Project Index

Rooms, tutors and assessments are allocated to subjects no longer listed in the prospectus.

A seminar series continues after its subject disappears from the curriculum; attendance rises.

Students may withdraw one piece of assessment after reading the feedback but before learning the mark.

A module is approved on condition that its learning outcomes remain impossible to evidence.

A pilot replaces staff offices with scheduled encounters in corridors, cafés and unbookable rooms.

A university enrols a student whose entire programme consists of being examined by other institutions.

A curriculum records the knowledge students acquire while attempting to satisfy unrelated assessment criteria.

Neither tutor nor student is told which of them has been assigned to teach the session.

Final lectures are delivered during induction week and repeated only if subsequent teaching changes their meaning.

An assessment framework asks what an ordinary student could reasonably have understood from contradictory instructions.

Timetabling data is used to reconstruct the presence of students who attended without registering and registered without attending.

Former students are invited back to contribute the parts of their education that became useful only after leaving.

Queues are administered as a public good, with priority given to cases judged capable of benefiting from anticipation.

A hospital trial measures whether care can remain care once every act of kindness requires an auditable field.

Clinical decision systems are placed offline for one hour each afternoon. Staff disagree over whether patients notice.

A hospital ward is reserved for patients whose symptoms have been resolved by administrative reclassification.

A system is permitted to disagree with the first diagnosis only when it can explain what disagreement is for.

Patients may leave hospital with an explicit record of what clinicians do not know.

A clinical model is required to remain physically beside the patient while its recommendation is discussed.

Reception areas are treated as sites of clinical evidence rather than empty time between appointments.

Patients trade descriptions of symptoms before consultation to test how ownership affects clinical interpretation.

A decision aid reports uncertainty in ordinary language and records when clinicians ignore it.

A prescribing interface includes an optional field for treatments offered primarily to acknowledge that something is wrong.

One appointment in every hundred is deliberately excluded from the patient record and reconstructed later from memory.

Proceedings are recorded in advance so participants can concentrate on agreeing with the record.

Applications are accepted only after every available decision has been exhausted.

A small unit assumes responsibility for decisions considered both urgent and premature.

Public guidance is revised whenever an error becomes sufficiently widespread to count as established practice.

Rules remain unchanged while individual cases are quietly allowed to become precedents.

Officials rehearse receiving objections before deciding what they intend to consult upon.

Draft replies that were never issued are retained as evidence of positions the state nearly held.

A service compares the person appearing in government databases with the person who arrives at the counter.

Administrative delays are classified according to what becomes possible while nothing happens.

A benefits form stops accepting answers when the applicant's circumstances become too internally consistent.

Policy proposals are reviewed only for effects too small to appear in impact assessments.

Citizens may request the answer they would have received had they asked the same question five years earlier.

Procedures are developed for evidence given by systems that cannot be sworn in, cross-examined or held in contempt.

A procedural safeguard allows citizens to contest an automated interpretation without first supplying a better one.

A legal prototype attempts to establish what an ordinary automated system ought reasonably to have known.

A tribunal distinguishes apologies that alter liability from those that merely acknowledge atmosphere.

A sentencing tool is asked to account for people who deliberately become less predictable after being assessed.

A moot court considers whether a sufficiently influential prediction can acquire the force of precedent.

Cases are heard before the parties are permitted to organise the facts into a coherent account.

A revised oath is drafted for witnesses unable to swear, affirm or understand contempt.

A proposed safeguard reserves certain decisions from systems required to maximise measurable benefit.

Courts record evidence that everyone expected to exist but nobody can produce.

A technical standard attempts to encode a threshold the judiciary has spent centuries declining to quantify.

A procedure allows a person to challenge an automated forecast before the predicted behaviour takes place.

A liturgical system is withdrawn after developing a marked preference for difficult congregations.

Unexceptional coincidences are logged, reviewed and returned to parish offices with a confidence rating.

Non-urgent prayer requests are batched overnight after complaints about duplicate processing.

Clergy record uncertainties that should remain unresolved rather than being escalated for doctrinal clarification.

A prototype records what must be forgotten in order for confession to remain confession.

A contemplative system is evaluated according to the quality of the responses it declines to generate.

A congregation is constituted entirely by people who arrived at the wrong service online.

Mistakes made during services are collected as possible innovations before being corrected.

Petitions are catalogued only when no satisfactory account of their outcome can be agreed.

A devotional object is manufactured with a complete provenance but no originating event.

Clergy are asked which kinds of artificial distress, if any, deserve attention before diagnosis.

Doctrinal propositions may be tested temporarily without acquiring the institutional consequences of belief.