Surrogation¶

Explain why a well-designed metric drifts from its purpose even absent any gaming: under load an accountable manager mentally replaces the strategic construct with its measure, treating the number as the thing itself rather than a partial proxy for it.

Core Idea¶

Surrogation is the documented cognitive tendency — named and characterised in the managerial accounting literature by Choi, Hecht & Tayler (2012, 2013), building on Lipe & Salterio (2000) — for managers and evaluators who are given a quantitative measure designed to represent a broader strategic construct (customer satisfaction, employee engagement, strategic value) to mentally replace the construct with the measure: to think, plan, and act as though the measure is the underlying thing rather than a partial, imperfect operationalisation of it. The substitution is not deliberate; under cognitive load, time pressure, and accountability for measurable performance, the concrete numerical measure becomes the operative referent while the abstract construct recedes to background.

The behavioural consequences are systematic: effort flows to whatever the metric rewards even where that diverges from what the strategic construct requires; the measure is manipulated when it becomes possible to do so at the construct's expense; tacit content of the construct that the measure does not capture is lost from decision-making. Surrogation is the cognitive mechanism operating inside a single decision-maker's head; Goodhart's law names the corresponding system-level dynamic when many actors collectively optimise a measure until it decays as a proxy for the construct it was meant to track. The experimental signature of surrogation is that managers provided with both a strategy map (making the construct explicit and salient) and a balanced scorecard show less surrogate behaviour than those given the scorecard alone — the construct's presence must be actively maintained, because under load the mind defaults to the measure as its operative handle.

Structural Signature¶

Sig role-phrases:

the strategic construct — the broad, partly tacit outcome of interest, abstract and hard to measure
the measure — the concrete, partial quantitative operationalisation adopted to track the construct
the accountability structure — evaluations, bonuses, or reputations tied to the measurable number, putting weight on it
the load conditions — cognitive load, time pressure, and accountability for measurable performance, the regime in which a concrete number is a cheaper handle than an abstract construct
the construct-into-measure collapse — the non-deliberate substitution inside one decision-maker's head: the measure becomes the operative referent and the construct recedes
the behavioural fallout — goal-displacement toward what the metric rewards, manipulation of the measure at the construct's expense, loss of the construct's tacit content
the salience remedy — keep the construct actively present (strategy map alongside scorecard, narrative accountability alongside numerical, triangulated measures) so none installs itself as the stand-in

What It Is Not¶

Not gaming or deliberate manipulation of the metric. Surrogation is non-adversarial: no one consciously decides to manage the measure instead of the strategy. Under load, time pressure, and accountability for measurable performance, the concrete number simply becomes the operative handle while the construct recedes. This is why the fix is not tighter controls and audits — those assume intent that is not present — but keeping the construct actively salient.
Not Goodhart's law. Goodhart names the system-level dynamic in which many actors collectively optimise a proxy until it decays as a measure of the construct; surrogation is the cognitive substitution inside a single decision-maker's head. They are sibling faces of one proxy/target parent, not the same thing at different scales, and they take different remedies — the cognitive form yields to salience, the system form to incentive redesign.
Not a defect in the measure itself. Surrogation does not say the metric is a bad proxy or fails construct validity; a perfectly reasonable measure surrogates once it is held as the goal rather than as a stand-in. The failure is in the mental treatment of the measure, not in the number's quality — so improving the metric does not address it, and a good metric is no protection.
Not the substrate-free proxy/target-substitution pattern. Stripped of construct, measure, scorecard, and accountability vocabulary, surrogation becomes "the proxy replaces the thing in the chooser's head" — the cognitive-bias face of a broader proxy/target-substitution family whose system-incentive face (Goodhart) and clinical-endpoint face are separate members. The cross-domain reach belongs to that parent; "surrogation" as named carries managerial-evaluation furniture that does not travel.

Scope of Application¶

Surrogation lives across managerial accounting, organisational behaviour, and the judgment-and-decision-making literature wherever a single accountable evaluator holds a measure for a partly tacit construct under load; its reach is bounded to that managerial-evaluation substrate (the broader proxy/target-substitution pattern — whose system-incentive face is Goodhart's law and whose measurement-causal face is the surrogate-endpoint problem — carries the cross-domain reach, not the named cognitive effect).

Balanced scorecard and performance measurement — the home turf: managers given a scorecard come to manage the scorecard rather than the strategy it encodes (the Lipe–Salterio and Choi–Hecht–Tayler experiments).
Sales and customer-service incentives — NPS, CSAT scores, and ticket-resolution times displace "customer satisfaction" as the operative goal.
Public-sector performance management — hospital waiting-time targets, school league tables, and police clearance rates show surrogation in administrators' decisions and front-line behaviour.
CEO compensation tied to share price — "long-term shareholder value" is mentally replaced by "next-quarter EPS" or options vesting at a strike, and decisions diverge from the construct in documented ways.
OKR systems — a key result becomes the operative goal while the strategic objective it operationalises recedes to background.
Academic evaluation — the h-index and citation counts displace "research impact" in tenure and hiring decisions.

Clarity¶

Naming surrogation explains why a well-designed measurement system drifts from the purpose it was built to serve even when no one is gaming it. The usual suspicion when a metric stops tracking its construct is manipulation — someone cheating the number. Surrogation identifies a prior, non-adversarial failure: the manager does not consciously decide to manage the measure instead of the strategy; under load, time pressure, and accountability for measurable performance, the concrete number simply becomes the operative referent while the abstract construct it stands for recedes. That reframing matters because it sends the fix to a different place. If the drift were gaming, tighter controls and audits would be the answer; because it is a cognitive substitution, the remedy is to keep the construct actively salient — pairing a strategy map with the scorecard, maintaining narrative accountability for the construct alongside numerical accountability for the measure, triangulating several measures so no single one can install itself as the construct's stand-in.

The concept's sharpest contribution is to hold the construct and the measure apart as distinct objects, and to locate the failure precisely in their collapse — the measure ceasing to be treated as a partial, imperfect operationalisation and being treated as the thing itself. That lets a practitioner ask the diagnostic question directly: is this number still being held as a proxy, or has it quietly become the goal? It also fixes the level at which surrogation lives. It is the substitution inside a single decision-maker's head, distinct from the system-level dynamic in which many actors collectively optimise a proxy until it decays — so an organisation can recognise the cognitive form early, in an individual's planning, before it compounds into the collective one.

Manages Complexity¶

The ways a measurement system can drift from its purpose look, case by case, like a catalogue of separate organisational pathologies: a balanced scorecard managed instead of the strategy it encodes, NPS and CSAT displacing customer satisfaction, hospital waiting-time targets and school league tables and police clearance rates distorting front-line behaviour, CEOs steering to next-quarter EPS rather than long-term value, OKR key results eclipsing their objectives, the h-index standing in for research impact. Surrogation compresses that catalogue by holding two objects apart — the construct (the broad, partly tacit outcome of interest) and the measure (its concrete, partial operationalisation) — and locating every case's failure in a single event: their collapse, the measure ceasing to be treated as a proxy and becoming the operative referent. Instead of re-diagnosing each metric-drift episode from its domain particulars, the analyst tracks one thing and asks one question — is this number still being held as a proxy, or has it quietly become the goal? — and reads the familiar consequences off the answer: effort flowing to whatever the metric rewards even where that diverges from the construct, manipulation of the measure once that becomes possible at the construct's expense, loss of the construct's tacit content from decision-making. The decisive compression is a branch on cause that routes the fix, because surrogation is defined as non-adversarial. When a metric stops tracking its construct, the reflex is to suspect gaming, and gaming would call for tighter controls and audits; surrogation identifies the prior, non-deliberate failure — under cognitive load, time pressure, and accountability for measurable performance, the concrete number simply becomes the handle while the abstract construct recedes — for which controls are useless and the remedy is instead to keep the construct actively salient (pair a strategy map with the scorecard, hold narrative accountability for the construct alongside numerical accountability for the measure, triangulate several measures so none installs itself as the construct's stand-in). A second branch fixes the level: surrogation is the substitution inside a single decision-maker's head, distinct from the system-level dynamic in which many actors collectively optimise a proxy until it decays — so the cognitive form can be caught early, in an individual's planning, before it compounds into the collective one. The open question "why does this measurement system betray its purpose, and what do I do about it" thus reduces to one construct-versus-measure distinction, a proxy-or-goal diagnostic, and two branches — cause and level — that between them select the intervention.

Abstract Reasoning¶

Surrogation licenses a set of inferential moves in managerial accounting and judgment research, all built on holding two objects apart — the construct (the broad, partly tacit outcome of interest) and the measure (its concrete, partial operationalisation) — and locating failure in their collapse.

The signature diagnostic move runs from observed metric-drift to the precise event behind it. Rather than re-diagnose each episode from its domain particulars, the analyst asks one question — is this number still being held as a proxy, or has it quietly become the goal? — and reads the familiar consequences off the answer: effort flowing to whatever the metric rewards even where that diverges from the construct, manipulation of the measure once that becomes possible at the construct's expense, and loss of the construct's tacit content from decision-making. The decisive part of the diagnosis is a cause-branch that distinguishes surrogation from gaming. When a metric stops tracking its construct, the reflex is to suspect manipulation — someone cheating the number — but surrogation identifies a prior, non-adversarial failure: under cognitive load, time pressure, and accountability for measurable performance, the concrete number simply becomes the operative handle while the abstract construct recedes, with no one deciding to manage the measure instead of the strategy. The analyst infers the cause from the absence of intent and the presence of load, not from the behaviour alone.

The interventionist move is routed entirely by that cause-branch, and its value is preventing a mismatched fix. If the drift were gaming, tighter controls and audits would be the answer; because surrogation is a cognitive substitution, controls are useless and the remedy is instead to keep the construct actively salient — pair a strategy map with the scorecard, hold narrative accountability for the construct alongside numerical accountability for the measure, and triangulate several measures so no single one can install itself as the construct's stand-in. Each is a falsifiable prediction: making the construct explicit and salient should reduce surrogate behaviour relative to presenting the measure alone, whereas adding audit pressure (which assumes intent) should not. The experimental signature confirms the prediction directly — managers given both a strategy map and a balanced scorecard show less surrogate behaviour than those given the scorecard alone — so the construct's presence must be maintained, because under load the mind defaults to the measure.

The predictive move forecasts when and how surrogation will appear from its enabling conditions. The substitution is predicted under cognitive load, time pressure, and accountability for measurable performance — the regime in which a concrete number is a cheaper handle than an abstract construct — and is predicted to recede when those pressures relax or the construct is made salient. From the substitution the analyst predicts the downstream pattern (goal-displacement, measure-manipulation, lost flexibility) without modeling the particular metric, so NPS displacing satisfaction, EPS displacing long-term value, and the h-index displacing research impact are read as one forecast rather than separate findings.

The boundary-drawing move fixes the level and the scope. Surrogation is the substitution inside a single decision-maker's head, distinct from the system-level dynamic in which many actors collectively optimise a proxy until it decays — so an organisation can catch the cognitive form early, in an individual's planning, before it compounds into the collective one, and an intervention aimed at the individual's salience is the right lever for the cognitive level while incentive redesign is the lever for the system level. The boundary also marks where the concept is scaffolding-bound: stripped of construct, measure, scorecard, and accountability vocabulary, surrogation becomes "the proxy replaces the thing in the chooser's head," the cognitive face of a broader proxy/target substitution pattern whose system-incentive face and clinical-endpoint face are separate members — so the cross-domain reach belongs to that parent, while "surrogation" as named carries managerial-evaluation furniture that does not travel.

Knowledge Transfer¶

Within managerial accounting, organisational behaviour, and judgment research the concept transfers as mechanism, not as analogy, because every site in its range runs the same cognitive event — the construct collapsing into the measure inside a decision-maker's head under load — and is diagnosed and fixed the same way. Balanced-scorecard management (managing the scorecard rather than the strategy), sales and service incentives (NPS/CSAT/resolution-time displacing customer satisfaction), public-sector performance management (waiting-time targets, league tables, clearance rates), CEO compensation (next-quarter EPS displacing long-term shareholder value), OKR systems (a key result eclipsing its objective), and academic evaluation (the h-index displacing research impact) are not separate organisational pathologies — they are one construct-versus-measure collapse, read off the same proxy-or-goal diagnostic, attributed by the same non-adversarial cause-branch (load and accountability, not intent), and corrected by the same salience remedies (pair a strategy map with the scorecard, keep narrative accountability alongside numerical, triangulate measures). The experimental signature carries with it: making the construct salient reduces surrogate behaviour relative to the measure alone, in any of these settings. What makes this mechanism rather than metaphor is the shared substrate — a single evaluator, accountable for a measurable number that stands in for a partly tacit construct.

Beyond managerial evaluation the entry is a clean case of shared abstract mechanism (B). The genuinely substrate-spanning structure is the parent — proxy/target substitution: a measure adopted because it correlates with a construct comes to function as the construct, after which the two decouple. That pattern really recurs across distinct substrates as co-instances, not loose resemblance: Goodhart's law is its system-incentive face (many actors collectively optimise a proxy until it decays), the surrogate-endpoint problem in clinical research is its measurement-causal face (a biomarker standing in for a clinical outcome), and KPI gaming is its operational face. Surrogation is specifically the cognitive-bias face of that family — the substitution localised inside one head, with no one deciding to cheat. So when the cross-domain lesson is needed — "a proxy installed for its correlation with the target tends to displace the target, and then they part ways" — it should be carried by the proxy/target-substitution parent, not by "surrogation," whose distinctive cargo (the scorecard, the strategic-construct vocabulary, the accountability-and-bonus setting, the load-induced default) is managerial-evaluation furniture that does not travel.

Two boundary notes keep the transfer honest. First, the entry's own level-branch is itself the line between two members of the family: surrogation (one decision-maker, cognitive) versus Goodhart (many actors, system-incentive) are sibling faces of the same parent, not the same concept at different scales — and confusing them sends the wrong fix, since the cognitive form yields to salience while the system form yields to incentive redesign. Second, the cross-domain instances (clinical endpoints, policy targets, citation metrics) genuinely exhibit the substitution, but they exhibit it as instances of the parent pattern, not as instances of "surrogation" as defined in managerial accounting; describing a surrogate clinical endpoint as "surrogation" imports the managerial framing onto a phenomenon that the parent already covers more cleanly. The clean boundary, then: literal transfer of surrogation across managerial-evaluation contexts wherever an accountable evaluator holds a measure for a construct under load; and beyond that, the reach belongs to proxy/target substitution (with Goodhart and the surrogate-endpoint problem as its other faces), not to the named cognitive effect. (See Structural Core vs. Domain Accent.)

Examples¶

Canonical¶

Choi, Hecht, and Tayler's experiments (2012, 2013) provide the defining laboratory demonstration. Participants acted as managers evaluating performance under a balanced-scorecard system in which numerical measures (for example, a customer-satisfaction survey score) stood in for a broader strategic construct (actual customer satisfaction). The key manipulation was whether managers were also given a strategy map that made the underlying construct explicit and salient alongside the scorecard. Managers given the scorecard alone showed markedly more surrogate behavior — treating and rewarding the measure as though it were the construct itself — than managers given both the scorecard and the strategy map. Making the construct visible reduced the substitution, confirming that under the accountability-and-load conditions of the task the mind defaults to the concrete number unless the abstract construct is actively kept present.

Mapped back: Customer satisfaction is the strategic construct; the survey score is the measure; the evaluation-and-reward setup is the accountability structure creating the load conditions. Managers with the scorecard alone exhibited the construct-into-measure collapse and its behavioural fallout. That the strategy map reduced surrogation is the salience remedy validated experimentally.

Applied / In Practice¶

The Net Promoter Score (introduced by Fred Reichheld in a 2003 Harvard Business Review article as "the one number you need to grow") is a widely deployed real-world site of surrogation. NPS was designed as a proxy for customer loyalty — a single survey item asking how likely a customer is to recommend the company. In practice many organizations came to treat "raising the NPS number" as the objective itself: staff are bonused on it, dashboards track it, and frontline employees openly solicit high scores ("I'll get a 10 if you..."), while the underlying loyalty the number was meant to represent recedes from view. The drift is typically not cynical gaming but the ordinary result of managers, accountable for a concrete number under time pressure, mentally equating the score with the loyalty it stands for.

Mapped back: Customer loyalty is the strategic construct and the NPS survey item is the measure; bonuses and dashboards are the accountability structure. Managers under routine pressure exhibit the construct-into-measure collapse — treating the score as loyalty itself — with score-soliciting as the behavioural fallout. Because it is non-adversarial, the fix is the salience remedy (triangulating other loyalty signals), not anti-gaming audits.

Structural Tensions¶

T1: Indispensable measure versus corrupting measure (operationalization is the double edge). A strategic construct — customer satisfaction, engagement, strategic value — is abstract and partly tacit, so it must be operationalised into a concrete measure to be managed, tracked, and rewarded at all. But that same operationalisation is what invites surrogation: the moment a number stands in for the construct, an accountable manager under load can let the number become the construct. The tension is that the measure is at once the only handle on the construct and the mechanism of its betrayal — you cannot manage what you do not measure, yet measuring installs a proxy that tends to displace the thing measured. There is no metric-free escape, because refusing to operationalise forfeits accountability entirely; the drift is a cost of the very legibility that makes strategic management possible, not a defect a better process eliminates. Diagnostic: Is the measure still functioning as a tracked handle on the construct, or has the operationalisation itself become the operative goal?

T2: Non-adversarial drift versus deliberate gaming (same symptom, opposite remedy). When a metric stops tracking its construct, the reflex is to suspect manipulation — someone cheating the number — and gaming calls for tighter controls and audits. Surrogation identifies a prior, non-adversarial failure: under load, time pressure, and accountability for a measurable number, the concrete measure simply becomes the operative handle while the construct recedes, with no one deciding to cheat. The tension is that the two produce the same visible fallout — effort flowing to what the metric rewards, the construct's tacit content lost — yet demand opposite fixes: controls for gaming, salience for surrogation, and each remedy is useless against the other failure. Worse, they co-occur, so an analyst who sees goal-displacement cannot read the cause off the behaviour alone; the diagnosis turns on the absence of intent and the presence of load, which is inferred, not observed, making the cause-branch that routes the fix the hardest step. Diagnostic: Is the metric-drift here driven by deliberate gaming (route to controls) or by non-intentional load-induced substitution (route to salience)?

T3: Salience remedy versus the load that defeats it (the cure fights its own cause). The prescribed fix is to keep the construct actively present — pair a strategy map with the scorecard, hold narrative accountability alongside numerical, triangulate measures so none installs itself as the stand-in. The experimental signature confirms it: managers given both a strategy map and a scorecard surrogate less than those given the scorecard alone. But surrogation is caused by cognitive load, time pressure, and accountability for measurable performance — exactly the conditions under which maintaining construct salience is hardest and most likely to be dropped. The tension is that the remedy demands sustained effort in precisely the regime that erodes effort: the manager most in need of holding the construct in view is the one whose load has already defaulted them to the number. So the fix is not self-sustaining; salience must be actively re-maintained against the standing pressure that keeps reinstating the substitution. Diagnostic: Under this actor's actual load, is the construct being actively kept salient, or has the pressure that causes surrogation also quietly defeated the salience remedy?

T4: A good metric is no protection versus the improve-the-metric reflex (the failure is in the head, not the number). Surrogation is not a defect in the measure's quality: it does not say the metric is a bad proxy or fails construct validity. A perfectly reasonable, well-validated measure surrogates the moment it is held as the goal rather than as a stand-in. The tension is that this cuts directly against the natural response. When a measure drifts from its purpose, the reflex is to refine it — a better proxy, more dimensions, a cleaner survey — but since the failure lives in the mental treatment of the number, not in the number, improving the metric leaves surrogation untouched and can even deepen it by lending the measure more authority to be mistaken for the construct. So the more defensible the metric, the more readily it installs itself as the construct's equal, and effort spent perfecting it is effort not spent keeping the construct salient. Diagnostic: Is the proposed fix improving the measure's quality (which does not address surrogation) or preserving its status as a mere proxy (which does)?

T5: Autonomy versus reduction (a named cognitive effect or one face of proxy/target substitution). Surrogation is specifically the cognitive-bias face of a broader family — proxy/target substitution: a measure adopted for its correlation with a construct comes to function as the construct, after which the two decouple. That parent recurs across distinct substrates as genuine co-instances: Goodhart's law is its system-incentive face (many actors collectively optimise a proxy until it decays), the surrogate-endpoint problem its measurement-causal face (a biomarker standing in for a clinical outcome), KPI gaming its operational face. Surrogation's own cargo — the scorecard, the strategic-construct vocabulary, the accountability-and-bonus setting, the load-induced default inside one head — is managerial-evaluation furniture that does not travel. The tension is between a sharply characterised cognitive effect with its own experimental signature and remedy, and the recognition that its cross-domain lesson ("a proxy installed for its correlation displaces the target, then they part ways") belongs to the parent, whose Goodhart and clinical-endpoint faces take different fixes. Diagnostic: Resolve toward the proxy/target-substitution parent (and its Goodhart or surrogate-endpoint face) when the substitution is collective or causal; toward surrogation when it is one accountable evaluator's load-induced substitution.

Structural–Framed Character¶

Surrogation sits at framed-leaning on the structural–framed spectrum — a cognitive effect whose underlying mental mechanism is real but whose named identity is managerial-evaluation furniture. Its structural pull is genuine but narrow: surrogation names a real psychological event, the load-induced default in which a concrete measure becomes the operative referent inside one head, and that substitution is an empirically demonstrated cognitive regularity (the strategy-map-versus-scorecard experimental signature), which gives it a mechanistic, non-arbitrary core and makes its within-domain transfer recognition rather than analogy — NPS, EPS, and the h-index cases are one and the same cognitive collapse, not separate pathologies. But four criteria pull toward framed. Human_practice_bound is high: surrogation is defined inside the managerial-evaluation substrate — a single evaluator, accountable for a measurable number that stands in for a partly tacit strategic construct — and it dissolves without that structure of constructs, scorecards, and accountability; there is no surrogation without something to be accountable for. Institutional_origin is pronounced: the entry is a construct of the managerial-accounting literature (Choi, Hecht & Tayler, building on Lipe & Salterio), with its scorecard-and-strategy-map experimental apparatus. Evaluative_weight is moderate: surrogation names a drift, a failure of a measurement system to serve its purpose — carrying a defect valence, though softened by being explicitly non-adversarial (no one is blamed for gaming). And vocab_travels is low: construct, measure, scorecard, strategic value, and accountability are managerial-evaluation terms pinned to that substrate.

The portable structural skeleton is a single one: proxy/target substitution — a measure adopted for its correlation with a construct comes to function as the construct, after which the two decouple. That skeleton genuinely recurs across substrates as sibling faces — Goodhart's law (system-incentive), the surrogate-endpoint problem (measurement-causal), KPI gaming (operational) — which is exactly why it does not lift "surrogation" off the framed-leaning position: the cross-domain reach belongs to the umbrella proxy/target-substitution parent that surrogation instantiates, and not to the named effect, while surrogation's distinctive content (the scorecard, the strategic-construct vocabulary, the accountability-and-bonus setting, the load-induced default inside one head) is precisely the managerial-evaluation accent that stays home. Its character: a moderately defect-flagged, practice-bound cognitive effect, structural in the proxy-substitution skeleton it borrows from its parent family but framed by the managerial-evaluation apparatus — the construct/measure distinction, the scorecard, the accountability setting — that makes it specifically the cognitive-bias face called surrogation.

Structural Core vs. Domain Accent¶

This section decides why surrogation is a domain-specific abstraction and not a prime, and it carries the case for its domain-specificity — there is no separate section for that.

What is skeletal (could lift toward a cross-domain prime). Strip the managerial evaluation and a thin relational structure survives: a proxy adopted for its correlation with a target comes to function as the target, after which the two decouple. The pieces that travel are abstract — a target of real interest, a stand-in installed because it tracks the target, a substitution in which the stand-in becomes the operative referent, and a subsequent divergence between proxy and target. That skeleton is genuinely substrate-portable, which is exactly why it recurs as the general proxy/target-substitution pattern that surrogation is one face of — a family whose system-incentive face is Goodhart's law (many actors collectively optimise a proxy until it decays), whose measurement-causal face is the surrogate-endpoint problem (a biomarker standing in for a clinical outcome), and whose operational face is KPI gaming. But that shared core is the structure surrogation shares — it is not what makes surrogation distinctive.

What is domain-bound. Almost every distinctive thing about the concept is managerial-evaluation furniture and none of it survives extraction intact: the strategic construct / measure distinction and its scorecard vocabulary; the accountability structure of evaluations, bonuses, and reputations tied to a number; the load conditions (cognitive load, time pressure, accountability for measurable performance) under which a concrete number becomes a cheaper handle than an abstract construct; the non-adversarial, single-head character that distinguishes it from gaming and from the system-level dynamic; and the salience remedy (strategy map alongside scorecard, narrative accountability, triangulated measures) with its strategy-map-versus-scorecard experimental signature. These are the worked vocabulary, the instruments, and the empirical cases the discipline actually studies. The decisive test: remove the accountable evaluator holding a measure for a construct under load — the scorecard, the bonus, the strategic construct — and there is no surrogation, only a bare proxy-replaces-target pattern; a clinical biomarker or a collectively-gamed policy target genuinely exhibits the substitution, but as an instance of the parent, not of surrogation as managerial accounting defines it. What is left once the evaluation scaffolding is stripped is a looser thing with no construct/measure vocabulary, no load-induced cognitive default, and no salience fix.

Why this does not clear the prime bar. A prime is a relational structure whose vocabulary travels and whose cross-domain transfer is recognition of the same mechanism, not analogy. Surrogation's transfer is bimodal. Within managerial accounting, organisational behaviour, and judgment research the mechanism travels intact across every site — balanced-scorecard management, NPS/CSAT service incentives, public-sector waiting-time targets and league tables, EPS-driven CEO pay, OKR key results, the h-index in academic evaluation — because each runs the identical cognitive event (a construct collapsing into a measure inside one accountable head under load), read off the same proxy-or-goal diagnostic, attributed by the same non-adversarial cause-branch, and corrected by the same salience remedies; that is recognition, not analogy. Beyond managerial evaluation it does not travel under its own name: the clinical-endpoint, policy-target, and collective-optimisation cases exhibit the substitution as instances of the parent pattern, so describing a surrogate clinical endpoint as "surrogation" imports the managerial framing onto a phenomenon the parent already covers more cleanly. And when the bare structural lesson is needed cross-domain — a proxy installed for its correlation displaces the target, then they part ways — it is already supplied in more general form by the proxy/target-substitution parent, with its Goodhart (system-incentive) and surrogate-endpoint (measurement-causal) siblings taking their own fixes. The cross-domain reach belongs to that umbrella family; "surrogation," as named, carries the scorecard, the accountability setting, and the single-head load-induced default that do not and should not travel.

Relationships to Other Abstractions¶

Current abstraction Surrogation Domain-specific

Parents (1) — more general patterns this builds on

Surrogation is a kind of Proxy-Target Divergence Prime

Surrogation is proxy-target divergence specialized to an accountable evaluator who mentally substitutes a concrete measure for its strategic construct under load.

Hierarchy path (1) — routes to 1 parentless root

Surrogation → Proxy-Target Divergence → Proxy–Target Fidelity → Representation → Abstraction

Not to Be Confused With¶

Goodhart's law. The system-level sibling face: when many actors collectively optimize a proxy, it decays as a measure of the construct. Surrogation is the cognitive substitution inside one decision-maker's head — the measure becoming the operative referent under load, with no collective optimization required. They are sibling faces of one proxy/target parent and take different fixes: surrogation yields to salience, Goodhart to incentive redesign. Tell: is the failure a single evaluator mentally equating measure with construct (surrogation), or many actors driving a proxy until it stops tracking the target (Goodhart)?
The surrogate endpoint problem. The measurement-causal sibling face, from clinical trials: a biomarker standing in for a clinical outcome, where an intervention's off-pathway biology makes the two diverge. Surrogation is a cognitive event, not a causal-mediation fact about interventions; describing a surrogate clinical endpoint as "surrogation" imports the managerial framing onto a phenomenon the parent covers more cleanly. Tell: is the divergence in a person's mental treatment of a metric (surrogation), or in a drug's effect failing to carry from biomarker to outcome (surrogate endpoint problem)?
KPI gaming / metric manipulation. The operational, adversarial face: someone deliberately games the number at the construct's expense. Surrogation is explicitly non-adversarial — no one decides to manage the measure instead of the strategy; the number simply becomes the handle under load. This is why the fix diverges: gaming calls for controls and audits, surrogation for salience. Tell: is there intent to work the metric (gaming, route to controls), or an unintentional load-induced substitution with no cheating (surrogation, route to salience)?
Goal displacement (means-ends inversion). The organizational-sociology phenomenon in which a means becomes an end in itself — a near-relative, since surrogation produces goal-displacement toward what the metric rewards. But goal displacement is the broader outcome (an instrumental procedure becoming terminal), while surrogation is the specific cognitive mechanism — a measure standing in for its construct — that is one route to it. Tell: is the claim that a procedure has become an end generally (goal displacement), or specifically that a measure has replaced the construct it proxies in an evaluator's mind (surrogation)?
The proxy / target-substitution parent (with goodharts_law as a face). The substrate-neutral pattern — a measure adopted for its correlation with a construct comes to function as the construct, then the two decouple — that carries the cross-domain lesson across test scores, GDP, and KPIs. Surrogation is the cognitive-bias face of this family. Tell: is there a single accountable evaluator holding a measure under load (surrogation), or the substitution occurring collectively/causally/operationally? If the latter, name the parent (or its Goodhart/surrogate-endpoint face), not surrogation. (Treated more fully in a later section.)

Neighborhood in Abstraction Space¶

Surrogation sits in a moderately populated region (43^rd percentile for distinctiveness): it has near-neighbors but no dense thicket of look-alikes.

Family — Organizational Power & Coordination Drift (9 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-07-12