Original Macintosh Icons Made Personal Computing Understandable

Susan Kare arrived at Apple as an art-history PhD who knew little about computers. That made her part of the audience the Macintosh group hoped to reach. The stated goals were that anyone could use the machine, that it should work more like an arcade game than a system requiring manuals, and that it should feel friendly.

Macintosh advertising expressed the same ambition: “Since computers are so smart, wouldn’t it make more sense to teach computers about people, instead of teaching people about computers?” Kare says the line was a North Star for the Macintosh group, and that rereading it recently made her think about AI and current technology.

Her first assignment made the practical problem clear. She was shown an early screen crowded with text controls, including “Do It,” “Eject,” and “Rename,” and asked to improve it through typefaces and symbols. The interface already had one feature that seemed magical to her: users could click a floppy-disk-shaped object with a cursor and drag it around. But the task was to give the system a more understandable visual vocabulary. “Do It” became “OK,” partly because people read the original wording as “dolt.”

The system had to communicate actions at a glance. Kare’s early MacPaint tools—a paint bucket, lasso, pencil, spray can, and hand—were recognizable objects and gestures rather than technical diagrams of software functions. A lasso could suggest selecting an irregular region without requiring a manual to explain selection.

The technical limits were severe: black and white in a 16-by-16-pixel grid. Kare’s view is that designers should understand constraints and become interested in them rather than resent what they cannot have. A tiny monochrome grid could not support much detail, but it could still hold a recognizable idea. The early tool icons show the discipline in practice: each had to retain an identifiable object, gesture, or consequence after almost everything descriptive had been removed.

Paul Rand’s advice supplied a broader standard. Rand, whom Kare later worked with at NeXT, told her to focus on making things meaningful and memorable. Kare connects that advice to his formulation that “simplicity is not the goal. It is the product of a good idea and modest expectations.” The form follows from the idea; stripping away detail does not, by itself, make a design simple.

Kare prepared for her Apple interview by borrowing typography and font-sample books from the Palo Alto library and carrying them under her arm so she would seem as if she knew what she was talking about. Her prior experience was limited but practical: she had used a phototypesetting machine to make museum-exhibit labels and had become adept at applying Letraset transfer lettering straight and with careful kerning.

Andy Hertzfeld, her high-school friend and a system-software programmer at Apple, asked her to bring bitmap-like examples made on graph paper. He did not specify what they should be. Kare drew hands, objects, puns, letters, and figures on grids. Some apparently random sketches later had a family resemblance to production work: a pointing hand became a digital pointing hand, and an early face with a speech cue resembled an eventual alert icon. The interview notebook later entered MoMA’s permanent collection.

Once at Apple, Hertzfeld wrote an icon editor that let her toggle individual bits, inspect an enlarged version of an icon, see it at normal size, and generate the hexadecimal code needed to display it. It had no shape tools and nothing like a modern undo history, but she could revise an image without an eraser and immediately judge whether a change worked at actual display size.

Kare describes that as a major shift in the work. Her 32-by-32-pixel portraits showed what iteration could accomplish: with roughly a thousand dots and enough adjustments, she could create some semblance of a person’s likeness. The requirement was not simply to arrange pixels attractively while enlarged. It was to make a face read when reduced to its intended size.

Her first system font, Chicago, worked under a similar discipline. Its capital letters were nine pixels tall. The lettering already on screen looked jagged to Kare, and she tried to tidy it through horizontal lines, vertical lines, and 45-degree diagonals. She still winces at the lowercase x, but she could fit the rest into a more orderly system.

The Macintosh also supported proportionally spaced fonts. Kare says she did not initially realize what a significant technical advance that was, but it corrected a visual problem she had seen elsewhere: a wide letter such as an “m” could be cramped while a narrow “i” occupied excessive space. Chicago became the title-bar font and later appeared on the first iPod, an unexpected descendant of the original Macintosh interface.

The system also shipped with Monaco, New York, and Geneva. Kare and Hertzfeld had initially named fonts after towns on the Paoli Local near their Philadelphia-area high school: Paoli, Wynnewood, and Rosemont. Steve Jobs accepted city names in principle but told them to use “world-class cities,” producing New York and Geneva. Kare’s deliberately mixed “Ransom” face, later called San Francisco, remained a playful exception: it was based on newspaper-cutout lettering associated with ransom notes, not a model for system readability.

Kare traces her preference for simple icons to Scott McCloud’s Understanding Comics. McCloud argued, in her summary, that a detailed face looks like somebody else, while a simple cartoon face lets more people project themselves onto it. She brought that idea to the Happy Mac.

A 128K Macintosh took time to boot. The team wanted a 32-by-32 image that could reassure someone while the computer was sitting there and starting up. Kare chose a smiling computer, partly from her memory of a smiley-face button she had owned around age 14. The smile signaled that everything was okay without a line of explanatory text.

The Happy Mac also avoided the specificity of a portrait. Kare uses a pencil as another example. A simple drawing of a pencil can stand for writing more effectively than a highly rendered chrome pen with a reflection: the chrome pen may be elegant, but it is a particular sort of pen. A simple pencil leaves the user with the broader idea of writing.

She compares the desired economy to a school-crossing sign. Its children are silhouettes. There is no technical reason the sign could not include plaid lunchboxes, shoelaces, or children singing, but those details would distract from immediate recognition.

Kare’s own printer icon demonstrated the cost of getting that balance wrong. The original image showed sprocket-holed paper because the ImageWriter printer actually used paper that attached to sprockets. It was accurate hardware detail, but too tied to one product design. A later, simpler printer icon had a longer prospective life because it depicted printing rather than a particular printer. Kare applies the same concern to the floppy-disk save symbol: product designs age, while a metaphor can last longer.

Application and document icons used another compact distinction. The MacPaint application icon showed a hand actively painting; the associated document used a related paintbrush image. Kare describes the system as a verb and a noun. The application does something, while the document is the file associated with the action. MacWrite and MacDraw followed the same pattern. It seemed workable when the Macintosh had only a handful of applications; the team was not yet designing for a world with thousands.

Other images were deliberately stripped of particulars. The alert icon was a cropped profile, not a person with a square head. Kare intended it to reveal nothing about gender, age, hairstyle, or other specifics. The waiting cursor became a wristwatch instead of an hourglass because Kare knew sand timers as egg timers and wanted an updated metaphor for the 1980s.

The icons that failed were often emotionally vivid. Kare was asked to make an alternative to the Happy Mac for the supposedly unlikely case that something went wrong. She drew a “sick” Macintosh with X-shaped eyes and a sad mouth. For a total system failure, she made a Bugs Bunny-style bomb with a lit fuse, assuming users would never encounter it. They did. While Kare was working in Apple’s software-group “fishbowl,” a receptionist relayed a call from someone worried that the bomb on her screen meant the computer might explode.

The bomb was unmistakable, but it communicated the wrong thing at the moment a user most needed reassurance. Kare’s practical conclusion was not to avoid personality. It was to iterate.

The Command symbol began as a branding constraint rather than an effort to create an enduring keyboard mark. Early Macintosh menus used Apple logos beside shortcut commands. Jobs saw the repeated marks and said there were too many Apples on the screen—an “Apple farm.” The logo, he argued, was precious and should be treated accordingly.

Kare was told that the shortcut modifier was a command key or a feature key. “Command” led her first toward police badges, whistles, deputy-sheriff stars, and the Ten Commandments. None offered a promising direction. She went back to the library and looked through books of symbols.

There she found the looped-square mark labeled “feature.” It looked a little like a four-leaf clover and, to Kare, also resembled a cloverleaf highway interchange—a way to get somewhere quickly. It had a more practical advantage as well: it was easy to draw in a 16-by-16 grid. The team used it as the menu-command symbol, and it remains on Apple keyboards.

Kare initially felt uneasy about using an abstract mark because abstract symbols can be harder to remember than metaphors that reinforce their meaning. A later trip to Sweden made the mark seem less arbitrary. She saw it on road signs indicating nearby sites of interest, such as monuments. Someone told her that the symbol derived from Borgholm Castle, a medieval ruin whose aerial footprint is a square with four round corner towers. The sign that had seemed abstract had a literal architectural source.

Kare places this discovery within a larger art-historical perspective: people have made compact symbols for a very long time. She points to ancient carvings, Egyptian hieroglyphics, Viking inscriptions, Byzantine mosaics, and stitched samplers as earlier examples of text, images, and marks adapted to constrained surfaces. She also sees the Cairo font on the 1984 Macintosh as an early form of emoji: it let users combine pictures and text before texting made that commonplace.

Her childhood interest in tiled crafts fits the same pattern. Kare shows an ashtray she made at age seven from a small grid of tiles, joking that perhaps she had been trained early to want to make things on grids. But the grid itself is not the point. It is the material through which a metaphor becomes legible.

Icon design, in Kare’s account, is not an exact science in which a brief dictates one inevitable result. Her alarm-icon studies for Swatch make the point. Asked to design an alarm, she began with a clock, then tried versions that were noisier, more literal, louder, simpler, or more like a travel alarm. She also explored a rooster, bell, elephant, and other possibilities. A clock on a clock became too self-referential.

The alternatives did more than offer a client different styles. Each carried a different theory of what an alarm should evoke: time, sound, waking, travel, urgency, or an object already familiar from daily life. Kare says she likes to show what a client requested or what a programmer used as a placeholder because such starting points can lead somewhere useful. The job is often to refine an existing thought rather than discard it wholesale.

Hertzfeld gave her a related rule for showing work to Jobs: never present only one thing. If there is one proposal, the response can simply be no. If there are several, the question becomes which one the reviewer prefers. That gives the other person a way to participate and produces more usable feedback than an undifferentiated rejection.

The move from monochrome Macintosh graphics to the 16-color palette of Windows 3.0 widened the range of possible distinctions. Kare says even a limited palette, including colors she did not consider especially good, made a difference: red could identify alerts, and categories could separate at a glance. For a wallpaper design, however, the additional colors created a different risk. She dithered them to make a softer pattern rather than what she calls a garish chessboard. More available visual material did not remove the need for restraint; it changed what restraint required.

Her Facebook virtual-gifts work supplied a more direct form of response. Users could buy individual images for a real dollar and send them to one another. Kare and collaborators released a new gift every day for nearly four years, and she says they could often tell within 15 minutes after midnight on the East Coast whether an item would become successful.

She expected expensive-looking objects—diamond earrings and Rolex watches—to feel more worth a dollar. Cute images did better. Teddy bears, hearts, and the kiss mark succeeded; the kiss mark was the all-time bestseller. Multiple limited-edition penguins sold out. The lesson was not merely that user data should overrule design judgment. Kare’s team had generated a large range of possible gifts, then found that the social meaning people wanted to send was often affection or charm rather than simulated luxury.

The common principle across these cases is that designers need both variation and exposure to response. Generate alternatives, make the different meanings visible, and then learn from collaborators, clients, or users which meaning actually lands.

Kare’s collected advice remains a work in progress rather than a single design doctrine. But several lessons bear directly on interface work: use a limited vocabulary well, put an idea into a form people can assess, preserve simplicity for common tasks, and avoid making users feel deficient when a product is difficult.

Rand’s advice on fonts was blunt: “You only need about five fonts. You just have to understand how to use them.” Kare noticed that he also reused a relatively stable set of colors. The point was not that few fonts or colors are inherently better. It was that a designer should develop judgment within a vocabulary instead of assuming that novelty or abundance will solve a weak idea.

His logo advice followed the same logic. Companies without a billion dollars to spend teaching people an arbitrary mark, Rand told her, should try to unite the name with an image that reinforces it. Kare contrasts this with a relatively detached symbol such as AT&T’s globe: a company with enough resources can make a broad abstract mark familiar, but most cannot. Rand’s work offered more immediate associations. Morningstar used a rising sun; IBM was striped and businesslike; the UPS mark combined a shield and a package. Rand reportedly showed the UPS image to his six-year-old daughter, who answered, “It’s a present, Daddy.” He regarded that as confirmation that the mark communicated.

His work for NeXT was, for Kare, evidence of how strongly he trusted developed judgment. Jobs initially proposed a contest: five designers or firms would receive $15,000 each, and the company would continue with the winner. Rand refused that arrangement. He said NeXT would pay him $100,000, he would create one logo, and they would like it. Kare says that is what happened. The story is not an argument that every design decision should be made by one person without alternatives; Kare’s own process argues the opposite. It shows Rand’s conviction that a client was paying for a coherent point of view, not merely for a menu of visual options.

Daniel Nordh’s instruction, “Do the drawing,” supplies the counterweight to prolonged discussion. Kare takes it to mean: make an idea visual. Instead of leaving a proposed interaction, symbol, or layout in verbal debate, create something that can be seen and judged. In icon work, a sketch makes the question concrete: what will someone infer from this particular image at the scale where it will be used?

Bud Tribble’s version of Alan Kay’s user-interface principle—“Simple things should be simple. Complex things should be possible.”—gave Kare a way to think about product layers. Not every function needs to be equally prominent. A system can make ordinary actions direct while preserving more complex possibilities beneath them. The principle does not equate a good interface with one that has few capabilities; it distinguishes between access to common functions and the availability of uncommon ones.

Alan Kay also told Kare, “You could make those fonts because you didn’t know you couldn’t.” She took pleasure in the line because it captured what happened with proportionally spaced Macintosh type: the team did not begin by accepting inherited limits about what a small computer display could do.

Jobs’s advice about users belongs to the same design problem. He warned against telling people something was easy: if users fail at something described as easy, they feel dumb; if they are told it is hard and succeed, they feel smart. That observation shaped how Kare remembers the team showing the Macintosh to people and paying attention to what they could actually figure out. Friendliness was not simply a matter of cheerful graphics. It depended on whether the product gave people a chance to understand and succeed without being patronized.