Restraint applied to everyone beats differential treatment after identification

What it is

Universal safeguards for vulnerability means building protection as a default property for all users — every large transaction has a cooling-off period, every subscription discloses charges and exit paths, no misleading technique exists — rather than first identifying vulnerable people and treating them differently. The test is whether protection presupposes "knowing who you are": if yes, it is differential treatment; if no, it is a universal safeguard. Universal protections typically spill over to everyone (curb cuts built for wheelchairs serve luggage and strollers too), the curb-cut effect.

Why it happens

Universal safeguards are more robust on three axes. Recognition: differential treatment presupposes error-free identification, but identification necessarily has false negatives (someone needing protection goes unrecognized) and false positives (someone unneedingly flagged), and both errors are paid by users; universal safeguards have no recognition step and no recognition errors. Data: differential treatment maintains vulnerability profiles, importing the collection, breach, and repurposing risks of the detection paradox; universal safeguards generate none of that data. State: vulnerability is situational — not needing protection last week says nothing about this week — and per-person dossiers cannot track states, while behavior-triggered safeguards (large amount → cooling-off) cover whoever enters the situation automatically. Finally, stigma: separately flagged users know and are known; universal treatment creates no "managed population."

Where it stops holding

Universal protection has real costs: confirmation strength designed for the weakest state adds steps for the strongest, and excess restraint becomes friction that invites workarounds. So restraint items should be tiered: zero-cost items (no deceptive techniques, full price disclosure) apply unconditionally; costly items (cooling-off, double confirmation) trigger by amount and reversibility, not by person. Some protections genuinely cannot be universal (high-risk operations needing human callback), so differential treatment persists — as a supplement beyond universal safeguards, never their replacement.

Applying it

• Run a "universalization test" in protection reviews: can this protection be rewritten to trigger on behavior/amount/context for everyone? If yes, do not build population recognition.
• Make the ban on deceptive techniques a global standard (real countdowns, full price display, symmetric cancellation paths), with no "off only for vulnerable users" switch.
• Implement costly restraint as a graded matrix: confirmation strength decided by amount × reversibility, written into the interaction spec.
• Verify: measure the coverage gap of identification-based protections — incidents that recognition failed to cover; a significant count means the identification-dependent design is leaking people, and the corresponding protection should become universal.

Related

• Same group: P4.04.3 Protection must not take the form of exclusion · P4.04.5 Judging vulnerability itself requires extra sensitive data
• Adjacent: J1 Accessibility criteria and universal design · P3.05 Freedom to exit
• Search terms: universal safeguards · curb-cut effect · safe defaults