Incomplete Contract¶
Core Idea¶
An incomplete contract is the structural pattern of a rule-set that deliberately leaves some contingencies unspecified, in the knowledge that exhaustive enumeration is impossible, prohibitively costly, or unenforceable, and that some other mechanism — residual control rights, renegotiation, default doctrines, trusted arbiters, norms, or good faith — will fill the gap when the unspecified contingency arises. The defining commitment is the acknowledged gap plus a gap-filling regime: incomplete contracts are not bad contracts but honest ones about the limits of foresight, language, verifiability, or enforcement.
Three features distinguish the pattern from "a contract with mistakes." Bounded foresight: the parties recognise that the space of future states is too large or uncertain to specify exhaustively. Verifiability gap: even when a contingency can be described, some variables — effort, intent, quality, future value — are hard for a third party to observe, so writing them in is pointless without an oracle. Designed residual authority: the contract specifies who decides when the unspecified case arises, even if it cannot specify what they will decide. The structural signature is therefore gap + gap-handler + governance of the handler: the gap is the unwritten zone, the handler is the rule, person, doctrine, or norm that activates when the zone is entered, and the governance is whatever constrains the handler from abusing the discretion. Reading any rule-set as "this part is specified, this part is left to a handler, this is how the handler is held to account" is the structural move the prime supplies. That move is abstractly general, but the prime's vocabulary and theory are deeply law-and-economics-bound, and its non-human instances arrive largely as analogies — which is why it sits at the framed end of the spectrum.
How would you explain it like I'm…
The On-Purpose Blank
Honest About The Future
Gap And Gap-Handler
Structural Signature¶
a rule-set covering some contingencies — an acknowledged gap of unspecified contingencies — the bounded-foresight and verifiability limits that force the gap — a designated gap-handler — a governance regime constraining the handler — the discretion-abuse failure modes
A rule-set is an incomplete contract when the following hold:
- A partially specified rule-set. An agreement, code, or specification that fixes outcomes for some contingencies explicitly.
- An acknowledged gap. A zone of contingencies deliberately left unspecified, recognised as such rather than overlooked — the unwritten part is a design choice, not a mistake.
- The forces that make the gap necessary. Bounded foresight (the state space is too large to enumerate), a verifiability gap (some variables — effort, intent, quality — cannot be observed by a third party), and enforcement cost together guarantee that residual gaps cannot be closed.
- A designated gap-handler. A named rule, person, doctrine, or process — residual control rights, a court, a regulator, a runtime, a learned prior — that activates when the unspecified contingency arises and decides what happens, even though the contract could not.
- A governance regime. Whatever constrains the handler from abusing its discretion: fiduciary duty, precedent, appeal, audit, reciprocity, market discipline.
- The discretion-abuse failure modes. Because the handler holds residual authority, the characteristic risks are hold-up, opportunism, handler overreach, capture, and drift.
These compose into one move: replace the impossible task of specifying every contingency with the bounded task of locating the gaps, designating a handler for them, and governing that handler.
What It Is Not¶
- Not a contract with mistakes. An incomplete contract leaves gaps deliberately and knowingly, having decided exhaustive specification is impossible or not worth it; a defective contract has gaps by error. The presence of a designated gap-handler is what marks the difference.
- Not an
interface. An interface specifies a complete contract over the cases it covers — every legal call has a defined behaviour; an incomplete contract explicitly admits cases it does not cover and routes them to a handler with discretion. - Not
optionality. Optionality is a held right to act later; an incomplete contract is an acknowledged gap in specification plus a regime for filling it. The gap is not anyone's option — it is an unspecified zone someone is empowered to resolve when it arises. - Not
authority_delegation_under_uncertainty. Delegation hands a decision right to an agent within a defined scope; an incomplete contract is the broader structure — gap, plus handler, plus governance of the handler — of which a delegated decision right is one possible gap-handling mechanism. - Not
discretionalone. Discretion is the latitude a handler enjoys in the gap; the incomplete contract is the full apparatus that creates the gap, assigns the handler, and constrains that discretion so it is not abused. - Common misclassification. Reading any unspecified term as a drafting failure. Catch it by asking whether the gap is acknowledged and a handler with governance is designated; if so, the incompleteness is a rational design choice, not an oversight.
Broad Use¶
The skeleton recurs across substrates, anchored in human institutions. In economics and contract theory it is the transaction-cost origin of incomplete contracts, the property-rights view of ownership as residual control, and relational contracting in long-term supply relationships. In law and governance it is constitutional vagueness ("due process," "reasonable") delegating gap-filling to courts, common-law evolution as ongoing gap-filling, and statutory gaps filled by regulatory agencies. In diplomacy it is constructive ambiguity — deliberate vagueness that secures agreement despite irreconcilable underlying preferences — and arbitration clauses naming a handler without resolving the dispute. In organisations it is open-ended employment where the employer holds residual decision rights over unnamed tasks. In software engineering it is specifications that intentionally underspecify (ordering of side effects, behaviour under malformed input, timing) and rely on conventions, defaults, or designated handlers such as the compiler or runtime. In biology it is genomes that underspecify developmental outcomes, with environmental cues and epigenetic state filling the gap. In AI alignment it is reward functions and prompts as incomplete specifications whose gap-handlers are learned priors and fallback rules — the specification problem being fundamentally contract incompleteness. These instances share the three structural features and the same family of failure modes — hold-up, opportunism, overreach by the handler, drift — though the non-human cases are read as analogies to the legal-economic original.
Clarity¶
The prime makes visible that the unwritten parts of any rule-set are not omissions but design choices. Once a practitioner sees this, the question shifts from "did we write down enough?" — a losing race — to "who handles the gap, and how is that handler constrained?" This reframes contract drafting, statutory interpretation, API design, and treaty negotiation around the same three knobs. The pattern also exposes the hidden completeness premise in naive specification work: the assumption that a thorough enough specification can close all gaps. The prime explains why that is structurally impossible — bounded foresight, verifiability limits, and enforcement costs guarantee residual gaps — and where to put one's energy instead, namely on the handler and its governance. The clarifying force is to convert an open-ended drafting ambition into a bounded design problem: locate the gaps, designate a handler, and constrain it.
Manages Complexity¶
The prime replaces the impossible task of specifying every contingency with the tractable task of designing the gap-handler. The combinatorial blow-up of contingency enumeration is absorbed by delegating to a process or a person who handles the open set at runtime — the same complexity-management move that virtual memory makes for address spaces, that exception handling makes for control flow, and that constitutional vagueness makes for legislation: trade exhaustive enumeration for principled delegation. The cost is a new question — who gets the residual authority and how is it constrained — but that is one design problem instead of infinitely many. The management payoff is that an unbounded specification burden collapses into a single, bounded governance design: rather than anticipating every future state, the rule-set names an authority to decide the unanticipated and a regime to keep that authority honest.
Abstract Reasoning¶
The pattern enables a set of substrate-independent questions. Where are the gaps? — which contingencies are unspecified, unverifiable, or unenforceable. Who is the gap-handler? — residual control rights, default doctrines, courts, regulators, the runtime, learned priors. What constrains the handler? — fiduciary duty, precedent, reciprocity, market discipline, audit, a principal-agent contract. What is the failure mode of unconstrained discretion? — hold-up, regulatory capture, scope creep, undefined-behaviour bugs, reward hacking. Can the gap be partially closed? — more specific clauses, better monitoring, machine-verifiable specs — and at what drafting or verification cost? These questions transfer across contract law, constitutional design, software specification, alignment, and organisational charter design. The reasoner asks, of any rule-set: where is it deliberately silent, who decides in that silence, and what holds that decider to account?
Knowledge Transfer¶
The intervention catalog carries portable design knobs. Default-rule design: choose defaults the majority of parties would have negotiated to, reducing the handler's load. Residual-control assignment: give residual rights to the party whose ex-post investment matters most, which applies to ownership, custody, project authority, and model-output control. Constraining-process design: bind the handler with precedent, appeal, reviewability, audit, or arbitration, applying to judges, regulators, project managers, and AI policy heads. Renegotiation-cost reduction: cheaper renegotiation tolerates more incompleteness, more expensive renegotiation pushes toward more upfront specification. And constructive ambiguity: when full specification would block agreement, deliberate incompleteness can be rational, applying to treaties, coalition agreements, and product roadmaps. The role mappings are direct: gap ↔ unspecified contingency / undefined behaviour / unfilled developmental outcome, handler ↔ residual-rights holder / court / runtime / learned prior, handler constraint ↔ fiduciary duty / precedent / oversight board / appeal, failure mode ↔ hold-up / capture / undefined-behaviour bug / reward hacking. A municipality contracting a 30-year wastewater plant makes three incomplete-contract choices — gap location (rates specified, future regulatory compliance left open), handler design (a regulatory adjustment commission plus reversion of residual control to the city), and handler constraint (state-board review, arbitration, a seven-year renegotiation right) — and the identical three-knob design recurs in API contracts (versioning policy plus a breaking-change committee plus an RFC process) and AI deployment (reward function plus fine-tuning plus an oversight board). Because the vocabulary and theory are inherently law-and-economics-bound and the non-human instances are analogies, the transfer is the deliberate carrying of a legal-economic design discipline — gap, handler, governance — into other substrates, rather than the recognition of a substrate-free structure already present in them.
Examples¶
Formal/abstract¶
Take the property-rights theory of the firm (the Grossman–Hart–Moore framework) as the rigorous instance, since it is where the prime received its sharpest formal treatment. Two parties — a buyer and a supplier — must make relationship-specific investments whose value depends on a future transaction whose terms cannot be fully written down. The partially specified rule-set fixes price and quantity for foreseeable states; the acknowledged gap is the open set of contingencies (a novel design change, an unanticipated cost shock) that bounded foresight and a verifiability gap — effort and quality are observable to the parties but not provable to a court — make impossible to enumerate. The theory's central move is the prime's designated gap-handler: ownership is defined as residual control rights, the authority to decide whatever the contract left open. Whoever owns the critical asset decides in the gap, and the governance regime is the surrounding law of property and the threat of inefficient renegotiation. The discretion-abuse failure mode the prime names appears precisely as hold-up: the residual-rights holder can, ex post, threaten to withhold cooperation to extract the other party's specific investment, which distorts ex-ante investment incentives. The intervention the framework derives is exactly the prime's: assign residual control to the party whose non-contractible investment matters most, so the unavoidable discretion lands where it does least damage — a result about who should own what that follows directly from reading the relationship as gap-plus-handler.
Mapped back: The property-rights model instantiates every role — specified terms, an acknowledged unverifiable gap, ownership as the residual-control handler, property law as governance, hold-up as the discretion-abuse mode — and shows the prime converting an impossible complete-contracting problem into a tractable question of authority allocation.
Applied/industry¶
Consider a 30-year municipal wastewater-treatment concession and a software API's versioning policy as two applied instances. The wastewater contract specifies current rates and service levels but cannot specify future regulatory compliance over three decades — the acknowledged gap forced by bounded foresight. The city designs a gap-handler: a regulatory adjustment commission empowered to reset terms when rules change, with residual control reverting to the city. The governance regime constraining that handler is layered — state-board review, mandatory arbitration, and a seven-year renegotiation right — directly answering the prime's "what holds the decider to account?" The forecast failure modes are hold-up (the private operator threatening service degradation to extract rate increases) and handler capture (the commission co-opted by the operator), both anticipated at design time by the constraining process. An API contract runs the identical three-knob design: the spec deliberately underspecifies behaviour under future feature requests (the gap), names a breaking-change committee plus an RFC process as the handler, and binds it with a published versioning policy and deprecation timeline as governance — guarding against the analog of overreach, an unannounced breaking change that strands downstream consumers. The transferable intervention: stop racing to specify every future state, and instead locate the gap, name the handler, and constrain it.
Mapped back: The concession and the API policy both run the prime end-to-end — specified core, acknowledged gap, designated handler, governing constraints, forecast discretion-abuse modes — confirming that the same gap/handler/governance discipline carries from a public-works contract to a software interface, even as the vocabulary remains law-and-economics in origin.
Structural Tensions¶
T1 — Specification versus Delegation. The prime trades exhaustive enumeration for principled delegation to a handler, but the boundary between what to write down and what to leave open is itself a hard choice. The tension is scopal: specify too much and you pay drafting and verification cost on contingencies that never arise; specify too little and you hand the handler more discretion than you can govern. The failure mode is the naive completeness premise — believing a thorough enough specification can close every gap — which loses the impossible race against bounded foresight. Diagnostic: ask whether a contingency is cheaply verifiable and likely; specify those, delegate the rest deliberately rather than by omission.
T2 — Handler Discretion versus Handler Constraint. The gap-handler must have enough authority to decide the unanticipated, yet enough constraint not to abuse it — and these pull against each other. The tension is that every increase in discretion needed for flexibility is an increase in the surface for hold-up, capture, and overreach. The failure mode is granting residual authority without a governance regime, so the handler extracts rents (the residual-rights holder threatens to withhold cooperation ex post). Diagnostic: for each grant of discretion, name the corresponding constraint — fiduciary duty, appeal, audit, reciprocity — and ask whether it actually binds before relying on the handler.
T3 — Ex-Ante Investment versus Ex-Post Authority. Residual control is exercised ex post, but the parties make relationship-specific investments ex ante, in anticipation of how that authority will be used. The tension is temporal: the handler's future discretion distorts present incentives. The failure mode is hold-up — a party under-invests because it foresees the residual-rights holder expropriating the returns once the investment is sunk. Diagnostic: assign residual control to the party whose non-contractible ex-ante investment matters most, and ask how anticipated ex-post discretion is shaping behaviour today, not just when the gap is entered.
T4 — Deliberate Ambiguity versus Latent Disagreement. Constructive ambiguity lets parties with irreconcilable preferences reach agreement by leaving the contested point unwritten — incompleteness as a feature. The tension is that the unresolved disagreement does not vanish; it is deferred to whenever the gap is entered. The failure mode is mistaking the agreement for consensus, so the latent conflict detonates at the worst moment, when the unspecified contingency finally arises and each party reads the silence in its own favour. Diagnostic: ask whether the gap papers over a genuine clash of preferences; if so, ensure the handler and its governance can withstand the deferred fight rather than assuming it was settled.
T5 — Gap-Handler Authority versus Verifiability Limit. A handler is designated precisely because some variables — effort, intent, quality — cannot be verified by a third party, yet the handler must still decide based on them. The tension is measurement: the gap exists because the relevant facts are unprovable, so the handler decides on signals it cannot fully confirm. The failure mode is designing a handler (a court, an arbiter) as if it could observe what is structurally unverifiable, then being surprised when its decisions are arbitrary or gameable. Diagnostic: ask what the handler can actually observe versus what it must rule on, and design proxies or monitoring to narrow that gap rather than assuming the handler sees clearly.
T6 — Renegotiation Cost versus Specification Cost. Incompleteness is rational when renegotiation is cheap and irrational when it is expensive — the right amount of upfront specification depends on what re-opening the contract later will cost. The tension is that the two costs trade off and are both uncertain at drafting time. The failure mode is leaving large gaps under the assumption renegotiation will be smooth, then facing prohibitive hold-up when one party can block re-agreement — or over-specifying to avoid renegotiation that would have been trivial. Diagnostic: estimate the cost and bargaining power asymmetry of future renegotiation; cheap, symmetric renegotiation licenses more incompleteness, expensive or lopsided renegotiation demands more upfront detail.
Structural–Framed Character¶
Incomplete Contract sits near the far framed end of the structural–framed spectrum — framed, aggregate 0.9, one of the most framed primes in the catalog. An abstractly general skeleton exists — a partially specified rule-set with an acknowledged gap, a designated handler, and governance of that handler — but the prime's vocabulary and theory are bound to a specific economic-legal lineage (Grossman–Hart–Moore), and its non-human instances arrive as deliberate analogies, so four of the five diagnostics read at full weight.
Walk them. Vocabulary travels (1.0): "residual control rights," "hold-up," "verifiability gap," "constructive ambiguity," "renegotiation" are contract-theory terms, and a software or biology instance is recognized as an incomplete contract only by importing that lexicon. Institutional origin (1.0): the construct is rooted in the institutions of contracting and property law; it does not merely originate in a discipline but presupposes contracting parties, courts, and ownership. Human-practice-bound (1.0): the gap, the handler, and the governance regime all presuppose human institutions — agreements, fiduciary duty, arbitration; the genomic and AI-alignment cases are analogies to that human-practice original. Import vs. recognize (1.0): reading a genome or a reward function as an "incomplete contract" overlays the gap/handler/governance frame onto it rather than recognizing a structure already self-evidently present. The single concession is evaluative weight (0.5): incompleteness is framed as honest rather than defective, carrying a mild positive evaluative tilt, but not a strong one. Four full points plus one half-point land exactly at the 0.9 aggregate and the framed label — the gap-plus-governed-handler discipline is genuinely portable, but it carries its law-and-economics frame everywhere it goes.
Substrate Independence¶
Incomplete Contract is a weakly substrate-independent prime — composite 2 / 5 on the substrate-independence scale, one of the more substrate-bound primes in the catalog. There is an abstractly general skeleton — a partially specified rule-set with an acknowledged gap, a designated handler, and governance of that handler — and the domain breadth reaches a fair distance: it is read into contract theory and property-rights economics, constitutional vagueness and common-law evolution in law, constructive ambiguity in diplomacy, open-ended employment in organizations, deliberately underspecified behaviour in software, developmental underspecification in genomes, and reward-function specification in AI alignment. But both the structural-abstraction and transfer bands are honestly low, and they cap the composite. The structural abstraction is weak (a 2) because the gap, the handler, and the governance regime all presuppose human institutions — contracting parties, courts, fiduciary duty, arbitration, ownership as residual control — so the vocabulary and theory are bound to a specific economic-legal lineage (Grossman–Hart–Moore) rather than to a medium-neutral relation. The transfer evidence is correspondingly weak (a 2) because the non-human instances arrive as deliberate analogies: reading a genome or a reward function as an "incomplete contract" overlays the gap/handler/governance frame onto it rather than recognizing a structure already present, so what transfers is a legal-economic design discipline carried into other substrates, not a substrate-free pattern recognized across them. A genuinely general skeleton wrapped in a heavy, human-practice-bound legal-economic frame with analogy-grade cross-domain reach earns a 2.
- Composite substrate independence — 2 / 5
- Domain breadth — 3 / 5
- Structural abstraction — 2 / 5
- Transfer evidence — 2 / 5
Relationships to Other Primes¶
Parents (1) — more general patterns this builds on
-
Incomplete Contract is a kind of Contract
An incomplete contract is, structurally, a contract (a rule-set fixing outcomes for some contingencies) plus the distinctive move of an acknowledged gap + designated gap-handler + governance. contract is the canonical genus; incomplete_contract is the specific kind that deliberately leaves gaps. Phase-C already records the link to contract. The file severs it from interface (complete contract over its cases), delegation (only the decision-right piece), and optionality (a held right) — none is the genus; contract is. Good conviction: the is-a holds even though the prime is heavily framed (law/econ), because every instance is literally a contract or rule-set.
Path to root: Incomplete Contract → Contract → Interface → Boundary
Neighborhood in Abstraction Space¶
Incomplete Contract sits in a sparse region of abstraction space (85th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely rather than landing on a neighbor.
Family — Boundaries, Rules & Discretion (14 primes)
Nearest neighbors
- Contract — 0.71
- Consistency — 0.70
- Imputation — 0.69
- Discretion — 0.68
- Credible Commitment — 0.67
Computed from structural-signature embeddings · 2026-06-14
Not to Be Confused With¶
The most instructive contrast is with interface, because an interface
is, in a sense, the complete-contract ideal that incomplete contracts
depart from. An interface specifies, for every legal interaction it admits, a
defined behaviour: callers may rely on the contract holding across all
covered cases, and any case outside the interface is simply not a legal
call. An incomplete contract takes the opposite stance toward uncovered
cases: it admits that contingencies will arise which it has not specified,
and instead of declaring them illegal it designates a handler — a person,
doctrine, or renegotiation process — to resolve them when they occur. The
deep difference is in how each treats the unanticipated: the interface
pushes it out of scope; the incomplete contract pulls it in and assigns
authority over it. A practitioner who models a relationship as an interface
when it is really an incomplete contract will be blindsided when an
unspecified contingency arrives and there is, in fact, no defined behaviour —
only a fight over who decides.
It is also distinct from authority_delegation_under_uncertainty, with
which it is intimately related but not identical. Delegation is the act of
handing a decision right to an agent, to be exercised within some scope
under conditions the principal cannot fully foresee. That is exactly the kind
of mechanism an incomplete contract uses to fill its gaps — "when the
unspecified case arises, party X decides." But the incomplete contract is the
larger structure: it comprises the acknowledged gap, the assignment of a
handler (of which a delegated decision right is one form), and the
governance that constrains the handler from abusing the discretion.
Delegation supplies the residual-authority component; the incomplete contract
also supplies the gap-identification and the accountability scaffolding.
Treating the two as the same loses the governance dimension — the checks that
keep the gap-filler honest — which is precisely where incomplete contracts
succeed or fail.
A third confusion is with optionality. Both point toward an unresolved
future, but they allocate it differently. Optionality is a right held by a
party to choose later whether and how to act; its value lies in keeping a
favourable choice open. An incomplete contract's gap is not a right anyone
holds — it is a zone of non-specification that no one has yet been given
license to exploit, and the contract's design problem is to decide who gets
authority over that zone and how they are held to account. Reading a gap as
an option mistakes an unallocated contingency for a granted entitlement, and
can hand one party de facto control over a zone the contract never meant to
cede.
For a practitioner the through-line is that the incomplete contract is defined by its honesty about the limits of specification plus its machinery for governing what fills the gap. An interface denies the gap exists by ruling it out of scope; delegation supplies only the decision-right piece; optionality treats the open zone as someone's entitlement. Keeping the full triad — gap, governed handler, accountability — in view is what distinguishes a robust incomplete contract from each of these narrower neighbours.
Solution Archetypes¶
No catalogued solution archetypes reference this prime yet.