Obstruction: the path against the operator's interest is made harder
Aliases: asymmetric friction · obstacle pattern · friction asymmetry
What it is
Obstruction makes the thing the user wants to do harder: more steps, more confirmations, deeper menus, longer waits — while the same-level action that serves the platform stays easy. It does not block (blocking is found instantly); it applies asymmetric friction: both directions remain open, but one is paved and the other full of speed bumps.
Why it happens
Behavioural cost decides completion: every extra step sheds some users, and the shedding lands exactly on the action the platform does not want — which is the design intent. The obstruction toolbox: lengthened flows (subscribe in one tap, cancel by phone), inserted interruptions (repeated confirmations and retention offers inside the cancellation flow), interface depth (unsubscribe buried four levels into account settings), waiting and cooling-off periods. The division of labour with interference: interference biases the choice at the perception layer, obstruction punishes the choice at the cost layer — the user may know precisely where the entrance is and simply have to walk ten minutes. The symmetry test is clean: put the "enter" and "exit" flows side by side, count steps and time, and the ratio is the obstruction index; significantly above one, and in the platform's favour, it is obstruction.
Studying it
Direct evidence comes from flow-audit research: large-scale walkthroughs of mainstream services' subscribe/cancel flows recording step counts, durations, and inserted interruptions; elasticity studies measuring flow length's sensitivity effect on completion; and regulatory rule-making documents (such as the "click to cancel" proceedings) that lean heavily on exactly these audits. Methodological caution: audits must run on real paid accounts — sandbox and demo environments miss the steps only paying users ever see, and those steps are usually the core of the obstruction.
Where it stops holding
Not all friction is obstruction: confirmation steps on high-stakes actions (account deletion, transfers) serve the user's own interest, and judgement follows the direction of the friction, not its absolute amount. Cooling-off periods in security contexts (password changes taking effect after a delay) have independent justification. Flow length also has an engineering-inertia origin — legacy systems nobody has rewritten — which is indistinguishable from deliberate design by appearance alone; "unchanged for years" may itself be evidence the obstruction was tolerated, so audit records the state and classifies by direction.
Applying it
- Symmetry audits: subscribe/unsubscribe, join/leave, enable/disable — walk and screen-record each pair, count steps, inputs, and waits, and produce an asymmetry list with a remediation order.
- Set a hard rule: no "exit" flow may take more steps than its corresponding "enter" flow; the rule goes into the design-review checklist.
- Verification: re-run the audit quarterly — obstruction regrows with each release cycle; a one-time cleanup is not an endpoint, the metric needs continuous watching.