Unfinishedness: Accountability Without Total Exposure

The request often sounds reasonable. Explain what happened. Document the decision. Show the supporting messages. Describe how you felt, why you hesitated, what you intended, whether you remain committed, what else was happening at home, and what this episode reveals about you. Each additional disclosure promises to resolve uncertainty. Each one also widens the institution’s claim over the person whose life is being translated.

At some point, evidence becomes exposure. The distinction matters because modern institutions regularly treat visibility as a moral good in itself. A transparent person appears cooperative; a private person appears suspicious. A complete record appears rigorous; a partial record appears deficient. Yet the record is never complete. It is a purpose-built representation assembled under unequal conditions, usually by an institution that can store it, circulate it, compare it, and use it again after the situation that justified its creation has ended.

Unfinishedness is the right not to be finalized by that representation. It does not excuse fraud, concealment of harm, or refusal to answer for consequential action. It limits what institutions may demand beyond the evidence necessary to evaluate a defined claim. A person may owe an explanation of a decision without owing a complete theory of the self. A worker may owe evidence of performance without owing an emotional performance of devotion. A patient may owe clinically relevant information without making every intimate fact available for administrative reuse. A citizen may owe compliance with a lawful process without surrendering a permanent, transferable account of motive, identity, and future character.

The record is useful because it is incomplete

Institutions cannot act directly upon the full density of human life. They need forms, files, categories, summaries, scores, and timelines. James C. Scott’s work on administrative legibility showed why: large systems coordinate by simplifying local complexity into representations that can travel. The simplification is not automatically illegitimate. A payroll system needs standardized fields. A court needs an evidentiary record. A hospital needs a chart. A company needs some way to assess whether work was done.

The danger begins when the representation forgets that it was made for a purpose. A medical note written to guide care becomes a durable character judgment. A performance review written for one cycle follows a worker into unrelated decisions. A conversational disclosure to an employee assistant silently affects later recommendations. A risk classification created under one threshold migrates into another system and appears there as established fact.

The file gains authority precisely because it travels more easily than the person. The person cannot accompany it into every meeting, model, database, or future interpretation. Context falls away. Corrections arrive late. The record becomes stable while the life it describes continues to change.

This is why accuracy, although necessary, is not enough. A statement can have been accurate when recorded and still become unjust when reused outside its purpose, detached from its uncertainty, retained beyond its legitimate life, or allowed to govern a person who no longer fits it. The ethical burden rises with portability and consequence: the farther a representation travels and the more it can decide, the stronger its controls must become.

Total visibility produces performance, not truth

Institutions often respond to uncertainty by requesting more information. The instinct is understandable. More data appears to reduce the chance of error. But a system that continually rewards disclosure changes the behavior it is trying to observe.

Michael Power’s account of audit culture helps explain the result. Once people know that legitimacy depends on producing inspectable traces, they reorganize work around what can be shown. Decisions are narrated in advance. Emails are written for unknown future readers. Notes become defensive exhibits. Ambivalence is edited into confidence. Humor disappears. Care is translated into competency language. The institution receives more documentation and less unperformed reality.

This is not proof that people are deceptive. It is evidence that they are adapting rationally to an evaluative environment. When every fragment may become character evidence, ordinary life becomes impression management. The demand for authenticity then produces its opposite: a self composed for legibility.

Consider a performance system that asks a worker to demonstrate commitment. Work product, reliability, and collaboration may be relevant. Visible enthusiasm, personal disclosure, after-hours availability, and a coherent story about professional identity are not equivalent forms of evidence. The latter reward those with time, safety, rhetorical fluency, and fewer competing obligations. They also invite the organization to confuse a person’s capacity to narrate devotion with the quality or integrity of the work.

Or consider an AI assistant that helps an employee understand leave policy. The employee asks a routine question, then admits feeling afraid and unreliable. The disclosure may help the system respond with care. It does not thereby become legitimate input for performance evaluation, team assignment, career recommendations, model improvement, or future personalization. The system has learned something. It has not acquired unrestricted jurisdiction over what it learned.

More visibility can therefore reduce truth in two ways. It can elicit strategic performance, and it can tempt the institution to draw conclusions its evidence cannot support. The solution is not secrecy. It is disciplined relevance.

Accountability and exposure are different practices

Accountability is answerability for action, obligation, and harm. It requires a defined question, an authorized forum, evidence proportionate to the question, and a path to challenge the conclusion. Exposure treats the person as the object of inquiry. Its appetite has no natural endpoint because every unanswered detail can be redescribed as the missing key.

The distinction becomes clearer when power is considered. Institutions exercise durable authority: they allocate resources, determine access, record findings, define eligibility, and shape future options. Persons subject to those decisions usually cannot see the full rule set, model, data flow, or interpretive process. Yet transparency is often demanded downward. The worker must explain the self; the organization need not explain the inference. The applicant must disclose; the scoring system remains proprietary. The patient must become legible; the downstream circulation of the record remains obscure.

A defensible transparency ethic reverses that default. Institutions should be more legible than the people they govern. Rules, decision criteria, data sources, model limits, retention periods, review paths, and accountable owners should face the light. The person should provide the minimum evidence needed to resolve the legitimate question.

This is directional sunlight: scrutiny aimed toward concentrated power, not indiscriminate illumination of ordinary life. It preserves the democratic value of transparency while refusing to turn transparency into a universal duty of self-exposure.

Six constraints for accountable knowledge

Unfinishedness becomes institutionally useful when it changes the design of inquiry. Six constraints follow.

1. Purpose before collection

An institution should state the decision it is trying to make before deciding what it wants to know. “We may need this later” is not a sufficient purpose. The evidence request should be tied to a present authority, a defined risk, and an identifiable use.

2. Relevance before intimacy

The fact that information is revealing does not make it probative. Institutions should distinguish evidence of action from material that primarily invites character inference, emotional conformity, or reputational judgment. A disclosure request should be rejected when its main function is to make the person easier to classify rather than the decision more accurate.

3. Context must travel with the record

Helen Nissenbaum’s account of contextual integrity provides the right orientation: privacy is not simply secrecy but appropriate information flow. A record should carry its originating purpose, source, date, uncertainty, authorized users, and limits on reuse. When context cannot travel, the record’s authority should contract.

4. Interpretation must remain contestable

Correction is not a comment box. A person needs to know what was inferred, which evidence supported it, who can alter the conclusion, and whether the correction reaches downstream records and systems. Human review is real only when the reviewer has time, evidence, independence, and authority to depart from the automated or institutional account.

5. Authority should expire

Retention is not neutral. A record kept indefinitely remains available for purposes no one anticipated when it was created. Institutions should specify when evidence, profiles, classifications, and generated reasons lose authority, require renewal, or must be deleted. Some residue may be justified; permanent possession should never be the silent default.

6. The burden of explanation belongs to power

When a system affects employment, care, access, reputation, or opportunity, the institution should explain the rule, source, inference, and review path. The person should not have to expose more of the self simply to discover how the institution has already interpreted them.

Attention without possession

The deepest objection to total exposure is relational. To know another person is not to complete them. Friendship offers one model: it attends across time, remembers, notices patterns, and sometimes speaks hard truths. Yet a good friend remains answerable to correction and does not treat knowledge as ownership. The friend knows that the person may change, surprise, contradict the record, or decline to explain.

Simone Weil described attention as a form of regard that does not seize its object. That restraint has institutional consequences. A system can be designed to notice what matters without claiming everything that could be extracted. A reviewer can ask for sufficient evidence without turning vulnerability into proof of sincerity. A record can preserve accountability without becoming a permanent identity.

Unfinishedness does not ask institutions to become intimate, forgiving, or vague. It asks them to become precise about their own authority. What decision is being made? What evidence is necessary? What remains uncertain? Who can contest the interpretation? How long may the record govern? What must be relinquished when the purpose ends?

Those questions make accountability harder, not softer. They prevent the institution from resolving its uncertainty by transferring the entire burden onto the person. They also preserve the possibility that the person will become someone the record could not predict.

A society that cannot tolerate partial visibility will eventually recognize only what its systems can render.

The right to remain unfinished is therefore not a right to disappear from consequence. It is the right to be judged through evidence proportionate to the claim, by institutions whose own authority is visible, limited, contestable, and capable of ending.

---

Selected references

• James C. Scott, Seeing Like a State.
• Michael Power, The Audit Society: Rituals of Verification.
• Helen Nissenbaum, Privacy in Context.
• Simone Weil, writings on attention in Waiting for God and Gravity and Grace.
• Édouard Glissant, Poetics of Relation.