Framework Overview

The evaluation of digital public infrastructure (DPI), digital public goods (DPGs), and public-sector AI stops at the resource layer. Decades of capability-approach, digital-divide, and state-capability research show that resources do not become outcomes automatically.

The Public Capability Stack (PCS) conceptualizes this conversion through six layers.

Layer 6: Life Outcomes

Substantive human freedoms and capabilities: what people can actually be and do differently, reached through service journeys.

Layer 5: Public Capabilities

Routine, owned, and budget-supported public service operations that deliver quality-assured technology functions.

Layer 4: Governance & Safeguards

Institutional mechanisms for dispute resolution, linguistic justice, data privacy, and oversight safeguards.

Layer 3: Conversion Factors

Personal, social, and environmental translators: workflow fit, service trust, literacy, and accessibility.

Layer 2: Demand Signals

Articulated, aggregated, and validated public needs translated into supply-side system commitments.

Layer 1: Digital Resources

Physical and software components: Digital Public Infrastructure (DPI) platforms, registries, standards, and AI models.

Earlier presented on this site as "Digital Capabilities for Life Outcomes (DCLO)"; the framework was consolidated and renamed in 2026.

#### The Central Mechanism: Demand Signalling & Matching

The Public Capability Stack operates through a five-step demand-signalling mechanism paired with its supply-side counterpart:

The 5-Step Demand-Signalling Flow

01
Articulation

Local actors voice concrete needs

Failure: Missing publics
02
Aggregation

Grouping signals into administrative queues

Failure: Signal fragmentation
03
Validation

Authenticating and checking eligibility

Failure: Verification locks
04
Translation

Refining needs to technical specifications

Failure: Translation gap
05
Commitment

Acquiring budget lines & service mandates

Failure: Orphaned pilots

Supply-Side Matching & System Alignment

The demand pipeline matches with a companion supply track: Solution Discovery $\rightarrow$ Registry Formation $\rightarrow$ Coalition Readiness. Digital Public Infrastructure is fundamentally a matching problem: capability conversion succeeds or fails at the point where validated social demand is paired with credible technical supply under a binding governance commitment.

The framework is a middle-range, configurational, and falsifiable framework designed to explain both capability conversion and structural failure (mimicry, missing publics, lock-in, and unsafe automation).

#### Evidence Base

The framework is built from two anonymised institutional cases from the international public sector: a service-level case on AI-enabled language services and an institutional-level case on system-wide technology governance. The analysis is evidence-first, mapping patterns across several hundred primary records before selecting theoretical frames.

---

LAYER 6: LIFE OUTCOMES (PATHWAYS) Agency, health, livelihood opportunities LAYER 5: PUBLIC CAPABILITIES Routine, owned, & accountable service functionings LAYER 4: GOVERNANCE & SAFEGUARDS Linguistic justice, confidentiality safeguards LAYER 3: CONVERSION FACTORS Personal, social, & environmental translators LAYER 2: DEMAND SIGNALLING Aggregation, validation, translation, commitment LAYER 1: DIGITAL RESOURCES (DPI) Digital Public Goods, software ecosystems, AI models
The Public Capability Stack (PCS) conceptualizes that digital systems become public capabilities only when demand signals, conversion factors, governance safeguards, and transition pathways align to convert digital resources into trusted, accessible, quality-assured, and sustainable service capabilities.

The Four Papers

PaperTitleDescriptionStatus
:---:---:---:---
Paper 0Between supply and demand: a systematic review of how public institutions convert digital-service needs into DPI, DPGs, and capabilitiesPRISMA-2020-guided systematic review across three search arms; the research gap is tested against four pre-registered falsification rules rather than asserted.In progress — identification complete
Paper 1The Public Capability Stack: specifying a capability-conversion framework and its demand-signalling mechanismThe conceptual paper of the dissertation. Layer operationalisations, the five-step demand signalling mechanism alongside its supply-side counterpart, falsification conditions, and the cross-case synthesis.Design stage
Paper 2Trusted enough for official use? AI language services and the conversion factors of multilingual public serviceQualitative case analysis combined with an embedded pre-registered factorial vignette experiment testing which institutional assurances (human review, confidentiality safeguards, disclosed quality) causally shift professionals' willingness to rely on AI outputs for official use.Design stage
Paper 3From ideas to infrastructure: demand–supply matching in system-level technology governanceInstitutional case study using theory-testing process tracing across bounded capability-conversion episodes, including one observed prospectively.Design stage
Measurement Note: A conditional, exploratory measurement component — aggregate indicator and panel data, possibly assembled into the DCLO screening composite — sits outside the four-paper core. It expands into a paper only if a named-dataset inventory gate passes; otherwise it is reported as an exploratory annex. Its inferences are limited to trends and lagged associations, never outcome causation.

---

Methods & Timeline

#### Methods My research utilizes an evidence-first sequencing methodology: - PRISMA-guided review with pre-registered gap-falsification tests (Paper 0). - Two-stage interpretive coding and theory-testing process tracing with explicit evidentiary tests. - Embedded factorial vignette experiment to causally map trust thresholds (Paper 2). - Exploratory indicator analysis using descriptive screening composites. - Expert stress-testing interviews (12–15 participants) under a domain-based sampling frame (post-ethics only). - Extended field immersion in rural India beginning mid-December 2026 serving as cross-cutting construct validation (village, district, and state are strictly anonymised).

Everything is version-controlled in a private research compendium made available to the committee.

#### Timeline Milestones Late 2026: Synopsis finalisation, ethical RERB package approval, and Paper 0 execution. Early 2027: Paper 0 submission and rural field immersion. Mid-2027: Factorial vignette experiment rollout and expert interviews. Late 2027: Dissertation assembly and defence preparation (defence targeted December 2027).

---

Indicator Explorer (Public Sandbox)

This sandbox presents aggregate state-year secondary indicator data from the open dual-track dashboard. No materials from the primary research corpus are stored or processed here. The DCLO Index is a PCS-informed formative composite indicator used as a descriptive screening instrument. It does not measure the PCS framework (which is specified as a configurational process framework, not a latent construct), and life outcomes are discussed only as mediated plausibility pathways — never as direct effects of digital resources.

Configurational Index Score (DCLO) by State

State-level public capability composite score (descriptive screening composite, averaged over time series)

Lagged Associations Plot

Lagged TWFE associations mapping DCLO composite components to life outcome indicators

*Exploratory associations — not causal estimates.*
Methodological Note: Chart 1 represents composite formative index weights (DCLO) derived across access, literacy, service enablement, trust, and outcomes. Chart 2 displays standardized lagged Two-Way Fixed Effects (TWFE) regression coefficients (β) with 95% confidence intervals of the DCLO Index predicting subsequent life outcomes, representing exploratory statistical associations while controlling for state-level and time fixed-effects.