Realized vs Possible Outcomes¶
Core Idea¶
Realized-vs-possible outcomes is the structural comparison between what a process actually produces and what the process could in principle produce — two sets, one a subset of the other, where the relationship between them is itself a load-bearing object of analysis. The realized set — what happens, what was achieved, what was visited — is observable; the possibility set — what could happen, what could have been achieved, what is reachable — is constructed from a model of the process, often with effort, and is the implicit reference against which the realized outcomes are interpreted. The essential commitment is to hold both sets in view at once and to treat the gap between them, its shape, and its causes as the primary analytical object.
Three structural pieces recur. First, a process with definable inputs, dynamics, or rules that determine what outputs are possible. Second, a possibility set — the full collection of outcomes the process could produce under its constraints: the reachable set of a dynamical system, the feasible set of an optimization, the support of a distribution, the action set in a game, the capability set in Sen's framework. Third, a realized set — the subset actually produced under the particular inputs, history, or play that occurred. The relationship is set inclusion (realized is a subset of possible), with the gap as the analytical object; the gap's shape — uniform, lumpy, biased, fractal — is informative about which possibilities go unrealized and why. The distinctive structural commitment is that the possibility set is constructed, not observed: different models of the process yield different possibility sets, and that construction is part of the analysis. Without the prime, analysis collapses one side — either the possibility set is unstated and the realized is treated as the whole story, or the realized is read as if it exhausted the system's behaviour.
How would you explain it like I'm…
Could-Have vs Did
What Could Have Happened
Realized Inside Possible
Structural Signature¶
the generating process — the constructed possibility set — the observed realized set — the set-inclusion relation — the gap as primary object — the gap shape — the construction-not-observation invariant
A configuration instantiates the realized-vs-possible comparison when each of the following holds:
- A generating process. Some process with definable inputs, dynamics, or rules determines what outputs are possible — a dynamical system, optimization, distribution, game, grammar, or person's circumstances.
- A possibility set. The full collection of outcomes the process could produce under its constraints: reachable set, feasible set, support, strategy space, capability set, full coverage.
- A realized set. The subset actually produced under the particular inputs, history, or play that occurred: trajectory, chosen point, sample, equilibrium actions, functionings, exercised paths.
- A set-inclusion relation. The realized set is a subset of the possibility set; the relationship between them, not either set alone, is the unit of analysis.
- The gap as primary object. What is possible but unrealized is the central analytical target — unused capacity, unexercised risk, unachieved functioning.
- A gap shape. The gap is uniform, lumpy, biased, or fractal, and its shape is informative about which possibilities go unrealized and why; selection effects may bias the realized set as a sample of the possible.
- The construction-not-observation invariant. The realized set is observed but the possibility set must be constructed from a model of the process; different models yield different possibility sets, so the construction is itself part of the analysis. Collapsing the two — reading the realized as the whole story — is the characteristic failure.
These components compose into a two-set comparison: an observed realized set nested inside a constructed possibility set, with the gap's size and shape as the object — admitting three interventions (expand the realized, shrink the possible, reshape the gap).
What It Is Not¶
- Not counterfactual reasoning (see
counterfactual_reasoning).counterfactual_reasoningasks what would have happened under a specific alternative antecedent — a single contrast world; realized-vs-possible holds the entire possibility set against the realized subset. One probes a chosen alternative; the other surveys the whole reachable set. - Not regret (see
regret).regretis the valued gap between the realized outcome and the best foregone one; realized-vs-possible is the unvalued set comparison. Regret presupposes a preference ordering over the gap; the prime is structural and indifferent to value. - Not a biased sample of the possible (see
sampling_representativeness).sampling_representativenessconcerns whether the realized set fairly represents the possible; that is one reading of the gap's shape, but the prime is the broader two-set comparison of which selection bias is one diagnostic. - Not opportunity cost (see
opportunity_cost).opportunity_costis the value of the best foregone alternative relative to the chosen one; realized-vs-possible enumerates all unrealized possibilities without ranking them by value. - Not the realized story alone. Treating realized outcomes as exhausting the system's behaviour collapses the comparison — the prime's whole content is that the realized must be judged against a constructed possibility set, not read as the whole.
- Common misclassification. Comparing two systems' realized outputs without normalizing for possibility-set size — confusing freedom with outcome (Sen's critique). Catch it by asking whether the underlying possibility sets are the same size; if not, the comparison must be on the gap or the capability set.
Broad Use¶
- Dynamical systems (reachability) — the reachable set is the possibility set; the trajectory traced is the realized set; safety verification operates on the gap.
- Optimization and decision theory — the feasible set is the possibility set; the chosen point is the realized; suboptimal choices sit inside but not at the Pareto frontier.
- Probability theory — the support of a distribution is the possibility set; a finite sample is the realized set.
- Game theory — the strategy space is the possibility set; equilibrium actions are the realized set; off-equilibrium paths constrain behaviour through threat structure.
- Linguistics — the grammar's generative possibility space is the possibility set; produced forms are the realized set; creativity lives in the gap.
- Sen's capability framework — the capability set is the possibility set; achieved functionings are the realized set; wellbeing assessment must look at capabilities, not only functionings.
- Engineering safety — designed, reachable, and exercised state spaces give a three-level distinction on which certification operates.
- Organizational strategy — Mintzberg's intended versus realized strategy makes the gap the unit of strategic learning.
- Software testing — full execution paths are the possibility set; paths exercised by the suite are the realized set; coverage reports on the gap.
- Information theory — channel capacity is the possibility set; the rate achieved is the realized; the gap is engineering inefficiency.
Clarity¶
Naming the realized-vs-possible distinction separates four kinds of question that ordinary analysis often fuses: descriptive (what did the system produce?), capacity (what could it produce?), gap (what is possible but unrealized?), and constraint (why is the gap the size it is?). The prime makes each a separate analytical move with its own evidence requirements. It also forces a methodological discipline: you have to construct the possibility set explicitly. The possibility set is rarely observable; it is built from the model of the process, and different modellers produce different possibility sets for the same process. Disputes about whether a system performs well often turn out to be disputes about the size of the possibility set, not about the realized outcomes.
The clarifying force is to expose a specific failure mode: conflating the realized with the possible. When analysts treat realized outcomes as the whole story, they miss systems whose realized performance looks good only because the possibility set was small, and systems whose realized performance looks bad only because the possibility set was large. The same realized output carries different meanings depending on the size and shape of the possibility set it came from — which is precisely Sen's argument that wellbeing requires capability assessment, not just functioning assessment.
Manages Complexity¶
The arrangement compresses a large family of substrate-specific concepts — reachability, feasibility, support, capability, exercised states, realized strategy, coverage — into a single structural comparison with reusable parts: a process, a possibility set constructed from the process model, a realized set observed from actual behaviour, and a gap that is itself an analytical object. The analyst asks the same questions in any substrate: how do I construct the possibility set; what is in the realized set; what is the shape of the gap; what would shift or expand it?
The intervention space sorts cleanly. To expand the realized set, intervene on process inputs or remove the constraints that limit it — increase coverage, broaden exploration, lift the capability constraint. To shrink the possibility set, narrow the process's allowed behaviours — safety constraints, scope reduction, operational-envelope restriction. To change the gap shape, intervene on the bias that determines which possible outcomes are realized — priority orderings, exploration policies, default trajectories. The leverage is that these three intervention families are the complete set of moves on the relationship, so the complexity of "how do I improve this system?" collapses into a choice among expanding the realized, shrinking the possible, or reshaping the gap.
Abstract Reasoning¶
Realized-vs-possible outcomes trains a reasoner to ask:
- What is the process, and what inputs, dynamics, or rules determine what it can produce?
- How is the possibility set constructed from the process model — and have I actually done that construction, or left it implicit?
- What is in the realized set, observed from actual behaviour?
- What is the shape of the gap — uniform, lumpy, biased, fractal — and what does that reveal about which possibilities go unrealized?
- Is the realized set a representative sample of the possibility set, or are selection effects operating in the gap?
- Am I comparing two systems' realized outcomes without normalizing for the size of their possibility sets?
The portable inferences are that realized outcomes under-determine the possibility set (knowing what happened does not tell you what could have happened; the possibility set must be constructed separately); that the gap is informative (a large gap signals unused capacity or unrealized risk, while nearly-equal sets signal a system operating near its limits); that selection effects operate in the gap, biasing inference when realized outcomes are unrepresentative; and that comparative assessment requires possibility-set normalization. A subtler inference is that many disputes about fairness or justice turn on whose possibility set is the reference, as in Sen's capability critique of welfare economics: the same realized resource carries different meanings depending on the possibility set it was drawn from.
Knowledge Transfer¶
Role mappings across domains:
- Process ↔ dynamical system / optimization / random variable / game / grammar / person's circumstances / program
- Possibility set ↔ reachable set / feasible set / support / strategy space / generative space / capability set / full coverage
- Realized set ↔ trajectory / chosen point / sample / equilibrium actions / produced forms / functionings / exercised paths
- Relationship ↔ set inclusion, with the gap as the primary object
- Gap shape ↔ uniform / lumpy / biased / fractal, informative about unrealized possibilities
- Interventions ↔ expand-realized / shrink-possible / reshape-the-gap
A control engineer computing a reachable set, a welfare economist comparing capabilities rather than incomes, a software tester reporting coverage, and a strategist comparing intended to realized strategy are doing the same structural work: construct the larger possibility set, observe the smaller realized set, and reason about the relationship as the unit of analysis. The transfers among substrates are documented and direct. The reachability technique from control engineering ports into software model-checking, where reachable program states must be enumerated or bounded. Sen's capability-versus-functioning distinction transferred from development economics into AI fairness, as the realization that outcome equality does not imply opportunity equality, with some fairness frameworks explicitly invoking Sen. The coverage-of-possibility-space discipline ported from software testing into empirical-method evaluation, comparing the explored region of a treatment space to the full space. Mintzberg's gap-as-learning framing transferred from strategy into organizational-learning theory and after-action review. The reachable-set technique ported from dynamics into privacy analysis, computing the set of inferences an adversary could reach from released data. The non-transfer caveat is that the prime depends on a constructible possibility set: where the set is genuinely unbounded or computationally intractable — open-ended creativity, undefined problem spaces — the gap-analysis machinery breaks down. What moves between fields is the literal set-comparison move — two sets, one nested in the other, the relationship as the object — together with its three intervention families and the methodological discipline of constructing the possibility set explicitly rather than reading the realized as the whole story.
Examples¶
Formal/abstract¶
Reachability analysis for a safety-critical dynamical system is the prime in its most rigorous form. The generating process is a controlled dynamical system \(\dot{x} = f(x, u)\) with state \(x\), control inputs \(u\) drawn from an admissible set, and an initial-condition set \(X_0\). The constructed possibility set is the reachable set \(\mathcal{R}\): the collection of all states the system can attain from \(X_0\) over the time horizon under every admissible control and disturbance. Crucially this set is constructed, not observed — it is computed from the model via forward set-propagation (zonotopes, ellipsoidal bounds, or Hamilton-Jacobi level sets), and a coarser model yields a larger, more conservative possibility set, instantiating the construction-not-observation invariant directly: the verification result depends on which model produced \(\mathcal{R}\). The observed realized set is the actual trajectory (or family of trajectories) the system traced under the inputs that occurred — a thin curve nested inside the fat reachable tube, making the set-inclusion relation literal. Safety verification is precisely an operation on the gap: certify the system iff the reachable set never intersects the unsafe region — i.e., the possibility set, not merely the realized trajectory, stays clear of the failure states. This is why testing alone (sampling realized trajectories) is insufficient and reachability (bounding the possible) is required: a system can pass every realized test yet have a reachable set that grazes catastrophe. The three interventions are all visible — shrink the possible (tighten control constraints so \(\mathcal{R}\) contracts away from the unsafe set), reshape the gap (bias the controller's exploration), or accept the realized as adequate only once the possible is certified.
Mapped back: The dynamical system is the process, the computed reachable tube is the constructed possibility set, the actual trajectory is the realized subset nested inside it, and certifying that no reachable state touches the unsafe region is gap-analysis — with model choice determining \(\mathcal{R}\) showing the possibility set is constructed, not observed.
Applied/industry¶
Software-test coverage and Sen's capability framework apply the same two-set comparison in two industries. In software testing, the generating process is a program; the possibility set is the set of all feasible execution paths (or branches, or states) the program could take across its input space — constructed by static analysis or control-flow modeling, not observed; the realized set is the subset of paths actually exercised by the test suite during runs. A coverage report is literally a report on the gap: "78% branch coverage" names a realized set nested in a constructed possibility set, and the uncovered 22% is the analytically interesting region — unexercised paths where defects hide. The gap shape is informative: coverage clustered in common code paths with a sparse tail over error-handling tells the tester which possibilities go unrealized and why, directing the expand-the-realized intervention (write tests targeting uncovered branches). In welfare economics, Sen's framework instantiates the prime with explicit normative stakes: a person's capability set is the possibility set (the functionings — states of being and doing — they could achieve given their resources and conversion factors), while their achieved functionings are the realized set. Sen's decisive argument is exactly the prime's don't-collapse-the-two discipline: two people with identical realized functionings (say, the same nutrition level) can have radically different capability sets (one chooses to fast amid plenty, the other starves with no alternative), so assessing wellbeing from realized outcomes alone — reading the realized as the whole story — misjudges both. The possibility-set normalization inference becomes a justice claim: comparing two people's realized resources without normalizing for the size of their capability sets confuses freedom with outcome, which is why this transferred into AI fairness as the recognition that outcome equality does not entail opportunity equality.
Mapped back: Feasible execution paths and a person's capability set are constructed possibility sets; exercised paths and achieved functionings are the observed realized subsets; coverage reports and capability assessments are both gap analysis; and Sen's insistence that identical realized functionings can mask unequal capability sets is the prime's warning against collapsing realized into possible, across a software and a welfare-economics substrate.
Structural Tensions¶
T1 — Observed Realized versus Constructed Possible (Measurement). The realized set is observed; the possibility set is constructed from a model, and never directly seen. This is the prime's load-bearing asymmetry — and its deepest hazard, because the construction is silently model-dependent. The failure mode is reading the realized as the whole story: treating what happened as exhausting what could happen, so a system looks excellent only because its possibility set was never built and its small realized output went unjudged against anything. Diagnostic: ask whether the possibility set was actually constructed or left implicit; if no one can state how the reference was built, the analysis has collapsed one side and any "performance" claim is unanchored.
T2 — Possibility-Set Size as Model Artefact (Scopal). Different models of the same process yield different possibility sets — a coarse model gives a larger, more conservative set; a fine one a tighter set. So the gap's size is partly a fact about the model, not the process. The failure mode is reifying a modelling choice as a property of the world: declaring huge unused capacity when the possibility set was merely inflated by a loose model, or certifying near-optimality when a tight model hid reachable states. Diagnostic: ask how sensitive the gap is to the model that generated the possibility set; if a reasonable alternative model would move the gap substantially, the gap is reporting modelling slack, not the system.
T3 — Realized as Representative Sample versus Biased Selection (Measurement). The prime invites reading the realized set as a sample of the possible, but the selection that produced it is rarely uniform — the gap has a shape, and which possibilities got realized is often biased. The failure mode is inferring the possibility set's properties from a skewed realized sample: concluding a region is safe because the trajectories that happened to run never entered it, when selection (not impossibility) kept them out. Diagnostic: ask what determined which possibilities were realized; if a priority ordering, default policy, or survivorship effect biased the draw, the realized set under-represents exactly the possibilities that matter, and absence in the realized set is not absence in the possible.
T4 — Comparing Systems versus Normalizing Possibility Sets (Scalar). Two systems' realized outcomes are comparable only after normalizing for the size of the possibility sets they were drawn from — the same realized output means different things from a large versus a small possible set. The failure mode is raw outcome comparison: ranking two performers, policies, or people by realized result while ignoring that one had a vastly larger feasible set, confusing freedom with outcome (Sen's exact critique). Diagnostic: before comparing realized outputs, ask whether the underlying possibility sets are the same size; if not, the comparison must be on the gap or on the capability set, and unnormalized outcome rankings reward whoever started with the smaller possible set.
T5 — Three Interventions Pull Opposite Ways (Sign/Direction). The complete move-set — expand the realized, shrink the possible, reshape the gap — contains directly opposed operations, and choosing wrong worsens the system. The failure mode is sign confusion: applying expand-the-realized (broaden exploration, lift constraints) to a safety problem that needed shrink-the-possible (tighten the operating envelope so the reachable set contracts away from catastrophe), thereby enlarging exposure while believing you improved coverage. Diagnostic: ask whether the goal is to use more of the possible (capacity) or to keep out of part of the possible (safety); these demand opposite-signed interventions on the same two sets, and a capacity reflex applied to a safety gap is actively dangerous.
T6 — Constructible versus Unbounded Possibility Set (Scopal). The whole machinery presupposes a constructible possibility set; where the set is genuinely unbounded, undefined, or computationally intractable — open-ended creativity, ill-posed problem spaces — gap analysis breaks down. The failure mode is forcing the frame anyway: fabricating a bounded possibility set for an open domain and then reasoning about a "gap" that is an artefact of an arbitrary boundary, mistaking the limits of enumeration for the limits of the process. Diagnostic: ask whether the possibility set can be bounded in a principled way; if the boundary is arbitrary or the set is effectively infinite, the prime does not apply, and gap percentages computed against a made-up ceiling are meaningless.
Structural–Framed Character¶
Realized vs Possible Outcomes sits at the structural end of the structural–framed spectrum, consistent with its frontmatter label and an aggregate of 0.0: it is a pure set-theoretic comparison whose vocabulary is mathematical and travels unmodified.
Every diagnostic points the same way. The home vocabulary travels freely: reachable set, feasible set, support, strategy space, and capability set are the same construction — a realized subset nested in a constructed possibility set, with the gap as the object — whether the substrate is a controlled dynamical system, an optimization, a probability distribution, a game, a grammar, or Sen's capability framework, and the reachability technique ports from control engineering into software model-checking without translation. The prime carries no evaluative weight: the gap is unvalued, applying equally to safety verification (where the unrealized region is catastrophes one is glad went unrealized) and to capability assessment (where it is goods one wishes had been realized). Its origin is formal — set inclusion between an observed and a constructed set — with no appeal to human institutions. It runs in physical and abstract substrates (reachable tubes of a controlled system, the support of a random variable) as readily as in social ones, so it is not human-practice-bound. And invoking it merely recognizes the two-set relation already implicit in any generating process rather than importing an interpretive frame; even Sen's normatively weighty capability critique is an application of the value-neutral structure, not evidence the structure itself carries value. On every axis the prime reads structural, exactly as the 0.0 aggregate records.
Substrate Independence¶
Realized-vs-Possible Outcomes is a maximally substrate-independent prime — composite 5 / 5 on the substrate-independence scale. The signature is a bare set-inclusion relation — a realized set nested inside a constructed possibility set, with the gap between them as the analytical object — which is as medium-neutral as a structure can be, so the structural abstraction is maximal. The domain breadth is wide and the structural force identical across it: dynamical systems (reachable set versus trajectory), optimization and decision theory (feasible set versus chosen point), probability (support versus sample), game theory (strategy space versus equilibrium play), linguistics (generative space versus produced forms), Sen's capability framework (capabilities versus achieved functionings), engineering safety (designed/reachable/exercised state spaces), organizational strategy (Mintzberg's intended versus realized strategy), software testing (paths possible versus paths covered), and information theory (channel capacity versus achieved rate). The transfer evidence is strong because the underlying mathematics — set inclusion and the geometry of the gap — is literally shared across these substrates, so the comparison is recognized rather than translated wherever a process's actual output is read against what it could in principle produce.
- Composite substrate independence — 5 / 5
- Domain breadth — 5 / 5
- Structural abstraction — 5 / 5
- Transfer evidence — 5 / 5
Neighborhood in Abstraction Space¶
Realized vs Possible Outcomes sits in a sparse region of abstraction space (60th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely rather than landing on a neighbor.
Family — Staged Processes & Drift (32 primes)
Nearest neighbors
- Outcome-Defined Adequacy — 0.71
- Eventual Realisation of Possibility — 0.71
- Stage Gate Process — 0.71
- Yield Loss — 0.70
- Accidental Vs Essential Complexity — 0.70
Computed from structural-signature embeddings · 2026-06-14
Not to Be Confused With¶
Realized-vs-possible outcomes is most readily confused with counterfactual_reasoning, since both concern what happened versus what didn't. The difference is cardinality and structure. counterfactual_reasoning evaluates a single alternative world: it fixes a specific antecedent change ("had the medication been given," "had the rate been raised") and reasons about the consequent that would have followed, comparing the actual world to one nearby possible world selected for relevance. Realized-vs-possible holds the entire possibility set in view at once — the full reachable set, feasible region, support, or capability set — and treats the realized subset's relationship to that whole as the object. Counterfactual reasoning picks one path not taken and asks where it would have led; realized-vs-possible asks about the shape and size of all the paths not taken collectively. They compose — one might use counterfactual reasoning to probe a particular point inside the gap — but they are not the same move. A practitioner who reduces realized-vs-possible to a counterfactual will examine one foregone alternative and miss the structural questions the prime is built for: how big is the gap, what is its shape, was the realized set a biased draw, and how does the possibility set's size condition the meaning of the realized output. Conversely, treating a counterfactual question as a full set-comparison over-builds a possibility set when only one contrast world was needed.
The prime is also confused with regret, which intuitively names the gap between what happened and what could have happened — exactly the prime's object, but valued. regret is the experienced or computed cost of the difference between the realized outcome and the best outcome that was available: it presupposes a preference ordering over outcomes, identifies the best foregone one, and measures the shortfall. Realized-vs-possible is structurally prior and value-neutral: it constructs the two sets and characterizes the gap without ranking any outcome above another, applying equally to safety verification (where the "unrealized" outcomes are catastrophes one is glad went unrealized) and to capability assessment (where they are goods one wishes had been realized). Regret can only be computed after the prime's comparison is made and after a value function is supplied; the prime supplies the set structure on which regret, opportunity cost, and value-of-information are then defined. Confusing the two imports a preference ordering where none belongs — reading every unrealized possibility as a loss to be regretted, when in safety the unrealized region is precisely what one wants to keep empty, and the correct intervention is to shrink the possible, not lament the gap.
A third confusion, subtler and methodologically important, is with sampling_representativeness. When the realized set is read as a sample of the possibility set, the question "is this sample representative?" is genuinely a realized-vs-possible question — specifically, a question about the shape of the gap and whether selection biased the draw. But sampling_representativeness is a special case and a diagnostic within the prime, not the whole of it. The prime also covers cases where the realized set is not a sample at all but a single deterministic trajectory (a controlled system's path through its reachable set), where "representativeness" does not apply, and cases where the interesting fact is the gap's size (unused capacity) rather than its bias (skewed sampling). Treating realized-vs-possible as merely a representativeness check narrows it to the measurement question and drops the capacity and safety questions; treating a representativeness problem as the full prime over-generalizes a sampling concern into claims about reachability or capability the data cannot support. The clean relation is that selection bias in the gap is one of the prime's diagnostics (its T3 tension), sitting alongside gap-size, model-dependence, and normalization.
These distinctions matter because each neighbor captures only one facet of the two-set comparison. counterfactual_reasoning probes one foregone path; regret values the gap against a preference ordering; sampling_representativeness audits whether the realized set fairly samples the possible. Realized-vs-possible is the structural substrate on which all three operate — the explicit construction of a possibility set, the observation of a nested realized set, and the treatment of their relationship as the unit of analysis. Keeping it distinct is what lets a practitioner avoid reading the realized as the whole story, normalize comparisons by possibility-set size, and choose correctly among expanding the realized, shrinking the possible, or reshaping the gap.
Solution Archetypes¶
No catalogued solution archetypes reference this prime yet.