This loop must be documented, supported and protected. It must not be treated as error and it must be recognised as civic expertise.
Protocols must be clear. Staff need to know when and how to exercise discretion. This includes override pathways, ethical guidelines, and reflective documentation. The author proposes a custodian’s register, a log where discretionary acts are recorded with reason and reflection. Not to monitor staff, but to honour judgement.
Training is essential. Staff must be equipped to recognise when systems fail, when context matters and when care requires departure from the default. This is not soft skill, it is civic skill. The author calls for discretion literacy; a curriculum that teaches presence, judgement and answerability.
Culture must shift. Institutions must move from compliance to care. Discretion must be valued, not punished and Staff must be trusted to hold complexity, not forced to flatten it. One notes that many acts of repair begin with a refusal to relay. Designing for discretion means making that refusal possible.
Finally, discretion must be relational and it must include the citizen. Overrides should reflect not just staff judgement, but citizen testimony. One proposes dialogic overrides; where the record includes both voices, both reflections and the shared threshold of decision.
Closing this section with a civic truth: discretion is not deviation, it is design. It is what allows systems to meet people, not just process them.
Repair Action: Students could submit appeals with narrative statements, not just forms. Teachers were trained to write contextual testimonies alongside predicted grades. A cross-disciplinary panel, including a student representative to review each case.
Impact: Appeals became more transparent and dialogic. Staff reported a renewed sense of professional agency. The process was adopted by other colleges in the region.
Why it matters: Repair was institutional, it rebalanced power, restored trust and made the record speak.
Designing for Discretion One proposes that discretion is not a flaw in the system, it is the system’s capacity to respond. In a civic field shaped by automation, discretion must be designed, not hidden, penalised or improvised.
Discretion allows staff to override, adjust, or reject automated decisions based on context, care and judgment. It is what makes systems livable. Yet in many UK institutions, discretion is discouraged, Staff are trained to relay, not to reflect. Overrides are undocumented and Appeals are procedural. The author insists: this is not efficiency, it is abandonment.
Designing for discretion begins with visibility. Every system must include a human loop; a point where staff can intervene.
