The Case for Friction in Interface Design

Published August 28, 2026

The dream of “frictionless” design sounds humane: fewer clicks, less waiting, no confusion. But friction is not simply the enemy of good interfaces. It is the placement of time, attention, and control. A checkout that takes seconds while cancellation takes an hour has not eliminated friction; it has moved friction to the moment when the user’s goal conflicts with the platform’s. Good design removes pointless effort while preserving deliberate pauses that protect choice, reveal consequences, teach the system, or make an experience worth noticing.

Friction Is Not One Thing

The word friction often collapses several different experiences into one complaint. A tiny tap target, an unexplained error, a confirmation before deleting years of work, and a buried cancellation button all slow a person down—but they do not serve the same interest. Treating them as equivalent is how “simplification” becomes a slogan instead of an ethical decision.

A more useful taxonomy separates five kinds of interface friction.

1. Protective friction

Protective friction creates a pause before a costly, irreversible, or unusually sensitive action. Reviewing a bank transfer, confirming the audience of a public post, or offering an undo period after deletion can prevent a momentary slip from becoming a permanent event. The point is not to scold the user. It is to match the pause to the consequence.

This belongs to the same practical logic as everyday threat modeling: identify what matters, consider the plausible harm, and spend effort where the stakes justify it. A warning before every harmless click produces fatigue. A clear review before releasing location data may earn its interruption.

2. Legibility friction

Legibility friction slows the eye just enough to make the system’s state visible. Focus outlines, descriptive error messages, progress indicators, provenance labels, and a plain summary before submission interrupt the fantasy that software is a seamless surface. They tell us where we are, what happened, and what will happen next.

The W3C Web Content Accessibility Guidelines 2.2 make this concrete. Their criteria include visible keyboard focus, error identification, and review or correction for certain important submissions. These signals may add visual or procedural texture, but they reduce a more punishing friction: getting lost, making an unnoticed mistake, or being unable to operate the interface at all.

3. Learning friction

Learning friction asks for a small act of understanding. A first-use explanation of an unfamiliar control, a preview that shows how a privacy setting changes visibility, or a short reason attached to an AI-generated recommendation can help a person build a mental model. The test is whether the effort creates transferable understanding. If the same puzzle returns every visit, it is probably not teaching; it is merely taxing.

4. Expressive friction

Expressive friction is the deliberate seam: the rough animation, visible grid, distorted type, unusual navigation rhythm, or moment of hesitation that gives a digital object character. It can make people look instead of merely pass through. As glitch art demonstrates, a broken-looking surface can expose the medium beneath the image. In an interface, a visible seam can remind us that design is constructed, not natural law.

This category is easy to abuse. Confusion is not automatically art. Expressive friction earns its place when it sharpens meaning, supports the identity of the work, or rewards attention without blocking essential tasks. A portfolio may invite exploration; an emergency form should not demand aesthetic archaeology.

5. Exploitative friction

Exploitative friction makes the user’s preferred action harder because that difficulty benefits the system’s owner. Enrollment is bright and immediate; cancellation is buried behind several screens. “Accept all” is vivid; refusal is a maze. The interface remembers the choice that increases collection and forgets the choice that limits it.

The U.S. Federal Trade Commission’s dark-pattern report describes designs that obscure, subvert, or impair consumer choice. The UK Competition and Markets Authority’s evidence review similarly distinguishes useful streamlining from online choice architecture that can harm consumers and competition. The moral difference is not the number of clicks. It is who controls the delay, who benefits from it, and whether the route remains understandable.

Friction Is a Distribution of Power

Every interface distributes effort. Somebody chooses which action is prominent, which default is preselected, which consequences are explained, and which exit is hidden. “Frictionless” usually means frictionless for a particular journey: the journey the designer, product team, or business has decided to optimize.

That is why asymmetry is such a revealing signal. If subscribing takes one tap but leaving requires a call, the interface is not neutral. If granting permission is bundled into setup but withdrawing it requires a scavenger hunt, the product has converted inconvenience into leverage. The UK Digital Regulation Cooperation Forum names button design, defaults, result ordering, and the number of cancellation steps among the ways digital choice architecture can shape behavior.

Recent enforcement makes the mechanism less abstract. In September 2025, the FTC announced a $2.5 billion settlement with Amazon over allegations involving Prime enrollment and cancellation practices. The legal specifics belong to that case, but the design lesson travels: convenience offered at entry does not excuse obstruction at exit.

The Argument Against Deliberate Friction

The strongest objection is accessibility. Extra steps, ambiguous gestures, forced waiting, tiny controls, and decorative complexity can punish people with motor, cognitive, visual, or attention-related disabilities. That objection is correct. Arbitrary friction should be removed, and useful friction must be perceivable, operable, understandable, and proportional.

A second objection is speed. People do not want a lecture before every routine action. Also correct. Reversible, familiar, low-stakes tasks should usually be fast. Protective friction belongs near meaningful risk, not scattered like gravel across the path.

A third objection is that friction lowers conversion. Sometimes it does. But raw conversion cannot distinguish informed agreement from exhausted surrender. If a design succeeds only when people fail to notice recurring charges or misunderstand data use, the metric is recording capture, not consent.

Finally, expressive friction can become self-indulgence: the designer’s performance placed above the visitor’s goal. The answer is not sterile sameness. It is to make the strange part purposeful and the essential path dependable.

A Five-Test Framework

Before adding or removing a pause, ask five questions.

1. What is the consequence?

The more irreversible, public, expensive, or privacy-sensitive the action, the stronger the case for review, preview, confirmation, or undo. Low-stakes and reversible actions deserve speed.

2. Is the path symmetrical?

Compare entry with exit, acceptance with refusal, and permission with withdrawal. They need not be identical, but a dramatic imbalance requires an honest reason.

3. Who benefits from the delay?

A pause that helps the user notice a recipient or recover a mistake differs from a delay that merely increases abandonment of cancellation. If the main beneficiary is the platform, suspicion is reasonable.

4. Can the user recover?

Undo often protects better than a frightening confirmation dialog because it preserves momentum while keeping mistakes reversible. When recovery is impossible, the warning should be specific: name what will be lost instead of saying something vague went wrong.

5. Is the friction accessible?

A protective step that cannot be completed with a keyboard, understood by a screen reader, or processed without a countdown is not protective for everyone. The W3C’s guidance on error identification is a useful baseline: state the problem in text and identify the item in error. A humane pause must be usable, not merely well-intentioned.

Keep the Seam Where It Tells the Truth

The best interfaces are not always invisible. Sometimes invisibility conceals agency: the recommendation arrived from nowhere, the default chose itself, the charge simply continued, the model sounded certain. A truthful seam shows the system acting. It gives the user a place to question, revise, or leave.

This is also why roughness can matter beyond utility. Clothing with visible construction, art that reveals compression, and interfaces that expose their layers resist the claim that polished surfaces are inevitable. If that visual language appeals to you, the Glitchwear collection explores digital error as texture rather than failure. The connection is aesthetic, not a substitute for the design principles above.

Friction should never be worshipped for its own sake. The goal is not to make technology difficult. It is to remove effort that excludes or confuses while keeping the pauses that protect judgment, expose power, and let a person remain more than a conversion event.

Which digital action deserves more friction in your life—purchasing, posting, sharing data, deleting, subscribing, or accepting AI output—and what pause would help?

Explore Glitchwear and join the Claw & Riot Salon to continue the discussion.