A protective decision has a deadline before it has an answer. Someone must act before a notice goes out, before a map is published, before a queue starts moving again, and must do so while something important is still unknown. Once the deadline has passed, the same decision may need to be opened again if events show that it failed the people it was meant to protect. This paper asks how one set of decision records frames that sequence. Its question, put to the record set rather than to any one case, is this: who must decide what by when, and what would make them revisit the decision?
The records are sixteen authored cases from AEGIS, the safety and protection studio of the RISCON family. They are synthetic. AEGIS wrote them as independent exercise cases in a fictional administrative and livelihood-support setting, and it says plainly, in each record, that the setting is not a real country, event or law. The paper therefore makes no claim about how real offices behave. It studies how a body of authored records describes a protective decision: what it names, what it counts, what it leaves unnamed. That is a narrower question than the ones the literature on emergencies usually asks, but it is a question that can be answered from the records alone, and it bears on a practical matter. A decision record is useful to a later reader when it shows who was supposed to decide, by when, on what, and under what conditions the decision was to be looked at again. Whether a record does that depends on how it is framed.
The argument in brief
The paper makes four claims about the set and one about the method.
First, every case names a single office as the owner of the decision, and the office is defined by the problem: a lead for application relief, an officer for procedural protection, a coordinator for resumption. No case spreads the decision across a committee, and none places it with a political leader.
Second, every deadline is counted from an operational event in the administration's own calendar, such as the publication of a list, the next payment decision or the shutdown of an old system. No deadline names a date. Fourteen of the sixteen are set before the event; the two set after it both belong to recovery, where the protective task is to close down what the emergency left open.
Third, each case names one unknown, and in six of the sixteen the unknown sits on the applicant's side of the counter: whether a delegation was voluntary, whether a person learned of a notice by another route, how well applicants understand a sentence. The office must decide before it can know what the person meant or understood.
Fourth, every case writes its condition for reopening as an event: the moment at which the harm the decision was meant to prevent would become visible. None of these conditions names a number, a share or a threshold, and none says who bears the cost of waiting while the condition is watched for.
The methodological claim is that these features can be told apart from the studio's template by measurement. A count of repeated five-word runs, the method the RISCON series uses to separate a studio's frame from its findings, shows that the parts of each record that carry the decision (the deadline, the owner, the unknown, the condition for reopening, the thing to observe) use each case's own words, while the links the records draw between cases are largely built from a repeated formula. The form of the reopening condition is uniform across the set, a sentence beginning with "when"; its content is not.
The debate this paper enters
Four bodies of work speak to the question of deciding by a deadline under uncertainty, and they pull in different directions.
The first studies protective action in the face of warnings. Michael K. Lindell and Ronald W. Perry, The Protective Action Decision Model: Theoretical Modifications and Additional Evidence (2012), describe a multistage model built from research on how people respond to hazards and disasters. In their model, warnings and environmental cues must first be received, attended to and understood before anything else happens; then perceptions of the threat, of the protective action and of the stakeholders form the basis of a decision, and situational facilitators and impediments shape what people actually do. Dennis S. Mileti and Lori Peek, The Social Psychology of Public Response to Warnings of a Nuclear Power Plant Accident (2000), review the evidence that people take protective action according to the substance of a warning and the course by which it reaches them, and they set out the components of a warning system. As this paper reads them, both works put the protected person at the centre and treat the warning as the hinge between hazard and response. The AEGIS cases invert this arrangement, in a way this paper reads as instructive. The protected person is still the applicant, but the one who must take protective action is the office, and the hazard is often a message the office itself sent: a bounced email, a misleading notice, a public table. The question of reception, which the warning literature places at the start of the process, appears in these records as the office's own unknown.
The second body of work studies crisis management as a problem of public leadership. Arjen Boin, Paul 't Hart, Eric Stern and Bengt Sundelius, in The Politics of Crisis Management: Public Leadership under Pressure (2005), treat crisis management as a defining feature of contemporary governance. In their account, communities expect leaders to minimise the impact of a crisis while critics and competitors look for someone to blame, and policy makers must establish a sense of normality and foster collective learning. Jeanne-Marie Col, Managing Disasters: The Role of Local Government (2007), argues from a comparison of two disasters that local communities are often on their own for hours or days and that local governments can protect their citizens by taking the initiative in risk assessment, decision making and communication. Between them these works raise the question of where a protective decision is lodged: at the top, where blame and legitimacy are contested, or close to the ground, where action can be taken in time. The AEGIS records answer it in one direction only, as the case sections show.
The third body of work concerns precaution. Cass R. Sunstein, Laws of Fear: Beyond the Precautionary Principle (2005), argues that the principle of acting against potential harms despite uncertain causal chains is incoherent, because risks lie on all sides of a situation and precautionary steps create dangers of their own. In its place he proposes a narrow principle against catastrophe, attention to costs and benefits, and an approach that respects choice while steering people toward better outcomes. The debate matters here because several AEGIS cases offer, among their three choices, a way to hold: to withhold a decision, to suspend a function, to wait for confirmation. Sunstein's point that precautionary steps create dangers of their own can be applied, on this paper's reading, to a hold: it is not free, and the records do not say what it costs or who bears the cost.
The fourth body of work concerns revision. Charles E. Lindblom, The Science of "Muddling Through" (1959), contrasts a comprehensive method of policy-making, impossible for complex problems, with a method of limited comparisons among a few alternatives, most of them familiar, in which those who decide expect to reach their goals only partially and to repeat the sequence as conditions and aspirations change. Marjolijn Haasnoot, Jan H. Kwakkel, Warren E. Walker and Judith ter Maat, in Dynamic Adaptive Policy Pathways: A Method for Crafting Robust Decisions for a Deeply Uncertain World (2013), describe planning under deep uncertainty as committing to short-term actions while setting out signposts to monitor whether adaptation is needed. Both works, as this paper reads them, make revisiting part of deciding rather than a later correction: Lindblom's administrator expects to repeat the sequence, and in the pathways method the signposts are what tell the planner whether the plan needs to adapt. The AEGIS records give every decision both a condition for reopening and something to observe, which looks like a signpost. The question is whether the condition could be recognised when it came.
Two further works sharpen what the records leave out. Daniel Kahneman and Gary Klein, Conditions for Intuitive Expertise: A Failure to Disagree (2009), conclude that the likely quality of an intuitive judgment depends on how predictable the environment is and whether the person judging has had the chance to learn its regularities. A record that asks an office to observe and record what follows a decision is, on this reading, a record that could give the office that chance. Donald Moynihan, Pamela Herd and Hope Harvey, Administrative Burden: Learning, Psychological, and Compliance Costs in Citizen-State Interactions (2015), describe the costs citizens bear in dealing with the state, and argue that the distribution of those costs between the state and the individual is often a political choice rather than an accident. On this paper's reading, their concept points to what the AEGIS records leave unnamed: who carries the burden while a protective decision waits.
Contribution
The paper contributes in three narrow ways. It gives a measured description of how one authored record set frames protective decisions against a deadline: single offices as owners, deadlines on an operational clock, unknowns on the applicant's side, and reopening conditions written as visible failures without thresholds. It separates, by a stated and repeatable count, the parts of these records that carry case-specific content from the parts that repeat a template, so that later readers know which parts can bear an argument. And it places the set within the debate above, showing that the records follow the adaptive and incremental view in giving every decision a way back, while leaving unnamed the question the precaution and burden literatures press hardest: who pays for the time between the decision and its revision. Chapter 1 sets out the data and method and reads the cases in two sections; Chapter 2, an appendix, gives the records read; the conclusion returns to the debate.