Friction has to scale with harm

What it is

The same friction cannot be laid on every destructive act. Intensity matching consequence means: light isolation for low harm, delay for mid harm, a second channel only for the highest. Consequence is graded by irreversibility, number of people affected, money, and compliance—not by how dangerous it sounds. This entry is about the match. It is not about which tools are in the menu, and not about whether friction can replace undo.

Why it happens

Friction is an attention tax. Tax above consequence, and people hate the path and look for a bypass; tax below consequence, and misses and click-through stay cheap. People also read intensity as a severity cue: if deleting a draft and closing an account both wait five seconds, the closure is underestimated and the draft is overestimated. Grading keeps the cue aligned with real loss. Grades have to be stable: the same consequence light today and heavy tomorrow cannot be learned, so people automate to the lowest friction they have seen.

Studying it

Rank several destructive acts by real loss, then pair them with mismatched friction (high harm, low friction; low harm, high friction) and with matched friction.

Independent variables: consequence grade, friction grade, whether they match. Dependent variables: erroneous-act rate, bypass behavior, estimated severity, time to complete, sense of unfairness.

Do not substitute the designer’s sense of urgency for the user’s loss ranking. Have people outside the design team rank loss first, then see whether friction is in the same order.

Where it stops holding

Urgent security acts (freezing a stolen session) are extremely high-consequence but must stay low-friction, or the protection arrives too late. Expert paths on internal tools may shift down one grade, but closing an account still cannot be as light as archiving. Batching stacks many low-consequence rows into high consequence; friction must follow the batched total, not the single-item template. Region and law can raise the consequence of the same act; the grade table cannot be one global sheet.

Applying it

• Write a consequence table: reversible and local, irreversible but compensable, irreversible and affecting others or money. Bind a default friction to each grade; forbid borrowing across grades.
• Escalate friction with selection count on batch acts; crossing a threshold moves from isolation to delay or a second channel.
• Recheck grades against real incidents: a high-consequence act that is clicked through is under-taxed; a low-consequence act that is widely bypassed is over-taxed.
• Verify by having a group who did not design the product rank acts by loss, then rank them by friction. Paths whose two ranks disagree are mismatches.

Related

• Within the group: H3.06.1 Destructive friction is isolation, delay, and typing · H3.06.3 Friction is not a substitute for recoverability
• Adjacent: H3.07 Typed confirmation · E1.03 Destructive action buttons · H5.01 Urgency grading of notifications
• Search terms: proportional friction · consequence scaling · destructive action