Chapter 15. Feedback: The World Responds Behind the Button

In Chapter 14, we brought people to their first action through a familiar gesture. Now it is the world's turn to answer that action.

When planners hear the word feedback, they usually picture visual effects: light bursting when a button is pressed, numbers rising when an enemy is struck, and the screen flashing after a success. They assume that more spectacular feedback is better feedback. That view sees only half of what feedback is.

Feedback is not an effect but an answer. When a person does something, the world replies, "Yes, I heard you." Whether that answer is flashy or restrained comes later. First comes the question of whether an answer exists at all. If someone presses a button and nothing happens, they begin by doubting whether they actually pressed it. If that doubt arises on the very first action, there will be no second action. When the world does not respond to a pressed button, what occurred was merely input, not experience.

That is why the most frightening thing on a first screen is not a bad answer but silence. A bad answer can be fixed. Silence keeps people from knowing even what they themselves did. An unresponsive world makes a person invisible.

Two Silences

Let us enter a fictional character game. A person taps a character on the head. Two kinds of silence can drive that person away.

The first silence happens immediately. A finger touches the screen, but nothing changes. The character stays still, and the button shows no sign of being pressed. The person waits for a very brief moment, then presses again, but it is still quiet. By the third press, they conclude that the screen has frozen or that they are pressing it incorrectly. In fact, the game has been working perfectly well. It simply failed to return the fact that the press had occurred. The input arrived, but there was no signal saying that it had arrived.

The second silence comes a little later. This time, touching the screen produces a pressed state on the button, and the character flinches slightly. But the flinch is where it ends. Nothing else happens, so the person thinks, "So what?" They know the press registered, but not what it meant: whether the character liked or disliked it, what they themselves did, or whether they did it well. They know the input arrived but not what meaning that input created. The first silence makes a person stop. The second makes a person leave. They are different silences and demand different answers.

In MEJE IDONG WORLD, when a fan finishes decorating a puppy IDONG and applies the final color, two things happen together. The touched area fills with color at once, and IDONG hops and laughs. The color appearing is the answer "Your hand reached me." IDONG's laughter is the answer "Now your IDONG exists." The first is a signal that the hand made contact; the second is a signal that meaning was created. The fan receives both answers and understands that what they did became something within the world.

Here, "together" does not mean that the two answers are mixed into one. It means they follow one another closely within a single flow. The answer that confirms the input comes first as the color appears. IDONG's laughter then returns the meaning of that action.

The Input Arrived, and Meaning Was Created

If feedback is left as a vague notion of "response," it becomes unclear what must be returned and when. We therefore divide feedback into three layers. Whenever a person acts, the world answers through these three layers.

The first layer is a physical answer: the exact place touched responds immediately. A button depresses slightly, a ripple spreads from the point of contact, or a small sound plays. This is the lowest layer of the answer: "Your finger touched here." It does not need meaning. It only needs to return the fact of contact immediately.

The second layer is the rule's answer. It tells the person how the action was processed within the game's rules. If pressing a button selected a character, an outline appears around that character. If the person applied a color, that color appears. This layer says, "Your input was reflected as this result." Here, the person learns that their action was not empty but became an event inside the game. The second layer also has a silence of its own: knowing that something was pressed but not how it was processed. If someone selects a character but the selection indicator is faint, they may choose the wrong one and continue without realizing it. Only much later do they discover that it was not the character they meant to choose. This silence of misselection does not stop the person like the first silence or make them leave immediately like the third, but it erodes trust later.

The third layer is the world's answer. The world shows what that action meant within it. The character smiles with delight, a new character becomes mine, or a small celebration follows. This layer says, "What you did meant this in this world." Almost every reason to remain emerges from this third layer.

Here we reach this chapter's central distinction. "The input arrived" and "meaning was created" are different events. The first two layers answer that the input arrived; the third answers that meaning was created. Many first screens stop after giving only the first layer. Their buttons press perfectly well, but they do not show what meaning the press produced. A person may know they are manipulating a screen without feeling that they are doing something within a world. Conversely, if a screen delivers only a grand third-layer answer and omits the first, people may already leave during the short silence immediately after touching it, never seeing the grand answer at all. The three layers must therefore arrive in order: first return that the touch was received, then that it was reflected, and finally what it meant.

For a tool-like experience, two clear initial layers can be enough to bring people back. The third layer matters especially in experiences that attach meaning to actions and build affection through a world, characters, or ownership.

Answers Have an Order

The timing of an answer is also a design choice. Feedback is often said to require an immediate response the moment something is pressed, with faster always assumed to be better. That is true of the first-layer answer confirming contact: it must return at the moment of touch. But if every answer, including the one that conveys meaning, explodes at the same instant, it becomes unclear which result came from which cause. If the answer that a button was pressed, the answer that a selection was applied, and the answer through which the world returns meaning are all mixed into one burst of fireworks, people miss what they did and what changed because of it.

The first and third answers must therefore divide their roles. The first answer, confirming contact, comes immediately. Here, "immediately" means within the limit at which a person experiences the response as directly attached to their action: a very brief moment of roughly one tenth of a second. When the answer returns within that limit, pressing and response feel like one event: "I pressed it, and it reacted." When it arrives later, pressing and response feel disconnected. The third answer, which says that meaning was created, follows one beat later in a form the user can connect to the action just taken. This beat is not a long wait. It is not a pause intended to make the user wait, but an ordering that reveals, "I did this, so that happened." If the answer comes too late, the person receives a result after forgetting their action. If every answer arrives at once, the action and result become tangled.

This matters in a first experience because the force that keeps people there comes from discovering causality. The small realizations that "the world responds when I press" and "what I do produces a result" make a person press once more. Feedback design is not the work of making a screen spectacular. It is the work of making that causality unmistakable.

Apps outside games craft these answers with equal care. In Chapter 14, we visited the language-learning app Duolingo as an example of postponing the sign-up threshold. Look at it again, this time for the texture of its answers. The moment a person chooses the correct answer, it returns a cheerful sound, fills one segment of a bar, and raises the learning streak. The answer to contact, the answer that it was correct, and the meaningful answer that "you continued again today" arrive in sequence with short spaces between them. (Apps change frequently, so the details of the presentation vary over time.) Because these small answers are clear, people feel that their work accumulates somewhere each time they solve a problem. It is not spectacle but clear causality that makes them solve one more.

Damage Numbers Are Noise to Some People

The convention to examine in this chapter is the habit of making feedback bigger without qualification. Strike an enemy and red numbers rise over its head. Land consecutive hits and a combo indicator grows in the corner. Collide with force and the screen shakes. Such spectacular feedback has long served as a device for pleasure in games.

This convention arose from the design of action games that value impact and satisfying controls. To players, rising damage numbers are both information and pleasure. They can read at a glance how powerful their strike was, and as the combo builds, they thrill at seeing their skill inscribed on the screen. When the screen shakes, they feel the weight of the impact in their bodies. For experienced players, this excess is not excessive but rewarding. They read the spectacle through context.

Ordinary people do not share this agreement. If they merely tap a character and numbers shoot above its head while the screen shakes, they become startled and wonder whether they pressed something wrong. What they want to read is, "Did my puppy like that?" But when the screen is covered with numbers, explosions, and shaking, the answer they actually need is buried under spectacle. A player reads a rising number as "the magnitude of what I did." An ordinary person reads it as "noise I do not understand." The same effect is pleasure to one and overload to the other. The spectacle is not meaningless. Its meaning exists only for people who know the agreement needed to read it; without that agreement, it is simply loud.

Even games made for players sometimes let enlarged feedback obscure what actually needs to be seen. The Diablo action RPG series prides itself on the satisfying feel of extravagant effects. Yet in battles with many participants, players have long complained that overlapping skill effects, explosions, and showers of loot make enemies and items impossible to see. (The severity varies by installment and patch.) Even experienced players who know how to read this spectacle as pleasure say that it is too chaotic to see through. For ordinary people who do not know that agreement at all, the same screen is noise from the very first action.

The cost this convention imposes on ordinary people is consumed attention and lost meaning. When a screen shouts too many things at once, a person cannot separate which one is the answer to their action. Too many answers begin to resemble no answer at all. Keep only the essence. What feedback was meant to convey is "What you did reached the world," not the spectacle of filling a screen. In a first experience, do not enlarge the answer; clarify it. One action, one answer. Make what the person did and what it meant the most visible things on the screen, and remove effects that obscure them. Add spectacle slowly after the person has learned the language of the world, when it can be read as pleasure. On the first screen, the thing that shines most strongly should not be the most spectacular effect, but the meaning of what the person just did. Condensed into one sentence: feedback on a first screen is information design, not reward design. Pleasure comes after meaning can be read.

▶ Three Questions to Apply to My Screen

Does an answer that the first action made contact—a pressed state, ripple, or sound—return at once?

Is there a separate answer after it that tells the person what the action meant in this world?

Is there one clear answer to one action, or do three or four effects overlap and conceal which one is the answer? When the first fails, people stop. When the second fails, they ask, "So what?" and leave. When the third fails, the answer is buried in noise.

So, Decide Which Answer Comes First

Starting from a list of effects is already too late. Feedback design begins by deciding how the world will answer each of a person's first actions. Is there an answer that the first action made contact? Does an answer that meaning was created follow it? Is there a tiny interval between the two in which a person can feel causality? Once these three points are settled, the form of the effects can be chosen as long as it does not obscure the answers. Choose effects first and fit answers into them, and the screen becomes spectacular while meaning is buried.

This decision returns to measurement. Whether the first action received a proper answer can be seen in whether people repeatedly tap the same place. If people often press once, receive no answer, and press two or three more times, the first-layer answer is missing. Evaluate the second layer through the frequency of immediate cancellation and undo. If people often undo and choose again right after choosing, the answer showing how the input was processed is unclear, and mistaken selections are occurring. Also observe the rate at which a first action leads to a next action. If many people press once and stop, they acted but did not receive the answer that meaning was created, leaving them stranded at "So what?" Whether the answer was made bigger or clearer is not a matter of intuition alone. It returns through whether the person presses once more.

Notice also that these silences do not carry the same kind of risk. If a button does not press, QA catches it and reviews record it in star ratings, so first-layer silence is reported as a bug. But a screen that can be pressed yet has no meaning leaves no trace in crash logs or customer-support tickets. Users depart in silence instead of complaining. Third-layer silence is most dangerous not because it is most expensive, but because it is recorded nowhere. That is why the rate of progression to a next action must be an essential gauge on the first screen. The measurement chapters beginning with Chapter 22 will take up this gauge in earnest.

Reference Content

Examples from other media that reveal this chapter's concepts.

Video Games

Group combat screens in the Diablo series: Players have often complained throughout the series that when many people fight together, overlapping skill effects, explosions, and showers of loot make it impossible to see the enemies or the items they want to pick up. It typifies "too many answers begin to resemble no answer at all": so many answers obscure which one belongs to my action.
Super Mario's jump and coin sounds: A brief effect that rises in pitch with a jump returns the action immediately upon contact, while collecting a coin produces an instant bright note. With only five limited audio channels, the composer attached a clear auditory answer to every important action, creating a classic example in which people immediately confirm through sound what they have done.
Candy Crush's bursting blocks and praise voice: The answer that contact occurred and the meaningful answer that the person "did well" return in sequence with a short interval. This recurring subject of examination in the book also appeared in Chapter 13, where we separated and read the signified behind its praise. General Apps
Duolingo's correct-answer sound and streak indicator: The answer to contact, the answer that it was correct, and the meaningful answer that "you continued again today" arrive in sequence with short intervals.
Indeterminate wait indicators such as spinners and skeleton screens: A rotating indicator or a dim placeholder that reserves the space where content will appear makes the answer that something is progressing visible. People can wait without becoming anxious at silence only when they can see an answer, inversely proving the principle that whether an answer exists comes first. The boundary drawn in Chapter 13 also applies here. Such indicators honestly announce that something is alive without claiming to measure anything. A fake progress bar that invents a percentage unrelated to actual progress falsifies measurement. Give a signal that an answer exists, but do not fabricate measurement.

The remaining reference content appears in the "Chapter 15 Appendix" at the end of this chapter. (In the print edition, it is collected in Appendix D.)

Design Notes ▶ Try It Yourself

Choose five actions that a person will perform during their first experience with our game: the first button, the first choice, the first act of making, and the other actions their hand will touch earliest.

Write three lines for each action. What is the answer that the hand made contact? What is the answer that it was reflected within the rules? What is the answer that tells what it meant in this world? If any of the three lines is blank, silence occurs there. When the first layer is blank, the person stops. When the second is blank, the person selects the wrong thing without knowing it. When the third is blank, the person leaves.

After writing them all, mark the order in which the answers arrive beside each action. The answer to contact comes first; the answer that meaning was created follows just slightly later. Check that they do not erupt at the same instant. Finally, count whether every action has one clear answer. If three or four effects overlap on a single action, see whether all but the one that conveys the meaning best can be removed. The standard for choosing the one to keep is simple: with only that effect remaining, a newcomer should be able to put into words what they did and what it meant. Precise answer design for each input continues in the three-layer feedback sheet in Appendix C.

In One Line: Feedback is not a spectacular effect but the world's answer. When a person acts, the world answers in three layers: the physical answer that the hand made contact, the answer that it was reflected in the rules, and the answer that tells what it meant in this world. The first two signal that the input arrived; the third signals that meaning was created. They are different events. Excess such as damage numbers, combos, and screen shake is pleasure to players but noise to ordinary people. Do not enlarge the answer; clarify it, so the meaning of what the person just did is the most visible thing on the screen. Next Chapter: Until now, we assumed that a person was quietly concentrating on the screen. But people look at screens on a shaking bus, with one hand, amid noise. Even a clear answer is useless if the person is not in a state to see it. Next, we turn to what should shine through distraction and noise.

Chapter 15 Appendix: Reference Content Collection

The examples collected here did not fit in the main text. They are grouped under headings by medium, from video games to music, theater, and everyday machines. Each entry ends with a "What to Observe" note explaining what a designer should examine. Keep the framework established in the main text in mind: the three layers of the physical answer that the hand made contact, the answer that the input was reflected in the rules, and the answer that meaning was created; and the distinction between the silence that omits the first answer and stops a person, and the silence that omits meaning and drives a person away. You will then see which layer of an answer arrives in what order in each case and where silence arises. Read with an eye for separating examples that make answers bigger from those that make them clearer.

Video Games

Screen shake in Nuclear Throne: Every hit shakes the screen and sends enemies flying, amplifying the feel of impact. Developer Vlambeer is known to have organized and shared in a talk its techniques for building game feel by layering exaggerations such as shake, brief pauses, and knockback. What to Observe: Consider this alongside the main text's caveat that exaggeration is not automatically a good answer. The same shake is pleasure to a player who knows the agreement but noise to a newcomer. Ask what intensity is appropriate for whom.
Quick-time events (QTEs) that suddenly appear during cutscenes: A person thinks they are passively watching a video when a prompt such as "Press X" suddenly appears. If the screen does not respond immediately to the press, they doubt whether the input registered and press the same button repeatedly. The silence left by a missing first-layer answer leads directly to failure. What to Observe: Watch whether people respond to the very brief silence after an input by button mashing. Repeated presses are the most honest signal that the first-layer answer is missing.
Hit stop in fighting games: At the instant an attack lands, both characters freeze very briefly before moving again, letting a person physically feel that the hit connected. Stronger attacks stop for longer, conveying the weight of the answer as well. This makes an answer clear with a beat of stillness rather than enlarging it with spectacle. What to Observe: Notice that answers have weight, not just speed. The scale of the answer is taught through the duration of the pause without adding anything to the screen, clearly revealing the difference between making something bigger and making it clearer.
Character-by-character speech sounds in the Animal Crossing series: A short vocal sound plays as each character appears in a dialogue box, creating the sense that the character is actually speaking. It is said to have become so central to the series' identity that fans call it "Animalese." What to Observe: Follow the path by which a low-level answer that sonifies progress itself also supports the third-layer meaning of a character's vitality. Even a modest answer becomes communion when it does not break.
The puppy-petting response in Nintendogs (2005): Touch a puppy with the stylus on Nintendo DS, and it immediately reacts toward that spot and makes a pleased sound. The answer to contact and the meaningful answer that "my puppy liked it" arrive together, creating a model of the third layer in which a person feels genuine communion with life inside a screen. What to Observe: See how the answer to contact and the answer of meaning follow one another within a single action. Because the response begins exactly where the touch landed, the causality stays clear.
Tetris Effect and Lumines: Every action of placing blocks and clearing lines interlocks with music, vibration, and visuals, making the entire world respond together. Each action returns as a musical answer, expanding the third layer until a person feels their hand and the world breathing on the same beat. What to Observe: Look beyond the answer to one action and see how far the third layer can expand when the whole world breathes to the user's rhythm. The many answers do not become noise because every one is aligned to a single beat.
Hit markers and elimination sounds in FPS games: When a distant enemy is hit, a brief cross appears over the reticle and a tick sounds. Eliminating the enemy produces a second, different sound. At distances where the target is too small to tell from the image alone whether a shot landed, dedicated signals separately return the fact of a hit and the fact of a kill. What to Observe: Treat this as an example of separating the second-layer answer—reflection in the rules—into an independent signal. Examine which signal takes over when the on-screen performance cannot convey that answer.
Spreading cracks while breaking a block in Minecraft: As a block is mined, cracks spread gradually across its surface, maintaining an uninterrupted answer that the long action is still in progress. What to Observe: Notice that an answer must continue even for an action whose result has not yet arrived. The longer the gap between the first answer and the final result, the more an intermediate answer must fill the silence.

Live-Action Film

Sound feedback in ASMR and mukbang: Tapping, rustling, and chewing sounds themselves return a sensory answer that something was touched. A clear answer holds attention even without spectacle. What to Observe: This shows the main text's proposition that clarity alone can constitute an answer, in its purest form. Use it to practice separating the intensity of an answer from its clarity.

Music

Call and response: One side sings and the other immediately answers, demonstrating the causality of an immediate reply to an input through sound. What to Observe: Because the interval before the answer is promised in units of beats, the body can feel what timing allows a person to perceive causality.
Buildup and drop in dance music: Tension accumulates, one beat is left empty, and then the bass drops and releases it. Song structure itself shows that an answer must arrive in the order of building, emptying, and dropping rather than erupting all at once for people to feel causality and pleasure. A weak buildup also weakens the drop's answer. What to Observe: Pair this with the section arguing that the moment when an answer arrives is itself a design target. The short emptiness before the drop does the same work as the interval that clarifies causality.

Live-Action Television / Drama

Correct- and incorrect-answer sounds on quiz shows: A sound answers immediately when a button is pressed, and the meaningful answer that it was correct follows. The order of the two answers is clear. What to Observe: Use this as the simplest model of an answer confirming the press followed by an answer confirming correctness. Imagine what disappears if the two answers are merged into one sound, and the value of their order becomes apparent.
Reaction shots in sitcoms, such as looking at the camera in The Office: One beat after an absurd event, a character briefly glances at the camera. The answer "Can you believe this?" attaches to the event at the precise moment that releases laughter. What to Observe: See how editing timing separates an immediate answer from a meaningful answer one beat later. The same scene ceases to be funny if the reaction shot comes too early or too late, confirming that the timing of an answer matters as much as its content.

Board Games

Immediate reactions in Jenga and Halli Galli: The moment the blocks collapse or the bell rings is the answer itself. Action and result are attached as one event. What to Observe: The original form of the answer that digital feedback imitates exists in the physical world. Use the feeling of action and result forming one event as a reference point, then measure how far apart they have become on your screen.

Theater / Performance

Applause and laughter from the audience: The audience answers a performer's action immediately. If the answer is late or absent, the person onstage also loses their rhythm. What to Observe: Silence destabilizes not only the user but also the rhythm of the maker. A performer before a silent audience is a designer before a screen without metrics.

General Apps

Transfer-complete animations in Toss and Kakao: A small animation clearly returns the fact that the money was sent, letting a person know at a glance that it is done. What to Observe: The greater the anxiety around an action, as with money, the clearer the meaningful answer must be. If the answer is vague, people reopen the transfer history to check it. That act of reconfirmation is itself an indicator that the third layer is weak.
Smartphone haptics: A single vibration at the fingertip immediately returns the lowest-layer answer that the screen was pressed. What to Observe: When one lowest-layer answer underlies every screen, examine how much total silence disappears. This is a starting point for designing system-level and screen-level answers separately.
KakaoTalk's unread marker "1": After a message is sent, a number beside the bubble says that the recipient has not yet read it. The number disappears when they do. The answer that it was sent and the answer that it was read arrive separately. What to Observe: Treat this as a structure separating the second and third layers. Making the state of "read but unanswered" visible created a new kind of silence, showing that a design which reveals answers can also create new emotions.
Instagram's double-tap Like: Double-tap a photo and a large heart rises and disappears at the point of contact while the Like count increases. What to Observe: Notice that the answer begins exactly where the hand touched. The farther the answer appears from the point of contact, the weaker the causality becomes.

Machines / Appliance UX

The pressed light and beep of elevator and microwave buttons: These immediately answer that the input arrived, keeping people from pressing the same button again. What to Observe: Repeatedly pressing the same button is the cheapest indicator of first-layer silence. The sight of people mashing a door-close button is a landscape of silence.
The click of a car turn signal: Sound and light answer simultaneously that the signal is on, so the driver knows the state without looking at a screen. The sound originally came from a relay component. Even after cars moved to electronic components, many are said to deliberately replay it through a speaker. What to Observe: Look at the decision to preserve a sound specifically as an answer after its functional necessity disappeared. Here it becomes clear that an answer is not a by-product but a design target.
Haptics for smartphone setting wheels and toggles: Turning a date or time wheel produces a tiny tap at the fingertip with each step; switching on a toggle produces a click-like vibration. A single vibration imitates the detent of a physical dial, returning through the fingertip the first-layer answer a screen alone would lack. What to Observe: See how a screen restores the first-layer answer once supplied by a physical dial. Fine-grained answers at each step fill the silence of a continuous action.
Locking and unlocking melodies on digital door locks: Different melodies play when the door locks and unlocks, so the fact that it locked reaches even a person who has turned away and is walking off. What to Observe: Examine a channel that delivers an answer to someone who is looking at neither a screen nor a light. Assigning distinct sounds to two states reveals not only that an answer exists but also what state resulted.