which items are not layout, the file option is usually to the
increases above five (Helander active at a given time. Underlining is far left of the menubar.
1988)(Wickens 1992). With groupings effective if it does not interfere with the 3. Importance: T h e designer needs to
less than 5 degrees, the search duration is legibility of characters. Flashing will place the more important information
directly a function of the total number of both get attention and annoy if the in a prominent location. For example,
groupings on the screen (Helander user can not turn offthe flashing. if several entries are possible, the G U I
1988). There are several techniques to Therefore, one should use flashing should lead offwith the required ones
aid in the grouping of information, only to convey an urgent need. The and end with the optional ones.
which include: Apple Macintosh uses flashing to sig- 4. Frequency o f use: O n e should place
nal only program or data destruction. the most frequently utilized com-
1. Color: Presenting different groups Regardless of which type of highlight- mands at the beginning. For example,
with different color clearly creates ing, one needs to apply it conserva- in a menu list, the most frequently uti-
some degree of grouping among the tively. Overuse o f highlighting causes lized commands should be at the top
elements of the same color. GUIs that confusion among users and defeats its of the list.
utilize color well increase productivity. purpose. Additionally, if one high- 5. Generality versus Specificity: T h e
If like color items are in close proxim- lights the wrong information, the user more general items should precede the
ity, the visual association is stronger has more difficulty detecting the more specific items, especially when
than if the like color items are further important information (Helander there is a hierarchical relationship
apart. In addition to changing the 1988). among the data.
item's colors, one can use different col- 6. Alphabetical or Chronological: If there
ors for the background and fore- InformationSequencing is no other rules for ordering data ele-
ground. The effectiveness of this ment, then one should adopt some
O n e needs to lay out a screen in a manner other technique such as an alphabetical
technique decreases as the number of
that allows the user to easily find any or a temporal listing. Card (Helander
screen colors increases (Helander
information on it. Most designers advo- 1988) showed that selection time was
1988). Overuse of color degrades per-
cate the use of one the de facto G U I faster for alphabetical than for any
formance, however.
screen standards. This is because many other functional grouping. T h e goal of
2. Graphical Boundaries: Drawing users now expect certain modes of opera- any G U I is to allow the user to work
boundaries around elements is the tion in all GUIs. For example, most users through the computer and application
most c o m m o n method of grouping expect the top of screen to contain the to concentrate on the primary cogni-
elements in GUIs. Although there is headings for the pull-down menus. T h e tive task. T h e user should not be con-
no empirical evidence to show that top right is the default location for icons cerned with the user interface. Any
these groupings improve performance, representing the disk availability. In the attention devoted to the interface
users prefer this type of groupings Macintosh G U I , the bottom right con- interferes with the main task (Benbasat
compared to other methods. This tains the trash icons used for deleting and Todd 1993; N o r m a n 1988).
technique is especially popular with files. Within a window, there are also
the IBM SAA systems. Another many standard modes. A window title is Ramifications
method of grouping is to group tasks usually at the top. Scroll bars are on the
within icons. Icon grouping is easy right and bottom for vertical and hori- W h a t are the ramifications of G U I
because many icons can have c o m m o n zontal window movement. A box for design? O n e consistent result is that an
attributes. Icons are also small and closing the window is at the top left. increased operational knowledge transfer
therefore use less space (Sears 1993), Icons for resizing the window are at the between applications reduces training
less than 5 degrees o f arc. Another four corners (Helander 1988). Studies costs (Harding 1989). Training costs are
advantage of icons is that recognition show that most users initially scan the usually one to three times the cost of the
is faster for pictures than for text (Ben- screen starting at the upper-left corner. actual software and hardware (Boeri and
basat and Todd 1993). This makes it This corner should be the obvious start- Martin 1994). A good G U I design
easier for the novice to learn a system. ing point for applications invoked from reduces required training time to 20-30
Studies also show that icons have within the window. This permits a left- hours for a user to learn an application
smaller error rates than textual inter- to-right and top-to-bottom reading, (Comaford 1993). For businesses, this
faces and the same as for menu inter- which is standard for Western cultures. means that a good G U I saves m o n e y and

SIGCHI Bulletin Volume 30, Number 2 April 1998 25

time. Additionally, a good G U I improves most prominent location. The more gen- Hopkins, Don. Paper from The X-Windows
the user's perception of the application. eral items should precede the more spe- Disaster. Undated from the UNIXHaters
The user's first 15 minutes of usage for- cific items. If no other ordering exists, PSbb-Site. Accessed May 1996.
mulates the lasting impression of an one should alphabetize. The ramification Horton, William. 1990. Visual Rhetoric for
application. of good G U I design results in reduced On-line Documents. IEEE 7~ansactions on
training time and improved perfor- Profkssional Communication. Vol. 33. No.
Conclusion mance. Reduced training time means 3:108-114.
lower costs and improved user percep- Hubona, Geoffrey S. Evaluating. 1995. User
The primarygoal ofa G U I is to allow the tions. Bad G U I design prevents the user
user to concentrate on the task at hand. Interface Design with Belief Constructs.
from concentrating on the primary cog- Systems Sciences, 1995 Annual Hawaii lnter-
To do this, the G U I must make the inter- nitive task. This results in user frustra- national Conference. Vol. 4:700-709.
face between the human and the com- tions, decreased performance, higher
puter seamless. Modern GUIs adhere to Lin, Yi Bing and Dan Daly. 1994. A Flexible
costs, and possibly product and market-
one o f three de facto standards, which are Graphical User Interface for Performance
place failure. W h e n designing GUIs, one Modeling. MASCOTS 1994: Modeling,
the Apple Macintosh, the IBM SAA, and need to keep the objectives o f the G U I in
the M I T X-Windowing System. These Analysis, and Simulation International
mind and to generally avoid needless I /brkshop: 193-199.
standards are not perfect, but they are complexity (Marcus 1992). One must
good enough to preclude major devia- avoid useless innovation and concentrate Marcus, Aaron. The Future of Advanced User
tion. Future GUIs will probably utilize Interfaces in Product Design. T R O N
on improvements that enhance perfor-
one or more of these standards unless Project, 1992 Symposium: 14-20.
mance. Future trends in GUIs are toward
major performance enhancements result. voice recognition and hypertext format- Miller, George A. 1956. The Magical Num-
Utilizing key psychological factors, G U I ting language (Hayes and Barab 1989; ber Seven, Plus or Minus Two: Some Limits
designers can achieve a seamless com- H o r t o n 1990). The hypertext trend on Our Capacity for Processing Informa-
puter human interface. The three pri- allows the user to move directly from tion. Psychological Review. Vol. 101. No.
mary human factors that directly affect 2:343- 352.
data and concepts in one application to
G U I design are visual acuity, the limits of similar data and concepts in other appli- Norman, Donald. The Psychology of Everyday
absolute memory, and the grouping o f cation. These trends will further remove Things. New York: Basic Books: 1988.
information. At about 19 inches from an the G U I as an obstacle between the user
object, a person's visual acuity is about 5 Reiterer, Harold. The Development of
and the task. Design Aid Tools for a Human Factor
degrees of arc. There appears to be a limit
Based User Interface Design. Systems, Man,
to absolute memory o f about 7 items. References and Cybernetics, 1993 International Cona~r-
Grouping o f information based on the ence~ 361-366.
Gestalt principle appears to aid in infor- Baddeley, Alan. 1994. The Magical Number
mation processing. Use of these factors Sarna, David E. and George J. Febish. What
Seven: Still Magic After All These Years?
result in G U I design principles that gov- Psychological Review. Vol. 101. No. 2:353- Makes a GUI Work? Datamation Vol. 4.
ern the amount o f information to 356. (July 15 1994):29f.
present, the proper way to group this Benbasat, Izak and Peter Todd. 1993. An Sears, Andrew. 1993. Layout Appropriate-
information, and the proper placement Experimental Investigation of Interface ness: A Metric for Evaluating User Interface
and sequencing of this information on Design Alternatives: Icon vs. Text and Widget Layout. IEEE 7~ansactions on Soft-
the screen. A good G U I should present Direct Manipulation vs. Menus. Interna- ware Engineering. Vol. 19. No. 7:707-720.
information that is contextual and con- tional Journal of Man-Machine Studies. Vol.
Shiffrin, Richard M. and Robert M. Nosof-
sistent. It should avoid unnecessary detail 38:369-402.
sky. 1994. Seven Plus or Minus Two: A
and use concise wording to conserve Betts, Mitch. Standard GUIs Make Sense. Commentary On Capacity Limitations.
screen space. If familiar data formats Computerworld. Vol. 28. No. 11:75-76. Psychological Review. Vol. 101. No. 2: 357-
exist, the G U t should utilize them. A Bonsiepe, Gui. 1993. Interpretations of 361.
G U I needs to group information using Human User InteoGce. Visible Language.
Vol. 24. No. 3:262-285. Wickens, Christopher D. EngineeringPsychol-
color to associate like items. Graphical ogy and Human Pe~brmance. 2d ed. Harp-
boundaries are a very effective means to Boeri, Roberts J. and Martin Hensel. 1994.
Mar Human Factors in Business CD-ROM ers: New York: 1992. p. 24-109, and 116-
group like items, especially icons. Other 160.
Titles. CD-ROM Pro)~ssional. Vol. 8. No.
highlighting techniques include reverse 2:107-108. Anonymous. 1995. IgHndows 95 Annoyances.
video, brightness, underlining, and flash- http://www.creativelement.com/
Comaford, Christine. Graphical User Inter-
ing. One needs to sequence information faces: Keep Them Sleek and Simple. Com- win95ann. Dated 15 November 1995.
on the screen to facilitate the user. The puterworM. Vol. 25. No. 16:37-40.
presentation o f information should fol- Harding, Bruce A. Windows & Icons &
low the sequence that the user needs it.
Author's Address
Mice, Oh My! The Changing Face of Com-
C o m m o n information needs to be in puting. Frontiers in Education Confbrence
c o m m o n locations across windows and 1989:19 th Annual: 337-342. Department of Electrical Engineering
GUIs. The most important information Hayes, Frank and Nick Baran. A Guide to and Computer Science
needs to precede the lesser important GUIs. ByteVol. 4. (July 89):250-257. United States Military Academy
information. Frequently utilized infor- Helander, Martin ed. Handbook of Human- West Point, New York, 10997, USA
mation or commands need to be in the Computer Interaction. New York: 1988. [email protected]

26 April 1998 Volume 30, Number 2 SIGCHI Bulletin