This is Article 3 of Phase 1 of the Polity thought-leadership series on go-to-market, capital, and valuation for Web3 infrastructure. Article 1 defined go-to-market as the activation graph with regulatory permission as a first-class node. Article 2 chose between bootstrapping-loop archetypes and specified MVNS per class. This article takes up the curve that determines whether the chosen loop produces a network or a collapse: the per-class trajectory along which incentive-driven participation must be replaced by utility-driven participation before the protocol-scheduled subsidy taper completes.
|
TL;DR. The substitution curve is per class, not aggregate. Each participant class has its own incentive-decay schedule and its own utility-accumulation rate; the network is self-sustaining only when every class has crossed its own curve. The curve is a design object – chosen, instrumented, and managed against per-class telemetry – not a forecast. An incentive that is not designed to be replaced by utility is not a growth lever; it is a treasury depletion schedule. |
Articles 1 and 2 established the activation graph and the bootstrapping-loop choice. They specify how a Web3 infrastructure network is supposed to activate. They do not, on their own, settle whether activation will hold once the protocol-scheduled incentives that drove early participation begin to taper. That question is the substitution curve, and it is the variable on which most networks of the cohort have turned.
Participation in a Web3 infrastructure network has two components: an incentive-driven component, paid from a finite treasury and decaying on a protocol-defined schedule, and a utility-driven component, accumulating only as fast as the network matures. The substitution curve is the trajectory along which one replaces the other. It is a per-class object – one curve per participant class – and the network is self-sustaining only when every class has crossed.
Most projects in the cohort never define the curve and therefore never have the means to know whether they are crossing it. They launch with incentives, observe participation, interpret participation as product-market fit, and continue paying. When the treasury thins and emissions taper, participants leave. This is characterised retrospectively as a market cycle. It is, more often, a substitution that was never designed.
This article makes the curve formal – per-class crossover times, an explicit pre-crossover failure region under aggressive emission tapering, and an illustrative quantified example – so that the question “has utility replaced incentive?” can be asked precisely and answered class by class. The curve, defined deliberately and measured per class, is the bridge from the activation strategy of Article 2 to the capital derivation of Article 4.
This is the decisive curve most Web3 projects never cross.
|
Formal Definition – Substitution Curve Sᵢ(t). For each participant class i, define the participation value to that class as the sum of two components, each a function of network time t: Pᵢ(t) = Iᵢ(t) + Uᵢ(t) where Iᵢ(t) is incentive-driven participation value and Uᵢ(t) is utility-driven participation value. Iᵢ(t) is bounded above by the per-class incentive budget and decays on a schedule set by the protocol; Iᵢ(0) is large, Iᵢ(∞) is the maintenance-only baseline. Uᵢ(t) accumulates as a function of the network’s realised state – the depth of upstream classes, the count of completed flows, the regulatory scope (R_j active in the relevant jurisdictions) – and is bounded above by the network’s steady-state utility ceiling for class i. Crossover time tᵢ* is the smallest t at which Uᵢ(t) ≥ Iᵢ(t); for class i, the substitution is “crossed” once t > tᵢ*. Properties: (i) pre-crossover collapse – if for some t < tᵢ* the protocol-scheduled Iᵢ(t) tapers below the participant’s reservation level for class i before Uᵢ(t) reaches that level, Pᵢ(t) drops below reservation and participation collapses; this is the cohort failure mode in §3.1’s stylised illustration; (ii) the network achieves substitution iff a t* exists such that t > tᵢ* ∀ i ∈ V; (iii) Uᵢ(t) is endogenous to the activation strategy – faster MVNS achievement (Article 2 §2.3) shifts Uᵢ(t) leftward, lowering tᵢ*; (iv) crossing is necessary but not sufficient for steady state – sustained operation also requires the network to remain on the post-crossing branch under exogenous shocks (regulatory regime change, competitor entry, demand-side macro shifts) that can pull Uᵢ(t) back below the post-crossover Pᵢ floor. |
Schematic 3 – Incentive-to-utility substitution curve, with illustrative example: I(t) = 87 e⁻²ᵗ + 8 (so I(0) = 95) and logistic Uᵢ(t) per class with named steepness parameters. Parameters are illustrative not estimated; see §3.1 illustration discussion.
Quantified illustration. To make the dynamic concrete rather than to estimate it, consider an illustrative instance of the curve over a normalised runway t ∈ [0, 1]. An incentive schedule of the form I(t) = 87 e⁻²ᵗ + 8 begins at exactly 95 (87 + 8) at t = 0 and approaches a maintenance baseline of 8 over the runway, with a half-life of approximately 0.35 on the decaying component. Per-class utility is modelled as a logistic Uᵢ(t) = Mᵢ / (1 + e⁻ᵏⁱ⁽ᵗ⁻ᵗᵐⁱ⁾), with class-specific steepness, midpoint, and ceiling: Providers (kᵢ = 8.0, tₘᵢ = 0.30, Mᵢ = 98), Merchants (kᵢ = 7.0, tₘᵢ = 0.55, Mᵢ = 88), Clients (kᵢ = 6.0, tₘᵢ = 0.85, Mᵢ = 72). Solving Uᵢ(tᵢ*) = Iᵢ(tᵢ*) numerically for each class gives the implied crossover times t*[Prov] ≈ 0.32, t*[Merch] ≈ 0.52, and t*[Client] ≈ 0.76; network MVNS – defined as the smallest t at which all three curves have crossed – is gated by the latest-crossing class. These values are illustrative, not estimated from any specific cohort observation. Their purpose is to make the structure of the curve visible – that crossover times are class-ordered, that the latest-crossing class binds, and that an aggressive incentive decay produces a failure region for the latest-crossing class first. Sensitivity analysis on illustrative parameters is suppressed here and confined to Article 4 §4.2, where the parameters are explicitly directional cohort anchors rather than illustrative shapes. Article 5 walks through one worked example with named parameter sources for a hypothetical European tokenisation operator.
The incentive budget runs on a schedule; utility does not. Incentives come from a finite treasury and decay according to the protocol rules. Utility, by contrast, accumulates only as fast as the network matures – which is largely determined by the graph, the MVNS target, and the bootstrapping loop. If the incentive decay rate outpaces the utility accumulation rate, the network collapses when the subsidy ends. Matching the two rates is the core design problem.
|
Key Insight. An incentive that is not designed to be replaced by utility is not a growth lever. It is a treasury depletion schedule. |
Stylised illustration – yield-farming substitution-curve collapse, 2020–22. A pattern repeatedly observed across the 2020–22 yield-farming and points-programme cycle, in dozens of protocols across multiple chains. The cleanest named cases on the public record include: Iron Finance (June 2021), where TITAN’s bank-run-shaped collapse from $64 to near-zero in 24 hours followed an emission schedule whose post-taper retention had not been designed; Wonderland (TIME, January 2022), where a step-down in the protocol’s emission rate triggered TVL decay from over $750m to under $50m within months; and the SushiSwap-vs-Uniswap vampire-attack reversal of mid-2020, in which Sushi’s emission-pulled liquidity rotated back to Uniswap once the relative incentive narrowed. The shape across all three is consistent with the substitution-curve framework: peak TVL preceded the emission peak by days; post-emission TVL settled near pre-launch baseline rather than at any intermediate plateau; per-class participation tracked emissions one-for-one. The substitution curve was not crossed for any participant class – participation was incentive-driven from t=0 to t=end. Real utility did not accumulate, because the cohort’s designs did not specify what utility, for which class, was supposed to accumulate.
|
Boundary note – failure modes outside the framework. The MANTRA / OM token collapse of April 2025 is sometimes cited as cohort evidence of substitution-curve collapse. On the public record it is more accurately a token-distribution and exit-liquidity event over a single weekend, with concentrated insider/early-holder positions liquidating into thin off-exchange liquidity, rather than the slow per-class participation decay over months that the substitution-curve framework formalises. The substitution-curve framework as set out here does not purport to characterise the specific April 2025 mechanism, and this article makes no claim about the cause of that event beyond what is in the public record. Token-distribution-design failure is a category Phase 2 will address as a first-class subject; in Phase 1 it is conceded as outside the substitution-curve scope. |
Utility replacement is participant-specific. Operators are retained by different forms of utility from Clients, Providers, and Merchants. Each class has its own substitution curve. The project has not crossed the curve until every class’s curve has crossed. Aggregating them into a single “are we there yet?” question is the same category error Article 1 warned against. Telemetry-completeness commitments – consistent measurement of per-class participation, incentive accrual, and utility realisation at protocol level – are what make these curves observable and therefore manageable; Article 4 treats them as inputs to the capital requirement derivation.
|
Constraint – Substitution Curve Integrity. A Web3 infrastructure project SHALL maintain an explicit, per-participant-class substitution schedule, with incentive-decay and utility-accumulation rates treated as matched design variables. Aggregated substitution models – a single curve for the whole network – are structurally insufficient and should not be relied upon in capital or valuation decisions. |
|
Failure Mode – Substitution Curve Collapse. Condition: incentive decay rate exceeds utility accumulation rate for one or more participant classes; Iᵢ(t) tapers below the participant’s reservation level for class i before Uᵢ(t) reaches that level. Mechanism: participants are retained by subsidy in proportion that exceeds the utility the network has yet generated for them; when subsidy tapers on schedule, participation collapses to whatever level utility alone can sustain – typically near zero in cold-start configurations. Observable symptom: per-class participation drops sharply within weeks of each scheduled emission step-down; characterised by the team as “market conditions” or “rotation”; in fact a substitution that was never designed. Cohort prevalence: repeatedly observed across the 2020–22 yield-farming and points-programme cycle (Iron Finance, Wonderland, the Sushi/Uniswap reversal). Cohort tell – peak participation preceding the emission peak by days and post-emission participation settling near pre-launch baseline rather than at any intermediate plateau – is consistent across instances. |
|
Consequence. Failure to cross the substitution curve is not a temporary setback; it is, for most networks in the relevant cohort, a structural terminal state. |
The substitution curve is not a forecast. It is a design object. Where projects in the cohort failed, they did so not because the curve was unobservable but because no one had specified what the curve was supposed to look like, per class, before launch. This section sets out the four design choices the operator must make explicitly – incentive-schedule shape, utility-accumulation rate, instrumentation, and decision points – and the questions each forces.
The incentive function Iᵢ(t) has shape parameters – initial level, decay rate, and maintenance floor – that are choices, not constants. Three shapes recur in the cohort. An exponential schedule, of the form used in the §3.1 illustration, decays continuously and is structurally appropriate where the operator has high confidence in the utility-accumulation rate it is matched against. A step-down schedule, where Iᵢ(t) is held flat across discrete epochs and reduced at scheduled boundaries, is structurally appropriate where the network can absorb supply shocks at the boundary and where the boundary itself is a useful coordination event. A milestone-gated schedule, where the next reduction is conditional on observed Uᵢ(t) reaching a defined threshold rather than on calendar time, is structurally appropriate where the utility-accumulation rate is uncertain and the cost of misjudging the decay rate is high.
The choice between the three is per class, not per project. A regulated Provider class, whose retention is structural and whose response time to incentive change is months, can absorb a step-down schedule. A Client class, whose retention is episodic and whose response time is weeks, cannot – the boundary becomes the cohort tell of substitution-curve collapse. The cohort error is to apply a single schedule shape to all classes; the corresponding discipline is to write the schedule per class against the response time and structural retention of that class.
Uᵢ(t) is the harder side of the curve, because it is endogenous to the activation strategy and must be estimated rather than scheduled. Three estimation methods are tractable in practice and one is not.
Comparable-network estimation. Map the per-class utility-accumulation rate observed in a comparable network onto the project’s own classes, with adjustments for graph-topology and constraint-set differences. The method is tractable when a comparable exists at sufficient depth; it is brittle where the comparable was activated through a different bootstrapping loop, because Uᵢ(t) is path-dependent on loop choice. As with MVNS estimation (Article 2 §2.2), the data conditions for executing this cleanly are met for almost no project at the planning stage.
Bottom-up flow estimation. Decompose per-class utility into the underlying flows it depends on – for a Provider class, this is typically transaction count, fee yield, and counterparty count – and estimate the rate at which each flow accumulates as a function of upstream-class depth. The method is more disciplined than the comparable approach but requires the operator to be explicit about which flows count as utility for which class. The act of writing the decomposition is itself diagnostically useful: a class for which utility cannot be decomposed into observable flows is a class for which the curve cannot be measured, which means the cohort error is already present.
Calibration through controlled subsidy withdrawal. Once a class is partially activated, the operator can run a bounded-duration subsidy reduction on a defined sub-segment, observe the participation response, and calibrate Uᵢ(t) at the observed t against a counterfactual. The method is the most accurate of the three, but is available only after launch and only on classes where bounded experimentation does not damage the activation path.
The intractable approach. The remaining option – observing aggregate participation and inferring U(t) by subtraction from I(t) – is the dominant cohort approach and the one that fails. It conflates classes, hides path-dependence, and produces an aggregate curve that can read as crossing while every individual class’s curve is still below its incentive line. The operator that relies on it has no instrument for noticing when the network is failing in real time.
A curve that cannot be observed cannot be managed. Per-class telemetry is therefore not an analytics layer added on after launch; it is a precondition for the curve to function as a design object at all. Three measurements are essential.
Per-class participation. Each participant must be classified into exactly one class at protocol level, with class membership observable from on-chain or operator-recorded state. An aggregated wallet count cannot be decomposed back into class participation after the fact; the decomposition must be present in the data model from t=0.
Per-class incentive accrual. The portion of value flowing to each class that comes from the protocol’s incentive budget must be separable from the portion that comes from realised network activity. The two are recorded together as wallet inflows; in the cohort, the failure to separate them is the leading cause of confusion between participation and adoption.
Per-class utility realisation. The flow-level decomposition specified in the bottom-up estimation must be carried forward as a live measurement, not a launch-time estimate. Uᵢ(t) at any t is then a sum the operator can read directly rather than a quantity inferred from aggregate activity.
Once the schedule is chosen, the rate is estimated, and the instrumentation is in place, the operator faces three discrete decisions that recur, per class, along the runway. Each is answerable from the curve and unanswerable without it.
Decision 1 – emission step-down. At each scheduled or milestone-gated boundary, the operator confirms that Uᵢ(t) for the class in question has reached the level required for retained Pᵢ(t) post-step-down. Where the gating is by milestone, the step-down is deferred until the threshold is met; where the gating is calendar-based, the operator must decide between proceeding on schedule, deferring on observed under-performance, or re-budgeting the incentive tail. The discipline is that the decision is made against per-class data rather than aggregate participation.
Decision 2 – class-level reallocation. Where one class is approaching its crossover materially ahead of plan and another is materially behind, the curve makes the reallocation case explicit. Reducing incentive on the leading class earlier than scheduled, and reinvesting in the lagging class, is a move the cohort consistently fails to make – not because it is hard to execute but because it is unobservable in projects without per-class instrumentation. With the curve in hand, the move is mechanical.
Decision 3 – abandonment of a class. Where a class’s observed Uᵢ(t) is materially below the trajectory required to cross before the runway expires, and where bottom-up examination indicates that the underlying flows are not accumulating, the curve makes the abandonment case visible. Narrowing the network – deferring a class, dropping a jurisdiction, removing a use case – is preferable to running the curve to its terminal state. The cohort failure mode is to defer this decision until the treasury has run; the framework’s discipline is to make it on the curve, while there is still capital to redirect.
|
Key Insight. The substitution curve is a design object, not a forecast. The operator chooses its incentive schedule, estimates its utility-accumulation rate, instruments the per-class measurements that make it observable, and uses those measurements to drive emission step-downs, class-level reallocations, and class abandonments along the runway. None of this is available without the per-class object. |
A bootstrapping loop, executed with a defined MVNS target and an explicit substitution curve, produces a real adoption profile – participant counts by class, engagement depth, subsidy-dependence, and expected time-to-utility-dominance. The profile answers, for each participant class: how many do we need, how fast do they arrive, how long they are subsidised, and at what rate utility replaces incentive? These four inputs, combined with infrastructure and operational costs, produce round size, burn profile, and milestone structure.
That derivation is the subject of Article 4. Article 5 then walks the full sequence – graph through round size – for one worked operator example.
The cold-start problem rarely yields to marketing intensity alone. It is more reliably addressed by treating it as a dependency-resolution problem with constrained, modelable, falsifiable inputs: the activation graph (with R_j as a graph node), the MVNS definition, the bootstrapping loop, and the substitution curve. Projects that do this work – rigorously, before the round, not during it – have a coherent activation plan and a derivable capital requirement; projects that do not, on the cohort evidence, generally do not.
|
Key Insight. Activation is where most Web3 infrastructure projects fail – not at the level of the architecture, but at the level of the cold-start problem treated as a distribution question rather than a dependency-resolution question. The projects that cross the substitution curve are the ones that designed it. |
These structures materially constrain outcomes; they do not dominate execution quality, behavioural variability, or external conditions. The framework moves the activation problem from intractable to analysable; it does not move it to determined.
Empirical illustrations are stylised and synthesised from publicly observable patterns; they describe no specific entity. The 2020–22 yield-farming referents (Iron Finance, Wonderland, the Sushi/Uniswap reversal) are public-record events; the broader regulated-finance cohort is documented in the Methodological Appendix V.3.7.0.
Next: Article 4 derives capital requirements end-to-end through the five-step chain.
This article is published for informational and educational purposes only. It does not constitute investment, legal, tax, or financial advice, an endorsement of any product or security, or any offer or solicitation. References to named projects in the cohort and to named public events (token launches, recapitalisations, discontinuations, regulatory milestones) are factual public-record observations recorded for analytical purposes; they are not assessments of those projects’ merits or prospects, and the framework expresses no view on the future trajectory of any individual project. Readers should conduct their own due diligence and consult qualified professionals before acting on any of its content.
Polity is a B2B technology vendor that develops substrate infrastructure for the operators of regulated digital finance networks. Polity does not provide investment advice, custody, crypto-asset services under Regulation (EU) 2023/1114 (MiCA), or any other regulated activity; operator-customers are the regulated entities for any service delivered on Polity substrate. Polity has a commercial interest in the adoption of this framework (operator-customers typically build on Polity substrate); this conflict of interest is disclosed for transparency and is not cured by it.
Forward-looking statements in this corpus – including the §4.6b research hypotheses – are based on current expectations and may prove materially incorrect. Past cohort patterns are not a guarantee of future outcomes. The §4.6b hypotheses are stated for falsification, not as investment recommendations, financial promotions, or solicitations within any applicable regulatory regime.
Published from the European Economic Area; not directed at any person in a jurisdiction where its publication would be contrary to local law. This corpus is not a marketing communication relating to any specific crypto-asset within the meaning of Article 27 of Regulation (EU) 2023/1114 (MiCA); it is a category-level analytical framework. Personal data of natural persons (typically corporate officers of named cohort projects) is processed under legitimate-interest grounds (GDPR Article 6(1)(f)) and confined to the public commercial record.
Third-party names and sources are public-record events; their inclusion implies no commercial relationship, endorsement, partnership, or affiliation. Characterisations of causes, mechanisms, or outcomes of named cohort projects are this framework’s analytical interpretation of the public record, not statements of fact about any entity’s strategy or condition. Inclusion in the companion Methodological Appendix implies no commercial relationship.