Multi-Level Governance of Renewable Energy Transitions Through the Viable System Model: A Hybrid Evidence-Based Framework

Abstract

Multi-Level Governance (MLG) frameworks effectively diagnose the complexity of regional renewable energy transitions but lack operational mechanisms for institutional implementation. This study develops a hybrid evidence-based architecture that integrates the Viable System Model (VSM) with computational intelligence and objective multi-criteria decision analysis. Methodologically, a systematic review following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) protocol of 339 peer-reviewed articles (2015–2025) feeds a Latent Dirichlet Allocation (LDA) model that extracts K = 30 strategic topics (semantic coherence $C_v$ optimized over K = 5–40), operationalizing System 4 environmental sensing. In parallel, a 1 km² pixel-based spatial model integrates a National Conflict Index (INC, 2019–2024) with technical feasibility layers (Global Wind Atlas v4.0, Solargis, Servicio Geológico Colombiano) and applies CRITIC (CRiteria Importance Through Intercriteria Correlation) objective weighting and TOPSIS (Technique for Order of Preference by Similarity to Ideal Solution) prioritization (System 3 control). Robustness is confirmed through $\pm 10$–$20\%$ weight perturbation (Spearman > 0.92). Empirically, the framework reveals a localization paradox: approximately 68% of optimal wind zones (>9 m/s at 100 m hub height) in the Colombian Caribbean overlap with the highest national conflict quartile, narrowing a theoretical capacity exceeding 100 GW (50 GW offshore wind, 30 GW onshore wind, 42 GW solar PV, 1.17 GW geothermal) to roughly 24 GW of governance-viable capacity. Scenario calibration (Accelerated 80/20, Balanced 50/50, Justice-Oriented 30/70 technical/conflict weighting) demonstrates that System 5 normative orientation materially reshapes territorial prioritization. The framework advances VSM from a qualitative diagnostic metaphor to a reproducible governance architecture for high-variety regional contexts.

1. Introduction

1.1. Governance Gap in Renewable Energy Transitions

Renewable energy transitions are commonly framed as technological substitution processes centered on decarbonization targets and capacity expansion. Yet the principal constraint is not technological feasibility but institutional coordination under spatial and multi-level complexity. Renewable infrastructures are territorially embedded: wind corridors intersect biodiversity zones, grid expansion confronts land-use conflicts, and solar clusters overlap with fragile institutional environments. What appears as a technical deployment challenge frequently reveals a structural misalignment between national strategic coherence and territorially situated implementation realities. This spatial friction is empirically evident in regions such as La Guajira, Colombia, which exhibits world-class onshore wind regimes (frequently exceeding 9 m/s at 100 m hub height) yet suffers project paralyzation due to top-quartile socio-political conflict exposure—most visibly the indefinite suspension of the Windpeshi wind farm by its developer in 2023.

Target-driven policy frameworks such as aggregate “Net Zero” commitments risk obscuring this coordination deficit by privileging macro-level outcomes over systemic integration mechanisms [1]. As Ison and Straw argue, governance systems not designed to process environmental variety become structurally fragile under Anthropocene turbulence [2]. The governance gap in renewable transitions is therefore cybernetic in nature: insufficient institutional capacity to absorb spatial heterogeneity and institutional asymmetry across scales.

1.2. Multi-Level Complexity and the Missing Regional Coordination Layer

Multi-level governance arrangements attempt to reconcile centralized coordination with decentralized autonomy. However, renewable energy systems intensify structural tensions between strategic coherence and territorial legitimacy. In particular, the regional layer—where transmission corridors, ecological systems, and inter-municipal infrastructures intersect—remains weakly operationalized in transition governance studies [3,4]. Without a formally empowered coordination mechanism at this meso-level, governance oscillates between hierarchical rigidity and fragmented localism. Addressing this structural gap requires a recursive architecture capable of aligning downward operational autonomy with upward strategic coherence.

1.3. Research Gap and Methodological Novelty

To sharpen the scientific contribution and avoid overstating novelty, four distinct gaps are identified across the intersected domains:

• Multi-Level Governance (MLG): While the national and local levels have received ample scholarly attention, the regional (meso) coordination layer remains weakly operationalized as a structurally complete arena for renewable transitions [3,4].
• Viable System Model (VSM): A recent systematic review explicitly identifies the lack of integration with quantitative and AI-based tools as a major limitation of current VSM research, which often remains diagnostically rich but operationally under-instrumented [5].
• Topic Modeling (LDA): Despite successful applications to extract latent thematic structures in energy contexts [6,7], LDA is rarely formalized as an anticipatory sensing mechanism embedded within governance cycles.
• Multi-Criteria Decision Analysis (MCDA): The renewable energy literature is dominated by subjective weighting methods, with critical underutilization of objective approaches such as CRITIC (CRiteria Importance Through Intercriteria Correlation) in uncertain decision environments [8,9].

The novelty of this study is therefore not the mere aggregation of these tools, but their cybernetic coupling: LDA is positioned as System 4 anticipatory intelligence and CRITIC–TOPSIS as System 3 objective allocation, directly responding to the research agenda articulated by [5] to augment the VSM’s qualitative architecture with computational intelligence and objective decision modeling.

1.4. Research Objective and Contributions

This study addresses the following research question: How can cybernetic governance architectures integrate computational intelligence and objective multi-criteria optimization to manage multi-level renewable energy transitions?

The objective is to design and empirically illustrate a scalable hybrid governance architecture grounded in the Viable System Model, operationalizing LDA as System 4 environmental sensing and CRITIC–TOPSIS as System 3 objective resource allocation within a recursive multi-level structure. The contributions are fourfold:

• Theoretical Contribution: A parametrized VSM architecture addressing the missing regional coordination layer in renewable transitions.
• Methodological Innovation: Integration of topic modeling (LDA) and objective MCDA (CRITIC–TOPSIS) within a unified cybernetic governance framework.
• Empirical Illustration in a High-Variety Context: Diagnostic application in Colombia, characterized by ecological diversity and institutional asymmetry.
• Alignment with Eco-Friendly Regional Transitions: A viability-based approach linking technical efficiency, territorial legitimacy, and normative sustainability identity.

Figure 1 provides an integrated overview of the proposed architecture, which the remainder of the paper develops and operationalizes.

2. Theoretical Framework

2.1. Multi-Level Governance in Energy Transitions

Multi-Level Governance (MLG) is a central lens for explaining why renewable energy transitions unfold unevenly across territories. It maps actor constellations, institutional layers, and policy interactions spanning national, regional, and local arenas, clarifying how vertical steering and horizontal networks co-produce transition dynamics. Yet MLG remains primarily diagnostic: it describes coordination problems and scale interactions, but rarely specifies the structural conditions under which governance can remain stable while adapting to spatial heterogeneity.

This limitation becomes visible when renewable deployment is treated as a planning pipeline rather than a high-variety governance problem. Infrastructure is territorially fixed and politically contested: siting decisions intersect with land-use regimes, ecosystem constraints, and legitimacy structures that differ sharply across regions. As a result, governance failures are often expressed as oscillations between centralized rigidity that preserves formal coherence but suppresses place-based adaptation, and fragmented localism that enables experimentation but undermines system-wide alignment. The empirical consequence is structural misalignment—the inability of institutions to coordinate across scales where technical and socio-ecological systems intersect.

A recurring weakness in transition governance scholarship is the limited operationalization of the regional scale. National strategies define targets and incentives, while municipal arenas absorb conflicts and implementation costs. Between them, the meso-level concentrates key spatial couplings of transitions: transmission corridors, renewable clusters, watershed and biodiversity systems, and inter-municipal externalities. Yet this scale frequently lacks an explicit mandate or operational design [3]. Figure 2 synthesizes this structural disconnection.

These constraints motivate the use of an architecture-oriented theory of governance: Organizational Cybernetics and the Viable System Model (VSM), introduced here as a design grammar for building multi-level governance systems that retain coherence under turbulence while preserving operational autonomy at lower levels.

2.2. The Viable System Model as a Cybernetic Architecture

The Viable System Model (VSM) is a structural theory of the necessary and sufficient functions required for a system to maintain identity while adapting to environmental change. VSM treats governance as a viability problem: survival depends on how effectively the system balances internal stability and external adaptation under high complexity. VSM has been positioned as a meta-language for diagnosing structural pathologies and redesigning institutions in complex socio-ecological environments [10]; its distinctive contribution is its recursive logic, by which viable systems must be viable at multiple nested levels.

In this hybrid framework, the five VSM functions are operationalized in measurable, energy-transition terms:

• System 1 (Operations): Territorially embedded deployment alternatives modeled as 1 km² pixel units (renewable generation assets, grid reinforcement, distributed initiatives).
• System 2 (Coordination): Mechanisms that dampen oscillations among System 1 units—interoperability standards, planning protocols, conflict-mediation arrangements—typically operating at the regional (departmental) scale.
• System 3 (Control): Resource allocation, performance monitoring, and constraint management, instrumented in this study by CRITIC–TOPSIS (Technique for Order of Preference by Similarity to Ideal Solution) objective prioritization.
• System 4 (Intelligence): Environmental scanning of policy, technology, and legitimacy signals, instrumented by LDA topic modeling on a 339-article governance corpus.
• System 5 (Policy/Identity): Normative closure that defines the constitutional identity of the transition, operationalized through scenario calibration of decision weights.

Practical examples support institutional adoption: regional authorities can act as System 2 intermediaries that share resources, harmonize licensing, and avoid duplication [11]; VSM-based interventions have demonstrated viability gains in academic and public-sector departments [12] and in Colombian community self-governance under resource asymmetry [13]. Figure 3 illustrates the recursive embedding of Systems 1–5 and the meso-level metasystem that this framework adopts.

VSM therefore contributes a prescriptive complement to MLG: it does not merely locate where coordination fails, but specifies the functional capacities that must exist at each recursion level for governance to remain viable.

2.3. Requisite Variety and Sustainability Governance

The theoretical bridge between VSM and renewable transitions is Ashby’s Law of Requisite Variety: only variety can absorb variety. Renewable transitions confront institutions with multidimensional environmental variety—biophysical constraints, socio-political contestation, heterogeneous territorial capacities, and shifting regulatory and market conditions. Governance failure is often a symptom of variety deficit: institutions are not structured to perceive, process, and respond to the scale and speed of environmental change [2].

Recent Organizational Cybernetics reviews indicate that VSM applications have frequently remained qualitative, limiting their scalability in complex governance networks where evidence-based prioritization is required [5]. For renewable transitions, a purely conceptual VSM is insufficient: viability requires that System 3 and System 4 be operationally instrumented so that governance can attenuate decision complexity and amplify intelligence complexity. MLG provides the cartography of multi-level complexity, whereas VSM provides the design logic for viability under turbulence; the subsequent sections operationalize this proposition.

3. Hybrid Evidence-Based Methodology

3.1. Research Design Overview

This study advances a hybrid evidence-based methodological architecture aligned with VSM requirements. The design operationalizes two complementary cybernetic functions: environmental intelligence amplification (System 4) and decision attenuation through objective prioritization (System 3). Latent Dirichlet Allocation (LDA) translates large-scale governance discourse into structured signals [6,7], while CRITIC–TOPSIS provides an auditable prioritization engine [8,9]. Topic-derived signals shape the attention structure of governance; MCDA outputs structure the allocation logic. Figure 4 summarizes this evidence-based workflow end to end.

3.2. Demarcation of Computational, Inferential, and Proposed Elements

To preserve empirical transparency and methodological reproducibility, the framework explicitly distinguishes what is algorithmically computed, analytically inferred, and proposed as institutional design:

• Computed: The K = 30 LDA topic structure with optimized semantic coherence ($C_v$); the CRITIC objective weights derived from data dispersion and inter-criteria correlation; and the TOPSIS closeness coefficients ($C{C}_i$) at 1 km² resolution.
• Inferred: The mapping of LDA topics to System 4 governance domains, and the translation of TOPSIS rankings into System 3 priorities. These inferences are bounded by deliberative expert validation (Section 3.4) rather than free researcher judgment.
• Proposed: The multi-level governance architecture (Systems 1, 2, 5), recommending that regional authorities (Departamentos) assume the coordination mandate, with national entities—the Mining and Energy Planning Unit (UPME), the National Environmental Licensing Authority (ANLA), and the Energy and Gas Regulation Commission (CREG)—acting as planning, licensing, and regulatory metasystems.

3.3. Systematic Review and PRISMA Protocol

3.3.1. Registration Statement

The evidence-synthesis component of this study was conducted as an internal sensing module of the proposed Viable System Model-based governance framework rather than as a stand-alone systematic review. Accordingly, the review was not prospectively registered in PROSPERO, OSF, or any other systematic-review registry, since at the time of corpus construction the protocol was embedded within the broader hybrid evidence-based methodology described in Section 3.1 and its primary purpose was to feed the LDA-based System 4 sensing layer (Section 3.4) and the CRITIC–TOPSIS spatial decision layer (Section 4). To safeguard methodological transparency and reproducibility, the full search equation, screening logs, eligibility decisions, and retained bibliographic metadata are reported in a manner consistent with the spirit of the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 guidelines [14]; a completed PRISMA 2020 checklist (Supplementary File S1) and the full list of the 339 included studies (Supplementary File S2; Table S1) are provided as Supplementary Material (see Supplementary Materials section).

The environmental intelligence corpus was constructed through a systematic review process following PRISMA. Searches were executed in Scopus and Web of Science using a pre-specified query at the intersection of renewable transitions, multi-level governance, and coordination/conflict dynamics:

TITLE-ABS-KEY((“renewable energy transition*” OR

“energy transition governance”) AND

(“multi-level governance” OR regional OR municipal) AND

(policy OR coordination OR conflict)) AND

PUBYEAR > 2014 AND PUBYEAR < 2026

The 2015–2025 window captures the post-Paris evolution of transition governance research. The query returned n = 1247 records; 271 duplicates were removed via the bibliometrix package; title/abstract screening excluded n = 641; and full-text eligibility assessment retained 335 database records. Together with 4 additional studies identified through citation chasing and Project GENTE Phase 4 expert input, this yielded a final corpus of n = 339 studies (285 articles and 54 reviews; Figure 5). Inclusion criteria were peer-reviewed status, presence of DOI and abstract, document types restricted to “article” or “review”, and substantive engagement with at least one of the three thematic axes. Full search strings, screening logs, and retained metadata are reported in the Supplementary Materials.

3.3.2. Summary of Included Studies (

Table 1. Top 10 most globally cited studies included in the PRISMA-based corpus (n = 339). Total citations and per-year citation rates are taken from the harmonized Biblioshiny report (MostGlobCitDocs sheet of BiblioshinyReport-Eq1.xlsx, Supplementary File S3). The full list of the 339 included studies is provided in Table S1 of the Supplementary Materials.

#	First Author (Year)	Total Cit.	Source	Scope	Governance Signal Family (System 4)
1	Gallo [15] (2016)	590	Renew. Sustain. Energy Rev.	Global	Regulatory coordination/policy-instrument mix
2	Cantarero [16] (2020)	412	Energy Res. & Soc. Sci.	Latin America	Spatial justice and conflict
3	Szulecki [17] (2018)	289	Environmental Politics	Europe	Energy democracy/System 5 normative anchoring
4	Levenda [18] (2021)	239	Energy Res. & Soc. Sci.	N. America/Global	Spatial justice and conflict
5	Yildiz [19] (2014)	221	Renewable Energy	Germany/Europe	Participation and community energy
6	Wahlund [20] (2022)	218	Energy Res. & Soc. Sci.	Europe	Participation and community energy
7	Hewitt [21] (2019)	207	Frontiers in Energy Research	Europe	Just transition/multi-level coordination
8	Pastukhova [1] (2020)	—	Geopolitics of the Global Energy Transition	Global	Regulatory coordination/strategic coherence
9	Hoppe [3] (2020)	—	Sustainability	Europe	Regional coordination layer (meso)
10	Chotimah [4] (2025)	—	E3S Web of Conferences	Indonesia/Global S.	Regional coordination layer (meso)

and Table S1)

Table 1 summarizes the ten most globally cited studies of the included corpus, providing for each entry the citation key, first author, year, source, geographic scope, methodological orientation, and the LDA-derived governance signal family to which the study contributes (cf. Section 3.4). The complete listing of the 339 included studies, with DOI, total citations, source, document type, country of corresponding author, and thematic-axis classification, is reported in Table S1 of the Supplementary Material (Supplementary File S2, Table_S1_Included_Studies.xlsx), generated from the harmonized Biblioshiny report (Supplementary File S3, BiblioshinyReport-Eq1.xlsx and sheet MostGlobCitDocs), the ten studies of Table 1 are individually cited at the point where their contribution informs the LDA topic interpretation, the policy discussion, or the scenario calibration; the remaining 329 studies are aggregated by signal family and cited collectively through the corresponding bracketed multi-citations of Section 3.4 and Section 5, with their individual entries listed in Table S1, following the citation approach recommended by the Editorial Office for corpus-level multi-citations.

3.4. LDA Modeling, Validation, and Topic-to-System 4 Mapping

Topic modeling was implemented in Python 3 (gensim 4.3). Text preprocessing included lowercasing, tokenization, removal of punctuation/numbers and English stop-words, phrase detection, and lemmatization to construct the document–term matrix. Candidate models spanning K = 5–40 were evaluated to balance semantic interpretability and generalization: coherence was assessed using the $C_v$ measure [22], and perplexity on a held-out set. The selected model (K = 30) corresponded to an inflection region where coherence reached a local maximum and perplexity gains became marginal (Figure 6).

To assess semantic separation and reduce the risk of topic leakage, inter-topic distances were examined using pyLDAvis [23] (Figure 7).

Topic-to-System 4 mapping and bias control. The 30 topics were consolidated into higher-order thematic domains representing governance-relevant signal families (participation and community energy, spatial justice and conflict, grid constraints, regulatory coordination, biodiversity/land-use trade-offs, and emerging technology/policy narratives). To prevent arbitrary labeling, the translation of statistical LDA topics into System 4 governance domains was structurally validated through deliberative expert panels (Project GENTE, Phase 4), aligning latent signals with operational policy dimensions. From a cybernetic standpoint, this human-in-the-loop interpretation is not a methodological flaw but a structural requirement: it constitutes the normative closure provided by System 5 [10]. Figure 8 shows the resulting translation of LDA outputs into System 4 strategic domains.

3.5. Temporal Dynamics and Requisite Variety in System 4

To maintain Ashby’s Law of Requisite Variety, the System 4 sensing mechanism cannot remain static. Temporal inspection of the LDA corpus (2014–2025) reveals fast discursive turnover: technical-descriptive topics dominated the early window, whereas transformative themes such as green hydrogen, gender perspectives in energy justice, and decarbonization pathways emerged rapidly and increased in salience between 2019 and 2025. Drawing on temporal LDA designs that track energy lifestyles seasonally [6] and procurement trends annually [7], we adopt a dual-update strategy for the regional intelligence function: (i) programmatic updates, in which the sensing corpus is re-estimated annually or bi-annually, synchronized with national and regional planning cycles (e.g., UPME’s reference expansion plans and Plan Energético Nacional), so that System 3 allocations are based on the latest socio-technical discourse; and (ii) event-driven (ad-hoc) updates, triggered by macro-level landscape shocks—regulatory overhauls, geopolitical energy crises, or extreme climate events—so that the architecture’s perceptual capacity remains anticipatory rather than merely historical. Operationally, this implies automating the NLP pipeline (preprocessing, K-selection, coherence check) as a repeatable component of System 4, ensuring continuous alignment between sensing capacity and the rate of environmental change.

4. Multi-Criteria Spatial Architecture

4.1. National Conflict Index Construction and Indicator Sourcing

The National Conflict Index (INC) is a spatially explicit constraint that internalizes socio-political instability into renewable planning. Conflict indicators (homicides of social leaders, massacres, forced displacement, mobility restrictions, and infrastructure blockades) were sourced from official institutional repositories: INDEPAZ, the Colombian Ombudsman’s Office (Defensoría del Pueblo), and OCHA. To avoid population-density bias, absolute conflict events for the period 2019–2024 were transformed into standardized rates per 100,000 inhabitants prior to spatial interpolation.

The index is computed on a raster grid where each 1 km² cell constitutes a decision alternative. The INC for cell i is computed as a weighted sum:

$$IN{C}_i={\displaystyle \sum _{j=1}^m}{w}_j{x}_{ij},$$

(1)

where $x_{ij}$ is the normalized value of conflict variable j at cell i, and $w_j$ is the criterion weight derived via CRITIC. The internal weights of the INC variables are therefore not subjectively assigned: they are extracted from the statistical structure of the data—contrast intensity (standard deviation) and inter-criteria conflict (Pearson correlation) of the standardized event rates [8]. However, the upstream selection of indicators, their treatment as a “cost” criterion in TOPSIS, and the post-hoc validation of their spatial impact underwent formal expert validation through interdisciplinary deliberative panels and participatory workshops conducted during Phase 4 of the foundational Project GENTE. These panels cross-validated the algorithmic CRITIC outputs against lived territorial realities and confirmed the reliability of the indicators to measure deployment risk for renewable energy projects [9]. From a cybernetic standpoint, this complementarity between algorithmic objectivity (CRITIC) and human-in-the-loop validation is not a methodological deficit but a structural requirement: it constitutes the normative closure provided by System 5 (Policy/Identity) [10].

4.2. CRITIC Objective Weighting and TOPSIS Spatial Prioritization

Following the call to expose computational equations in the main text, CRITIC weights are derived from the contrast intensity (standard deviation $s_j$) and the inter-criteria conflict (Pearson correlation $r_{jk}$) of the data:

$$H_j=s_j{\displaystyle \sum _{k=1}^m}(1-r_{jk}),W_j={\displaystyle \frac{H_j}{{\sum }_{k=1}^m{H}_k}}.$$

(2)

This penalizes structurally collinear indicators and assigns greater influence to variables that contribute non-redundant discriminative power [8,9]; the resulting inter-criteria correlation structure is shown in Figure 9. Min–Max normalization is applied prior to weighting:

$${\tilde{x}}_{ij}={\displaystyle \frac{x_{ij}-{min}_i{x}_{ij}}{{max}_i{x}_{ij}-{min}_i{x}_{ij}}}.$$

(3)

Spatial alternatives are then prioritized using TOPSIS. Letting $D_i^{+}$ and $D_i^{-}$ denote the Euclidean distances from cell i to the positive and negative ideal solutions in the weighted, normalized criteria space, the closeness coefficient is:

$$C{C}_i={\displaystyle \frac{D_i^{-}}{D_i^{+}+D_i^{-}}}.$$

(4)

Cells with higher $C{C}_i$ jointly maximize technical feasibility and minimize conflict exposure (Figure 10).

4.3. Technical Feasibility Modeling and Spatial Datasets

Technical feasibility was constructed at a unified 1 km² raster resolution from the data sources reported in

Table 2. Spatial datasets, resolution, and treatment in the multi-criteria pipeline.

Layer	Source	Native Res.	Treatment
Solar irradiation (Global Horizontal Irradiation, GHI)	Solargis	250 m	Resampled to 1 km; benefit
Wind resource (100 m)	Global Wind Atlas v4.0	250 m	Resampled to 1 km; benefit
Geothermal suitability	Servicio Geológico Colombiano (SGC)	National	Indexed to 1 km; benefit
Land use/cover	IGAC, SIAC	Variable	Boolean inclusion/exclusion
Protected areas	RUNAP, RAMSAR	National	Boolean exclusion
Conflict indicators	INDEPAZ, Defensoría, OCHA	Municipal	Standardized per 100,000 inhab.; cost

. Boolean exclusion layers were applied prior to MCDA to mirror real-world land-use trade-offs: National System of Protected Areas (RUNAP), RAMSAR wetlands, páramos, urban areas, and indigenous reserves. Continuous feasibility layers (solar irradiation, wind resource at 100 m hub height, geothermal suitability, slope, and grid proximity) were normalized as benefit/cost criteria for prioritization, in line with established Spatial Multi-Criteria Analysis practice [24].

4.4. Sensitivity, Robustness, and Scenario Calibration

Robustness was assessed by perturbing CRITIC-derived weights within $\pm 10\%$ and $\pm 20\%$ intervals and repeating the TOPSIS ranking under alternative normalization schemes. Stability was quantified using rank-order agreement (Spearman correlations) between baseline and perturbed rankings: across perturbations, agreement remained high (Spearman > 0.92), indicating that prioritization is not driven by fragile weighting configurations (Figure 11).

To make System 5 normative orientation explicit and reproducible, baseline CRITIC weights are recalibrated in three scenarios:

• Accelerated (Technocentric): 80% weight on technical/resource criteria, 20% on conflict (INC).
• Balanced: 50%/50% baseline distribution.
• Justice-Oriented: 30% weight on technical criteria, 70% on conflict, heavily penalizing deployment in territories with elevated INC.

5. Spatial Integration and Empirical Findings

5.1. Pixel-Level Viability Patterns

At 1 km² resolution, substantial heterogeneity emerges within single municipalities, exposing local contrasts that aggregated administrative statistics conceal. Spatial alternatives are classified into three operational categories using thresholds derived from national distributions:

• Opportunity Zones: Strong renewable resource (Global Horizontal Irradiation, GHI $>4.5$ kWh/m²/day or wind speed > 8 m/s at 100 m hub height) combined with conflict exposure below the second national quartile.
• Critical Zones: Strong resource potential combined with conflict intensity above the third quartile—technically feasible but with elevated coordination requirements and institutional risk.
• Neutral Zones: Moderate resource availability intersecting mid-range conflict indicators.

Figure 12 maps the spatial distribution of these three zone types.

5.2. Quantitative Evidence of the Localization Paradox

Overlay analysis reveals a pronounced spatial adjacency between high renewable resource corridors and zones of elevated governance complexity. Empirically, the bivariate analysis indicates that approximately 68% of territories in the Colombian Caribbean with optimal wind regimes (>9 m/s at 100 m hub height) fall within the top quartile of the INC (Figure 13). Consequently, the theoretical national renewable potential—comprising approximately 50 GW of offshore wind, 30 GW of onshore wind, 42 GW of solar PV, and 1.17 GW of geothermal capacity in volcanic systems, exceeding 100 GW in aggregate—narrows to roughly 24 GW of governance-viable capacity once spatial conflict and institutional exclusion layers are applied. The reduction is empirical rather than rhetorical: it is driven by the joint operation of the INC penalization and the Boolean exclusion of RUNAP, RAMSAR, páramo, and urban areas.

5.3. Regional Concentration of Critical Zones and the Windpeshi Case

Aggregating pixel-level classifications, the highest density of Critical Zones concentrates in Colombia’s Caribbean corridor. La Guajira exhibits onshore wind regimes among the strongest in Latin America (frequently exceeding 9 m/s at 100 m hub height) yet records conflict exposure within the highest national quartile. Parts of Cesar combine solar irradiation above 5.0 kWh/m²/day with socio-economic tensions associated with post-coal transition dynamics. These territories materialize the localization paradox: renewable infrastructure reaches peak technical efficiency precisely where governance complexity is greatest.

The indefinite suspension of the Windpeshi wind project (PROY-00836, jurisdictions of Uribia and Maicao) by its developer in 2023 illustrates this dynamic in practice. Despite Class 7 wind potential and an environmental license, implementation stalled due to persistent community blockades, unresolved benefit-sharing arrangements, and coordination breakdowns—an empirical demonstration that the absence of recursive meso-level coordination (System 2) converts high technical potential into a stranded asset.

5.4. Normative Scenario Divergence

Spatial prioritization diverges materially under the three calibrated scenarios. Under the Accelerated configuration, priority concentrates in raw resource abundance: offshore and onshore wind clusters in the Caribbean dominate the ranking. Under the Justice-Oriented configuration, territories with elevated conflict exposure are penalized more strongly; high-potential Caribbean zones decline in relative priority and investment shifts toward distributed solar corridors in central Colombia. The Balanced configuration produces diversified regional portfolios that reduce exposure to concentrated bottlenecks. Empirically, transitioning from the Accelerated to the Justice-Oriented scenario reorders the top-decile pixel set by more than one third, evidencing that System 5 normative orientation is not a rhetorical overlay but a quantifiable structural parameter [9]. Figure 14 contrasts the regional ranking outcomes across the three scenarios.

6. Cybernetic Governance Reconfiguration

6.1. Operational Recalibration of Systems 1–3

The spatial findings indicate that renewable transition bottlenecks are less a function of resource scarcity than of institutional absorption capacity. At the operational level, deployment alternatives (System 1) are no longer evaluated solely on technical performance: high-potential pixels coinciding with elevated INC are flagged for enhanced coordination capacity prior to deployment. Resource allocation (System 3) shifts from discretionary negotiation toward structured prioritization grounded in observable dispersion and inter-criteria structure. Coordination mechanisms (System 2) are recalibrated to activate proactively around Critical Zones rather than after conflict escalation. The result is a reduction in “inactive asset” accumulation—projects that achieve technical approval but stall during implementation due to unabsorbed socio-political complexity.

6.2. System 4 Feedback and Adaptive Adjustment

The topic-modeling layer provides a structured signal environment that complements spatial indicators by capturing shifts in governance discourse. When integrated into the decision cycle, these signals inform periodic recalibration of prioritization parameters: increasing salience of justice-oriented discourse can justify greater penalization of high-conflict zones, while stabilization in certain regions may relax constraint intensity over time. This expands institutional monitoring capacity and enables governance to process discursive and territorial signals before they crystallize into implementation paralysis [5,6,7].

6.3. System 5 Normative Calibration

Normative orientation influences how conflict exposure and technical potential are balanced. Scenario analysis demonstrates that the penalization intensity associated with conflict indicators produces materially different regional portfolios. A capacity-maximizing orientation privileges high-resource corridors even when governance risk is elevated; a justice-weighted orientation reallocates priority toward regions with lower conflict exposure; an intermediate configuration diversifies investment geographically. Identity commitments are not abstract declarations: they shape the geometry of territorial allocation as a structural parameter rather than an external policy overlay. Figure 15 synthesizes the interaction among operational deployment, coordination, allocation logic, anticipatory intelligence, and normative calibration.

7. Discussion

7.1. From Qualitative Cybernetics to Evidence-Based Governance

This study advances a shift from qualitative cybernetic interpretation toward evidence-structured cybernetic governance. The contribution does not alter the ontology of the VSM; it reinforces its epistemic capacity by embedding analytical instrumentation within its recursive structure, directly responding to the research agenda of [5]. The coupling of environmental sensing and structured prioritization transforms cybernetic design from metaphor into executable governance logic. Sustainability is reframed as an internal structural parameter rather than an external compliance layer.

7.2. Critical Engagement with the Empirical Results

The numerical evidence requires interpretation rather than restatement. The reduction from >100 GW theoretical potential to roughly 24 GW governance-viable capacity is not a verdict on Colombia’s resource base; it is a measurement of the institutional absorption deficit that the current MLG configuration imposes on technical opportunity. Equally, the 68% overlap between Class 7 wind zones and the top INC quartile is not an indictment of La Guajira but a structural diagnosis: the territories most attractive on engineering grounds are those whose viability depends most strongly on functional System 2 mediation. The Windpeshi suspension confirms that, absent such mediation, even licensed projects revert to stranded-asset status. Read jointly, these findings argue against capacity-maximization as a sufficient policy logic and in favor of conflict-sensitive prioritization aligned with regional coordination capacity.

7.3. Positioning Within Transition Governance Literature

Compared with established frameworks, the proposed architecture emphasizes structural prescriptiveness. The Multi-Level Perspective offers a powerful descriptive account of niche–regime–landscape interactions but does not specify the institutional design conditions for systemic viability. Adaptive governance models stress flexibility and polycentric coordination but rarely define how institutional complexity must match environmental complexity. The evidence-based VSM architecture differs in its insistence on multi-level coordination instrumented by computational intelligence and objective MCDA [3,4,25].

7.4. Broader Applicability and Scalability in the Global South

A central implication of this hybrid framework is its broad applicability across the Global South, where renewable transitions repeatedly encounter similar intersections of high abiotic potential, spatial friction, and institutional asymmetry. Multi-level governance literature highlights an empirical gap concerning how transitions unfold in developing contexts characterized by fragmented mandates, limited fiscal capacity, and fluid or informal institutional structures [4,5]. The VSM is a functional rather than organic-administrative architecture: it maps cybernetic capacities rather than rigid organizational charts, which makes it portable to fundamentally different institutional configurations.

Where formal regional authorities are weak or contested, the cybernetic functions of the model can be redistributed across hybrid governance arrangements. For instance, System 3 (control and resource allocation) need not reside within a central ministry: it can be operationalized by mixed public–private entities with a public mandate—analogous to Chile’s Agencia de Sostenibilidad Energética—which execute policies with operational agility while preserving accountability. Likewise, Systems 1 and 2 (operations and coordination) can be fulfilled by community energy trusts, indigenous councils, or cooperative micro-grid enterprises, as illustrated by rural electrification arrangements documented in Nepal and South Africa and by Colombian community self-governance under resource asymmetry [11,13]. By decoupling required governance functions from rigid administrative hierarchies, the framework offers developing nations a replicable and adaptable tool to diagnose structural voids and design conflict-sensitive governance architectures, regardless of their starting institutional maturity [3].

7.5. Limitations and Future Research

Several limitations delimit the scope of the proposed framework. First, prioritization is derived from a cross-sectional configuration, whereas renewable transitions unfold under evolving conditions; future research should explore adaptive recalibration mechanisms updating allocation parameters in near real time. Second, environmental sensing remains constrained by corpus composition; expanding sources to multilingual, non-indexed, and community-generated materials would enhance representativeness and reduce epistemic bias. Third, while expert deliberative panels validated indicator selection and topic-to-System 4 mapping, full territorial validation requires participatory implementation studies—moving from formal consultation to binding intercultural dialogue and shared-ownership models in the spirit of energy-justice scholarship [5,13]. Validation in this manuscript is therefore structural and computational: LDA coherence, $\pm 10$–$20\%$ TOPSIS sensitivity, and expert deliberative panels; longitudinal action-research validation remains an explicit next step.

8. Policy Implications for Colombian Energy Institutions

8.1. Anchoring the Framework in Real Institutions

The broad principles of multi-level governance must be anchored in concrete institutional realities [11]. In the Colombian context, the framework implies a redesign of current planning mechanisms. UPME currently operates as a centralized System 3/4 hybrid focused on technical generation–transmission expansion; ANLA acts as System 3 regulatory control; and CREG sets market and tariff parameters. The framework recommends empowering the Departamentos (regional governments) to fulfill the System 2 coordination role, integrating CRITIC–TOPSIS spatial dashboards and LDA-derived discourse signals to mediate between national targets and municipal realities.

8.2. Operational Anchors Already Available

Several recent regulatory innovations operate as System 3 attenuation mechanisms within ANLA. The specialized licensing tracks for solar (LASolar, Decree 1033 of 2025) and wind (LAEólica, Decree 1186 of 2025) rationalize impact assessment for non-conventional renewables and reduce coordination friction. ANLA’s public dashboards for non-conventional renewable energy sources (FNCER) operate as accountability sensors aligned with System 3/4 functions. UPME’s reference expansion plans should integrate the INC alongside existing technical layers to synchronize generation readiness with delayed transmission infrastructure (e.g., the Colectora line in La Guajira), preventing the accumulation of inactive assets analogous to the Windpeshi suspension.

8.3. Recursive Multi-Level Coordination

A recursive architecture implies that intermediate governance levels possess sufficient analytical and coordination capacity to harmonize infrastructure planning, licensing, and community engagement across municipalities. The two-decade delegation of mining and energy functions to the Secretaría de Minas of Antioquia (2001–2023) is an instructive Colombian precedent: empowering a meso-level entity with titulation, fomento, and oversight mandates demonstrably harmonized national policy with local realities and reduced coordination breakdowns. Interoperable data systems and shared spatial dashboards can support this alignment without recentralizing decision authority. Figure 16 summarizes this policy reconfiguration across governance levels.

Overall, policy adaptation in renewable transitions benefits from embedding differentiated spatial prioritization, anticipatory sensing, and multi-level coordination within a coherent architecture; the objective is structurally aligned deployment capable of translating technical potential into territorially viable outcomes.

9. Conclusions

This study transitions the Viable System Model (VSM) from a conceptual metaphor into a reproducible, evidence-based governance architecture for regional renewable energy transitions. Concretely, (i) an LDA model identified K = 30 strategic topics from a PRISMA-curated corpus of 339 articles, operationalizing System 4 anticipatory intelligence; (ii) a 1 km² CRITIC–TOPSIS spatial pipeline quantitatively confirmed the localization paradox, with approximately 68% of optimal wind zones in the Colombian Caribbean overlapping the highest INC quartile; and (iii) the joint application of conflict and exclusion filters narrowed a theoretical capacity exceeding 100 GW (50 GW offshore, 30 GW onshore, 42 GW solar PV, 1.17 GW geothermal) to roughly 24 GW of governance-viable capacity. Sensitivity testing ($\pm 10$–$20\%$ weight perturbation, Spearman > 0.92) confirmed ranking robustness, while the three normative scenarios (Accelerated, Balanced, Justice-Oriented) showed that System 5 calibration materially reshapes territorial portfolios.

These findings argue that transition bottlenecks in high-variety contexts are driven by governance absorption capacity rather than by technical resource scarcity. Embedding the INC inside formal planning instruments—UPME’s reference expansion plans, ANLA’s licensing tracks, and the coordination mandate of the Departamentos—is therefore not a social safeguard appended to technical optimization, but a precondition for technical viability. The architecture is intentionally modular: its components can be adapted to diverse Global South governance configurations, scaling from municipal networks to transnational corridors. Renewable transitions require not only abundant resources, but governance systems recursively designed to absorb the complexity they generate.