What ethical questions belong in an architecture review, and who should raise them?
Show the full answer Hide the answer
What is being tested
Whether you can turn an abstract concern into concrete review questions, and whether you accept that raising them is part of the role.
The questions
"What are we collecting that we do not need?" Every field collected can be breached, subpoenaed, sold or misused. Minimisation is both a security control and an ethical position, and it is decided in the data model — where it is cheap — rather than afterwards, where it is not.
"Who is harmed if this is wrong, and can they tell?" A ranking, pricing, matching or moderation system affects people's opportunities. Whether its decisions are visible, explicable and challengeable are architectural properties, decided at design time.
"Where does the cost of an error fall?" A false decline inconveniences a customer; a false approval may cost them money. Systems should be designed so errors fall where they are recoverable, and that is a design choice.
"Who cannot use this?" Client architecture, latency budgets and payload sizes determine whether people on slow connections, older devices or assistive technology can use the product. That is usually an oversight rather than a decision, and naming it converts it into one.
"How long do we keep it, and why that long?" Retention forever, because nobody set a policy, is a choice with consequences for real people.
"What is the worst use this could be put to" — by us, by a future owner, or by someone who compromises it?
Who raises them
The architect, because they are frequently the last person who sees the whole system and the first who could raise a concern cheaply. Raised early it is a design conversation; raised after launch it is an incident.
None of these questions requires a philosophy background. They require asking, and asking them routinely — as part of the review checklist rather than as an intervention — is what makes it normal rather than confrontational.
When the concern is not accepted
Record it. Not defensive paperwork — it is how an organisation's decisions remain visible to itself, and it is the mechanism by which a concern can be revisited when circumstances change.