Claim (frozen)
Temporary high-agency outsiders function as high-signal sensors for civic digital systems; the friction they encounter isolates design assumptions about local knowledge, sequencing, and belonging that permanent-resident and low-literacy studies typically confound.
Why I Do This
Chinese American, Greenfield (Indiana) ... library and eggs, mom’s daily drive to SRES (St. Richard’s Episcopal School, class of 2019) then Cathedral High School (class of 2023) in Indianapolis, including four years on iSquad, robotics, CHS Live, Student Council, Freshman Experience, tennis coaching, Stop Asian Hate organising, a coffee-shop shift lead, high-school pandemic POS (age 14 or 15; IU is class of 2027), Madrid systems friction and Serve IT (Laurie McRobbie Serve IT Nonprofit Clinic) handoffs, AI as recognition, Handshake (US campus jobs platform) 150K+, and paying attention as the real engineering work. Full arc on About.
Why temporary outsiders
Permanent residents and long-time users learn workarounds. Temporary high-agency outsiders ... students abroad, new arrivals who still have energy to push through a form ... hit the design assumptions while the stakes are real and the local script is not yet internalised. That friction is a signal, not a personal failure.
Supporting patterns (Madrid · public-safe)
Drawn from lived use of civic, university, commercial, and transit/utility systems in Madrid. These suggest failure modes; they do not prove causation beyond this observer’s semester.
- Assumed local context ... Forms expect address history, bank relationships, or national ID workflows a newcomer has never used.
- Unwritten sequencing ... The flow works only if step B happened before step A, but the interface does not say so.
- Language as access control ... Spanish-only defaults; “English support” that is a PDF rather than a completable interface.
- Hours and channels that don’t match outsider life ... In-person only, narrow windows, little or no status tracking after submit.
- Error messages that blame the user ... “Invalid document” instead of naming the fix.
Same failure modes show up in ServeIT clinic work: tools built for the maintainer’s mental model, not the visitor’s first visit.
What this does not claim
- Not a controlled experiment, multi-city sample, or statistical generalizability.
- Not invented sample sizes, % improvements, or pilot “results.”
- Not legal or policy expertise ... EU coursework supplied vocabulary only.
- Not clinical or health-system research. Public-safe notes only.
Working notebook: madrid-ai-ethics (MIT · personal research). Longer PDF writeup: When websites assume you belong. Practice bridge: ServeIT / Serve-AI.