From Crypto Trust to Digital Governance: Mapping Coordination Frictions in Global Policy Discourse

From Crypto Trust to Digital Governance: Mapping Coordination Frictions in Global Policy Discourse | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article From Crypto Trust to Digital Governance: Mapping Coordination Frictions in Global Policy Discourse hana kim This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8702396/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Blockchain is no longer only a crypto-finance experiment; it is a contested digital infrastructure shaping financial regulation, data governance, and digital sovereignty. Yet policy debates lack systematic evidence on how blockchain is framed as a governable object across regions and institutions—and where such framings create coordination frictions for cross-border regulation and standard-setting. This study maps global blockchain policy discourse using a dual-corpus, two-stage design that links broad metadata coverage with full-text inference. We classify documents into governance archetypes and quantify cross-group alignment using a low-dimensional semantic space derived from text, alongside a policy-risk index capturing evaluative stance. Three findings emerge. First, the discourse shifts from monetary framings toward regulatory consolidation and then toward coordination/standard-setting and public-infrastructure integration. Second, meanings cluster by region and issuer type despite shared anchors, indicating persistent interpretive divergence. Third, evaluative stance varies systematically across groups, revealing heterogeneous risk sensitivity. By separating governance logic from evaluative stance, the paper provides a scalable measurement framework and a diagnostic toolkit (e.g., coordination heatmaps and translation tables) to identify where regulatory alignment is most likely to break down. Humanities/Complex networks Social science/Complex networks Physical sciences/Mathematics and computing Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 1. Introduction Blockchain has moved from a niche protocol debate to a recurring object of governmental problem definition - a site where institutions negotiate compliance, identity, and cross-border coordination in data-driven economies. Yet existing scholarship typically observes this transformation through case studies, national regulatory histories, or sector-specific episodes, which makes it difficult to compare how policy meaning travels, fragments, or converges across jurisdictions over long horizons. This paper treats policy discourse itself as a governance signal that can be measured at scale. Using Overton, I assemble a large global corpus of blockchain-related policy documents and map how the field evolves in (i) institutional participation, (ii) topical scaffolding, and (iii) semantic and evaluative framing. The key challenge is that broad coverage and reliable full-text inference rarely coincide: metadata repositories provide breadth, while computational text inference depends on extractable, linguistically stable PDFs. To address this coverage-inference trade-off, I use a dual-corpus, two-stage design. Stage 1 uses Overton metadata (N = 17,019; 1994–2022) to map the long-run field and motivate historically grounded periods. Stage 2 constructs a full-text inference corpus (N = 393; 2014–2022) that supports three complementary modules: transparent governance-archetype measurement, document-embedding maps of semantic proximity, and evaluative risk framing via a transparent lexicon (with a transformer diagnostic reported as a robustness check). Importantly, selection into extractable full text is diagnosed and, where possible, stress-tested rather than assumed away. The analysis is organized around three research questions that convert field mapping into decision-relevant comparisons. RQ1 asks: how does the functional center of blockchain governance discourse shift over time - from regulatory control toward coordination/standards and identity/infrastructure - and when do inflection points occur? RQ2 asks: where do cross-jurisdiction coordination frictions concentrate, as measured by semantic distance and archetype divergence between region pairs? RQ3 asks: is evaluative tone (risk sensitivity) primarily a region effect, or is it better explained by governance archetypes and institutional source composition? The paper contributes to a interdisciplinary scholarship on social implications of emerging technologies in three ways. Substantively, it offers a longitudinal, comparative map of how an emerging technology is repurposed into digital-governance infrastructure through policy discourse. Methodologically, it provides a reproducible dual-corpus template that separates field coverage from text inference while making extractability an explicit measurement constraint. Practically, it translates empirical outputs into coordination tools: a friction heatmap that identifies where semantic misalignment is highest, a layered playbook that sequences minimum viable alignment before deeper harmonization, and a translation table that documents how similar governance functions are named differently across regions and institutions. The remainder proceeds as follows. Section 2 situates the study in scholarship on technology governance, policy discourse, and computational mapping. Section 3 describes the dual-corpus design and inference modules. Section 4 reports results as answers to RQ1-RQ3, with uncertainty and robustness diagnostics. Section 5 discusses implications for cross-border coordination and the study's limitations. 2. Literature Review 2.1 Blockchain as an Institutional and Governance Technology Blockchain emerged from its origins as a cryptographic protocol for peer-to-peer currency exchange (Nakamoto, 2008 ) to a broader institutional technology that reconfigures how coordination, verification, and rule enforcement operate in digital environments. Rather than functioning solely as a decentralized transaction mechanism, blockchain introduces architectures that reassign authority and compliance responsibilities from intermediaries to protocol-based consensus systems (Catalini & Gans, 2020 ; Allen et al., 2019 ; Rauchs et al., 2018 ). This shift has positioned blockchain as a governance-relevant infrastructure whose implications extend well beyond cryptocurrency markets, including adoption logics and organizational value drivers (Angelis & da Silva, 2019 ) as well as broader fintech ecosystem transformations (Lee & Shin, 2018 ; Iman, 2020 ). Studies in financial economics demonstrate that programmable logic and decentralized validation reshape transaction costs, corporate governance mechanisms, and contractual enforcement processes (Yermack, 2017 ; Cong & He, 2019 ). Related work in finance and computing highlights blockchain’s expanding role in financial infrastructures and market operations (Treleaven et al., 2017 ), while governance-oriented legal and management research discusses how blockchains may function as corporate governance tools and reshape capital-market paradigms (Gikay, 2018 ; Weber, 2018 ). Complementing these perspectives, patent-based evidence charts the technological development trajectory of blockchain and its diffusion across domains (Xu et al., 2019 ). In the public sector, governments have experimented with blockchain applications across identity systems, procurement, land registries, and record-management infrastructures, often motivated by transparency and auditability goals (Ølnes et al., 2017 ; Hileman & Rauchs, 2017 ; Hughes et al., 2019 ; Berryhill et al., 2018 ). Systematic reviews and conceptual work further consolidate this domain by mapping how blockchain is operationalized in government and what governance challenges recur across cases (Garrod & Samuel, 2021 ; El-Haddadeh et al., 2021; Tan et al., 2022 ), including more recent efforts to translate these insights into practical governance principles and a comprehensive framework for blockchain systems (Liu et al., 2022 ). These developments underscore blockchain’s emerging institutional importance as states explore alternative models of trust, information sharing, and accountability. Governance research highlights that blockchain redistributes—rather than eliminates—trust by shifting reliance from institutional intermediaries to protocol designers, validators, and platform operators (Hawlitschek et al., 2018 ; Tan & Low, 2019 ). Legal scholarship examines the tensions between immutable code and adaptive regulatory regimes, pointing to unresolved questions around liability, compliance, and the interaction between on-chain and off-chain governance (De Filippi & Wright, 2018 ; Werbach, 2018 ; Yeung, 2019 ; Wright & De Filippi, 2015). Political-economy perspectives similarly highlight the rise of “shadow governance,” where private consortia and standards bodies set de facto rules governing blockchain infrastructures (Campbell-Verduyn, 2017). In parallel, fintech and regtech scholarship emphasizes how regulators and financial institutions respond to these shifts through new supervisory technologies and regulatory design, including sandbox experimentation and “smart regulation” approaches (Anagnostopoulos, 2018 ; Zetzsche et al., 2017 ). These streams portray blockchain as a contested socio-technical institution characterized by distributed authority and heterogeneous governance imaginaries. Yet scholarship remains fragmented across sectors and regions, resulting in limited understanding of how governments collectively frame blockchain’s institutional significance or how this framing evolves over time. 2.2 Policy Documents as Arenas of Technology Framing and Public-Sector Interpretation Policy documents—national strategies, white papers, regulatory consultations, guidelines, and expert reports—serve as arenas where governments interpret technological change and articulate expectations for its governance. Discursive institutionalism conceptualizes such documents as sites where ideas, norms, and governance logics are circulated, stabilized, and contested (Schmidt, 2008 ; Shore & Wright, 2011 ). These texts reveal how institutions define opportunities and risks, legitimizing certain technological pathways while delimiting others. They also intersect with longer-standing discussions on how public-sector research is evaluated for societal relevance and how policy-facing knowledge infrastructures are organized (Van der Meulen & Rip, 2000 ), as well as debates over the use of metrics in research assessment and management (Wilsdon et al., 2015 ). The growth of large-scale infrastructures like the Overton database has enabled systematic mapping of science–policy linkages and cross-national policy agenda patterns (Szomszor & Adie, 2022 ; Bornmann et al., 2022 ). Such corpora provide visibility into which institutions contribute to policy discourse, the scientific evidence they cite, and how their priorities shift over time. Recent bibliometric work illustrates how these approaches can be operationalized to uncover science–policy interfaces in concrete policy domains (Schlierkamp et al., 2025 ). Computational text-analytic methods have deepened the capacity to analyze these documents. Topic models and clustering techniques have revealed thematic structures in climate adaptation, environmental governance, public health, and digital government policy (Lesnikowski et al., 2019 ; Ma et al., 2024 ; Araos et al., 2016 ; Biesbroek et al., 2015 ). Embedding-based semantic models illuminate conceptual proximity, discursive alignment, and narrative shifts across political and technological domains (Kozlowski et al., 2019 ; Li et al., 2023 ). Sentiment and evaluative-language models further capture how institutions express support, caution, or concern when articulating governance positions around emerging technologies (Ceron & Negri, 2016 , Wang et al., 2023), and adjacent digital-government research extends these tools to map public sentiment and policy debate dynamics across platforms and media environments (Panagiotopoulos et al., 2017 ; Jia et al., 2022 ; Nitzbon et al., 2023 ; Sucupira & de Albuquerque, 2022 ). Recent scholarship in Big Data & Society shows how large-scale analysis of policy and social-media discourse can produce decision-relevant governance insights—for example, by quantifying justification frames around surveillance infrastructures (Steinhardt & Göbel, 2026 ) or by unpacking how users negotiate risk–benefit trade-offs around generative AI (Tang et al., 2025 ). This paper extends this orientation to blockchain policy documents, pairing field mapping with explicit comparative tests that are directly interpretable for cross-jurisdiction coordination. Despite these methodological advances, blockchain remains underexamined as a policy-discourse domain. Existing studies focus largely on parliamentary debates, regulatory announcements, or specific national strategies (De Filippi & Wright, 2018 ; Yeung, 2019 ). While recent scholarship has increasingly conceptualized blockchain as a novel mode of organizational collaboration (Lumineau et al., 2021 ) and analyzed the distinct dynamics of its internal and external governance (Hsieh & Vergne, 2023), these conceptual advancements have yet to be fully integrated into a comprehensive policy-discourse framework. 2.3 Analytical Perspectives for Mapping Global Technology Policy Discourse Synthesizing insights from blockchain governance research and policy-discourse analytics reveals several major empirical gaps. First, although blockchain’s institutional relevance has expanded rapidly, comprehensive evidence remains limited regarding which actors—national governments, international organizations, regulatory agencies, NGOs, and expert institutions—have contributed to the production of blockchain policy documents over time. Existing analyses focus primarily on isolated regulatory episodes or single-country contexts (De Filippi & Wright, 2018 ; Yeung, 2019 ; Finck, 2019), limiting visibility into the global architecture of policy discourse. This fragmentation persists even as studies on decentralized network governance propose models of regulated self-regulation (Zwitter & Hazenberg, 2020 ) and public-sector research begins to consolidate actor–governance frameworks and implementation guidance (Garrod & Samuel, 2021 ; Tan et al., 2022 ). Second, the evolution of policy themes has not been systematically documented. Blockchain’s policy salience has shifted from cryptographic innovation to financial regulation, public-sector modernization, data governance, and systemic-risk management (Hileman & Rauchs, 2017 ; Campbell-Verduyn, 2017; Hughes et al., 2019 ). Closely related fintech and regtech trajectories—covering ecosystem change, supervisory technology, and new regulatory design—suggest additional pathways through which blockchain becomes policy-salient, but these have rarely been integrated into longitudinal discourse mapping (Lee & Shin, 2018 ; Anagnostopoulos, 2018 ; Iman, 2020 ; Zetzsche et al., 2017 ). Moreover, although technology-development studies document blockchain’s expanding innovation base (Xu et al., 2019 ) and organizational adoption rationales (Angelis & da Silva, 2019 ), the timing, magnitude, and stability of discourse transitions have not been mapped across a multi-decade global corpus. Third, semantic and evaluative framing remains underexplored. While prior work identifies divergent narratives surrounding blockchain—innovation enabler, transparency mechanism, surveillance risk, systemic-risk factor—empirical evidence on framing variation across institutional sectors and world regions is extremely limited (Ceron & Negri, 2016 ; Li et al., 2023 ; Wang et al., 2023). Embedding-based methods now make this analysis feasible, but they have not been applied to blockchain policy at scale, despite accumulating foundations in public-sector blockchain governance and systematic syntheses of blockchain’s role in public services (El-Haddadeh et al., 2021; Berryhill et al., 2018 ). Prior literature offers conceptual clarity about blockchain’s institutional and governance characteristics and provides robust methodological tools for analyzing large policy corpora, but it lacks systematic, multi-actor, longitudinal evidence on how blockchain is framed, interpreted, and prioritized across the global policy landscape. This study addresses these gaps by constructing a large-scale corpus of blockchain policy documents and examining its structural, thematic, and semantic dimensions over nearly three decades. 3. Data and methods 3.1. Two-stage analytical design Policy documents record institutional choices and the policy logics that justify them, making them primary materials for measuring policy meaning at scale. Text-as-data approaches therefore offer a principled route to recover latent positions, frames, and implied governance priorities from large collections of political and policy texts, while requiring explicit attention to assumptions and validation (Grimmer & Stewart, 2013 ; Grimmer et al., 2022 ; Ash & Hansen, 2023 ). This study uses a dual-corpus, two-stage analytical design that separates broad field mapping from full-text semantic and evaluative inference. Stage 1 constructs a wide-coverage metadata corpus from Overton to characterize the emergence of the blockchain policy field, its institutional composition, and the distribution of Overton topical descriptors across historically grounded periods and regions. Stage 2 then draws a smaller full-text analytic corpus consisting of policy documents whose PDF text can be reliably extracted and cleaned for modeling. Using this inference corpus, the analysis (i) represents policy meaning with document embeddings (Doc2Vec, building on distributed representations; Mikolov et al., 2013 ; Le & Mikolov, 2014 ), (ii) estimates evaluative risk framing using a transparent lexicon (FinBERT is used only as a diagnostic robustness check; Devlin et al., 2019 ; Araci, 2019 ), and (iii) assesses convergence and divergence across regions and institutional source types via centroid-based cosine similarity. This separation aligns coverage and interpretability by preserving breadth for field mapping while prioritizing text quality for semantic and stance inference. Stage 2 covers 2014–2022 (N = 393) and supports embedding-based semantic mapping and lexicon-based evaluative risk framing; a transformer sentiment model (FinBERT) is computed only as a diagnostic robustness check (Devlin et al., 2019 ; Araci, 2019 ). To make the coverage–inference tradeoff transparent, I report the sample construction in Table 1 and summarize the end-to-end analytical pipeline in Fig. 1 . Table 1 Sample construction Metric Value Stage 1 metadata corpus (N) 17019 Stage 1 years 1994–2022 Stage 2 full-text corpus (N) 393 Stage 2 full-text years 2014–2022 3.2. Stage1. Field mapping and scaffolding 3.2.1 Data sources and metadata corpus construction The dataset is built from Overton, an international database of policy and government documents. A dual-corpus structure is used to align field coverage with text quality. The metadata corpus includes all documents tagged with “blockchain” or assigned blockchain-related topical descriptors between 1994 and 2022 September 9th, retaining publication year, source country, institutional source type, and Overton topic labels. This corpus supports long-run field mapping and participation patterns in the policy space. 3.2.2 Temporal harmonization and historically grounded periodization Publication years are validated and harmonized into numeric form. Because blockchain policy evolved through punctuated shifts associated with widely recognized technological and regulatory transitions, time is modeled as historically staged rather than purely continuous. Documents are assigned to four periods: Early Cryptography and Trust (1994–2010), Financialization and ICO Emergence (2011–2017), Regulatory Consolidation (2018–2020), and Digital Government Integration (2021–2022). Periodization provides the temporal scaffold needed to interpret topic shifts, semantic movement, and changes in evaluative language as transformations in policy meaning rather than as noise in annual production (Schmidt, 2008 ). The period boundaries are additionally motivated by governance scholarship emphasizing transitions in the architecture of trust, the rise of code-mediated rule systems, and the intensification of regulatory contestation around blockchain (Werbach, 2018 ; De Filippi & Wright, 2018 ; Yeung, 2019 ). 3.2.3 Regional and institutional classification To compare governance logics across contexts, documents are grouped by macro-region and institutional source type. Countries are aggregated into US/CAN, EU/UK, Asia (China, South Korea, Japan, Singapore, India), IGOs, and Global South/Other. Institutional types follow Overton’s source-type labels (e.g., government, IGO, think tanks). This classification operationalizes the study’s core comparative claim: that policy meaning is produced within institutional settings and is therefore expected to differ across regulatory traditions and governance models rather than converge mechanically across jurisdictions (Ølnes et al., 2017 ; Campbell-Verduyn, 2017). It also supports comparative assessment of how trust and regulatory philosophies are articulated across contexts in blockchain governance (Werbach, 2018 ; Yeung, 2019 ). 3.2.4 Archetype operationalization: keyword rules and anchor concepts To aid interpretation, Table 2 reports the operational definitions and representative keyword anchors for each governance archetype. Additional implementation details and extended outputs are provided in the Supplementary Information. Table 2 Archetype operationalization Archetype Governance logic Representative keywords (illustrative) Anchor terms (seed concepts) Regulatory Control Governance through legal and compliance mechanisms compliance; regulation; licensing; enforcement; AML/KYC; sanctions compliance; security; securities; AML/KYC Digital Identity & Infrastructure Governance through public digital infrastructure and identity systems digital identity; authentication; credentials; registry; infrastructure; public services identity; trust; registry; public services Coordination & Standards Governance through coordination, interoperability, and international standards interoperability; standards; cross-border; harmonization; international cooperation standards; interoperability; cross-border; ISO Unclassified No dominant governance logic detected — — Notes: This table reports archetype definitions with representative keywords and anchor concepts for transparency. The full regex rule set and complete anchor lists used for classification are available upon request. Documents with no keyword match and low semantic proximity to all anchor sets are labeled Unclassified. 3.3 Stage2. Semantic and evaluative inference Accordingly, Stage 2 prioritizes extractability and linguistic stability, and all semantic, stance, and alignment estimates are computed exclusively on this inference corpus, while Stage 1 indicators are computed on the full metadata corpus. 3.3.1 Inference corpus and preprocessing (semantic structure → evaluative stance → alignment) A full-text corpus is derived from the metadata corpus by retaining only those documents whose PDF text can be extracted without structural corruption. Extracted text is cleaned to remove encoding artefacts and linked back to the metadata; documents with unusable or extremely short text are excluded. Table 3 summarizes the Stage 2 inclusion and exclusion rules that operationalize this design choice, yielding a corpus (393 documents) that is tailored for inference tasks requiring stable linguistic inputs rather than broad archival coverage. This construction reflects the trade-off commonly encountered in large-scale policy repositories (Szomszor & Adie, 2022 ). The approach also aligns with best practice in policy-text analytics, where the credibility of semantic and stance estimates depends on prioritizing extractability and interpretability over maximal inclusion (Lesnikowski et al., 2019 ; Ma et al., 2024 ). Preprocessing is intentionally conservative. Full texts are standardized through lowercasing, removal of disruptive formatting artefacts introduced by PDF extraction, whitespace normalization, and minimum-length filtering. As indicated in Table 3 , these quality checks are applied uniformly to reduce artefactual meaning and sentiment estimates driven by OCR noise, boilerplate, or layout corruption. This yields inputs that are comparable across heterogeneous document templates and minimizes the risk that model outputs are driven by layout noise rather than policy framing and evaluative language (Kozlowski et al., 2019 ). Emphasizing format heterogeneity and analytic cleanliness is consistent with established guidance from adaptation and policy text-analytics research, which repeatedly notes that extraction artefacts can otherwise dominate measured thematic and evaluative signals (Araos et al., 2016 ; Lesnikowski et al., 2019 ; Ma et al., 2024 ). Table 3 Stage 2 inclusion and exclusion rules for the inference corpus (2014–2022; N = 393). Rule type Criterion Rationale / design purpose Notes (where applied) Inclusion In-scope blockchain-related policy document in the Stage 1 Overton metadata corpus Fixes topical scope and preserves field-wide coverage for representativeness diagnostics Stage 1 → Stage 2 linkage Inclusion Full-text PDF retrievable and machine-extractable (no structural corruption) Ensures stable linguistic inputs for embeddings and stance inference Stage 2 construction Inclusion Text passes conservative preprocessing quality checks (e.g., minimum-length and noise filters) Reduces artefactual sentiment/meaning estimates driven by OCR noise or boilerplate Stage 2 preprocessing Exclusion Image-only / scanned PDFs or extraction failures Avoids unreliable semantic and sentiment inference from corrupted inputs Dropped at extraction step Exclusion Duplicates or near-duplicates (mirrored reports, re-hosted PDFs, minor versioning) Prevents overweighting a single document family in time-series and clustering summaries De-duplication prior to inference Exclusion Extremely short or non-substantive texts after extraction (e.g., stubs, announcements) Insufficient semantic signal for archetype assignment and model-based stance estimation Dropped by minimum-length rule Flag (robustness) DAO/DeFi-labeled documents Potentially distinct protocol-governance discourse; retained in baseline but excluded in a robustness replication See Section 4.5 and Appendix A3 3.3.2 Inference Module I: Semantic structure via document embeddings The first inference module recovers the latent semantic structure of policy discourse using document embeddings. I train a Doc2Vec (paragraph vector) model on the cleaned full-text corpus to represent each document as a fixed-length vector; Doc2Vec extends the distributed-representation logic of Word2Vec to sentences and documents (Mikolov et al., 2013 ; Le & Mikolov, 2014 ). These vectors encode differences in problem framing, governance priorities, and regulatory orientation that are often expressed through contextual phrasing rather than explicit keywords. Embedding-based representations are increasingly used to map meaning structures and discursive proximity among actors (Kozlowski et al., 2019 ), but I interpret them cautiously given conceptual and measurement limits when embeddings are used as proxies for meaning and culture (Mandell, 2022 ). For visualization, I project document vectors to two dimensions with t‑SNE (van der Maaten & Hinton, 2008 ) to produce embedding maps, and I compute group-level semantic centroids to quantify convergence and divergence across regions and institutional types. 3.3.3 Inference Module II: Evaluative framing via a transparent risk lexicon (with transformer diagnostic) To characterize evaluative framing in blockchain policy discourse, I operationalize a transparent risk-framing index based on a domain-specific lexicon of risk and compliance terms (see Supplementary Note S4). The index counts lexicon hits per 1,000 tokens in the cleaned full text, supporting direct interpretability and sensitivity checks to alternative dictionaries. As a diagnostic, I also apply FinBERT, a transformer model pretrained on financial text, to the same documents (Araci, 2019 ). In this corpus, however, the resulting FinBERT-based signed sentiment series is constant (0), consistent with domain mismatch between policy-legal governance language and financial sentiment benchmarks. Accordingly, the main analyses use the lexicon-based risk-framing index, while FinBERT is retained only as a diagnostic robustness check. This approach follows established text-based risk and uncertainty measurement traditions—from policy-uncertainty indices (Baker et al., 2016 ) to domain-specific risk dictionaries (Loughran & McDonald, 2011 )—and aligns with embedding-enabled political risk measurement that captures contextual risk beyond raw counts (Hassan et al., 2019 ). For external benchmarking, I compare the resulting series to a widely used news-based geopolitical risk index (Caldara & Iacoviello, 2022 ). For transparency, I convert FinBERT labels (Devlin et al., 2019 ; Araci, 2019 ) into a signed score s_d by mapping the positive and negative classes to their posterior probabilities with opposite signs and assigning 0 to the neutral class. Because the resulting series is constant (0) in this dataset, it is not used as an outcome variable in the main text; it is reported only as a diagnostic in Supplementary Note S4. 3.3.4 Inference Module III: Institutional and regional semantic proximity (convergence/divergence) To quantify convergence and divergence in policy meaning, group-level semantic centroids are computed by averaging document embeddings within each region and each institutional type. Pairwise cosine similarity between group centroids yields similarity matrices that summarize whether the discourse field exhibits shared semantic “centers of gravity” or persistent separation across governance contexts. This aligns with prior work on institutional differentiation in technology governance, where similar technologies can be stabilized under distinct policy imaginaries and regulatory philosophies (Werbach, 2018 ; De Filippi & Wright, 2018 ; Yeung, 2019 ). In interpretive terms, proximity patterns are read as evidence about whether the architecture of trust and the code–law interface are framed similarly across actors, or whether they remain regionally and institutionally segmented (Werbach, 2018 ; De Filippi & Wright, 2018 ; Yeung, 2019 ). 3.4 Aggregation and reporting strategy Results are reported using aggregation choices that preserve interpretability while aligning with the dual-corpus design. Topic prevalence is summarized using relative frequencies derived from the reconstructed document–topic matrix; sentiment is summarized using arithmetic means of the continuous index; semantic alignment is summarized with cosine similarity among group-level centroids. Where document volumes differ substantially across groups, reporting emphasizes structural patterns and robustness-consistent rank ordering rather than raw magnitude comparisons, consistent with best practice in large-scale policy-document analyses (Szomszor & Adie, 2022 ). This reporting strategy also mirrors cross-domain policy analytics where thematic mapping and evaluative inference are interpreted primarily as comparative signals rather than exact measures of “true” sentiment or meaning (Lesnikowski et al., 2019 ; Sucupira & de Albuquerque, 2022 ; Mandell, 2022 ). In addition to reporting raw aggregates, I attach uncertainty or contrast-based evidence to the two comparisons that underpin RQ2 and RQ3. For semantic proximity, centroid similarities are complemented with bootstrap confidence intervals and selected difference tests; for evaluative tone, group contrasts are supported by regression models that condition on archetype and issuer composition. Finally, to translate mappings into a coordination metric, I construct a Coordination Friction Index for each region pair and time window that combines (i) semantic distance (1 - cosine similarity), (ii) divergence in archetype composition, and (iii) differences in evaluative tone. This index is non-causal but decision-oriented: it is designed to identify where mutual recognition and standard-setting are likely to face the highest semantic and institutional alignment costs. 3.5 Robustness and validation Robustness checks evaluate whether the main conclusions are sensitive to reasonable perturbations in corpus construction and modeling choices. First, extraction-quality and minimum-length thresholds are varied to ensure that results are not driven by a small number of unusually short or noisy documents. Second, embedding specifications are perturbed within defensible ranges and similarity matrices are compared for rank-order stability, while interpretation remains attentive to known limits in embedding-based meaning inference (Mandell, 2022 ; Kozlowski et al., 2019 ). Third, sentiment results are re-estimated under alternative truncation rules and score constructions to confirm that regional differences are not artefacts of document length or input constraints (Araci, 2019 ; Sucupira & de Albuquerque, 2022 ). Finally, period boundary sensitivity is assessed by shifting cut points around major transitions to verify that observed turning points and regional contrasts persist under plausible alternative temporal partitions, consistent with historically grounded periodization in governance and regulatory-change studies (Schmidt, 2008 ; Werbach, 2018 ; Yeung, 2019 ). 3.5.1 Validating archetype assignment (keyword rules). Archetype labels are validated by (i) conducting face-validity audits on stratified random samples across regions and years to confirm that rule hits map to the intended anchor concepts, (ii) perturbing keyword lists and minimum-hit thresholds to verify that dominant-archetype shares and turning points are stable, and (iii) checking that substantive conclusions persist under the representativeness reweighting and the DAO/DeFi exclusion exercise reported later. These checks treat the archetypes as transparent measurement rules and evaluate whether substantive patterns survive reasonable rule variations rather than relying on a single fixed dictionary. 3.5.2 Validating embedding-based semantic structure. The embedding module is validated by (i) testing rank-order stability of region-by-region cosine similarity matrices under defensible perturbations to embedding specifications and preprocessing, (ii) verifying that the main cluster contrasts are not artefacts of a particular projection method by comparing 2D visualizations to the underlying high-dimensional similarity structure, and (iii) bootstrapping centroid estimates to confirm that cross-regional distance patterns are not driven by a small number of influential documents. Interpretation remains comparative and structure-focused, consistent with known limits of embedding-based meaning inference. Following best-practice guidance on measurement validity in automated text analysis (Grimmer & Stewart, 2013 ), I treat embedding-based meaning measures as potentially noisy constructs and therefore report sensitivity checks for preprocessing and hyperparameter choices (Denny & Spirling, 2018 ) and for embedding stability and interpretability in applied research (Rodriguez & Spirling, 2022 ). I further interpret distances and projections in embedding space as measurable shifts in cultural and policy meaning in line with the geometry-of-meaning framework (Kozlowski et al., 2019 ). 3.5.3 Validating evaluative framing (risk lexicon) and transformer diagnostic. Evaluative framing is validated in three ways. First, I re-estimate the risk-framing index using alternative lexicon variants (narrow vs. broad dictionaries and exclusions of highly polysemous terms) to verify that cross-region contrasts are not an artifact of a single word list. Second, I replicate key patterns after excluding DAO/DeFi-labeled documents and other protocol-governance subcorpora to ensure that regional contrasts are not mechanically driven by a distinct discourse segment. Third, I compute FinBERT-based signed sentiment as an external diagnostic (Araci, 2019 ); in this corpus the score is constant (0), reinforcing the decision to rely on transparent lexicon-based framing rather than interpret transformer sentiment as psychological positivity/negativity. Accordingly, I emphasize relative cross-group differences in risk framing and interpret them as systematic variation in governance logics. 4. Results This section reports empirical patterns from the dual-corpus design as answers to RQ1-RQ3. Stage 1 documents long-run field participation and topic scaffolding (1994–2022) (Fig. 2 ). Stage 2 (2014–2022) supports full-text inference of governance archetypes, semantic proximity, and evaluative tone. Where possible, key comparisons are accompanied by uncertainty diagnostics (bootstrap confidence intervals or permutation tests) to convert visual contrasts into interpretable evidence. 4.1 RQ1: How does the functional center of blockchain governance discourse shift over time? The governance-archetype measurement reveals a long-run transition in what policy documents treat blockchain as 'for'. In the early period, blockchain is frequently framed through regulatory control and financial integrity, consistent with a problem definition centered on compliance, fraud, and market stability. Over time, the center of gravity moves toward coordination/standard-setting and, later, toward identity and public-infrastructure integration. Table 4 summarizes archetype shares by period and region, and Fig. 3 traces annual evolution in the inference corpus (2016–2022 shown). To test whether these shifts are systematic rather than visual artefacts, Table 4 estimates multinomial (or fractional-logit) models in which archetype prevalence is explained by flexible time trends (year splines) and context (region and institutional source type). The estimated time profiles confirm that the governance center of gravity moves away from pure regulatory control toward coordination and infrastructure frames in the later period, even after conditioning on region and issuer composition. Table 4 Archetype shift models (time trends, region, and issuer composition) Post-period contrast Coordination & standards Identity & infrastructure Regulatory control 2020–2022 vs 2017–2019 2.900*** [1.430, 5.880] 0.260** [0.070, 0.920] 0.59 [0.32, 1.11] Notes: Odds ratios with 95% confidence intervals. The post-period contrast is 2020–2022 vs 2017–2019 (A option). 4.2 RQ2: Where do coordination frictions concentrate? Semantic proximity patterns indicate that blockchain policy meanings do not converge into a single global center, even when archetype anchors are shared. Figure 4 visualizes the embedding space and highlights clustering by region and institutional source type. Table 6 reports centroid-based cosine similarity across regions, which can be interpreted as a coarse measure of how costly translation and mutual recognition may be for cross-border coordination. Table 5 and Table 6 indicate that cross-region semantic alignment tightened in 2020–2022 for some pairs, while others diverged, foreshadowing where policy coordination costs are likely to be highest. Table 5 Regional semantic similarity (2017–2019; cosine similarity between region centroids) Region EU/UK US/CAN Asia IGO Global South/Other EU/UK 1.000 0.600 0.830 0.830 US/CAN 0.600 1.000 0.660 0.610 Asia 0.830 0.660 1.000 0.660 IGO 1.000 Global South/Other 0.830 0.610 0.660 1.000 Notes: Entries are cosine similarities between region-level centroids in the TF–IDF + SVD semantic space (higher = more semantically aligned). To quantify uncertainty and enable sharp comparisons (e.g., whether EU-Asia similarity is lower than EU-US similarity), Supplementary Table S4 reports bootstrap confidence intervals for centroid similarities, and Supplementary Table S5 reports selected difference contrasts. These uncertainty-aware comparisons support interpretation of coordination frictions without requiring causal identification. Uncertainty for the similarity estimates and selected pairwise contrasts are reported in Supplementary Tables S4–S5. Table 6 Regional semantic similarity (2020–2022; cosine similarity between region centroids) Region EU/UK US/CAN Asia IGO Global South/Other EU/UK 1.000 0.660 0.560 0.570 0.540 US/CAN 0.660 1.000 0.540 0.680 0.260 Asia 0.560 0.540 1.000 0.470 0.240 IGO 0.570 0.680 0.470 1.000 0.240 Global South/Other 0.540 0.260 0.240 0.240 1.000 Notes: Same embedding pipeline as Table 6 A; differences reflect period-specific document sets. Building on these similarities, Fig. 5 reports a Coordination Friction Heatmap for region pairs and time windows. Higher friction indicates greater expected translation and alignment costs (semantic distance, archetype divergence, and risk gaps). The heatmap is intended as a decision aid for standard-setting and mutual-recognition efforts: it identifies where coordination is most likely to stall because jurisdictions are discussing the same technology in functionally different ways. 4.3 RQ3: Is risk sensitivity a region effect or a framing effect? Evaluative tone varies systematically across contexts. Table 7 reports average lexicon-based risk index (per 1,000 tokens) by region and archetype, showing that some regions maintain more cautious, risk-sensitive framing even when discussing similar governance functions. This pattern is consistent with the idea that institutions differ not only in what they prioritize (archetypes) but also in how they normatively evaluate opportunities and risks. Table 7 Evaluative risk framing by region and archetype Region Coordination & standards Identity & infrastructure Regulatory control Unclassified EU/UK 2.81 (N = 24) 9.43 (N = 4) 5.78 (N = 85) 2.64 (N = 26) US/CAN 5.40 (N = 4) 3.96 (N = 12) 9.85 (N = 104) 2.46 (N = 32) Asia 4.50 (N = 8) 3.00 (N = 6) 5.31 (N = 6) IGO 1.77 (N = 3) 7.94 (N = 17) Global South/Other 2.41 (N = 13) 4.04 (N = 3) 2.79 (N = 30) 2.39 (N = 18) Notes: Lexicon-based risk index per 1,000 tokens. FinBERT sentiment is constant in this corpus (0), so I use a transparent risk lexicon for evaluative framing. To distinguish region effects from framing and issuer effects, Supplementary Table S1 regresses document-level risk index on region indicators, archetype indicators, and institutional source-type controls (with year fixed effects or flexible trends as appropriate). The results show whether regional caution/optimism persists net of governance frame composition, thereby clarifying whether region proxies for deeper differences in how governance problems are framed or simply reflects different mixes of document types. Notes: Linear model with year fixed effects. Coefficients reported with robust standard errors in parentheses where available. Alternative specifications and full outputs are reported in Supplementary Note S3 (Supplementary Tables S6–S7). 4.4 Validity and robustness Validity checks for the two-stage design are summarized in Table 8 A, and Table 8 B reports robustness checks for key descriptive patterns. Supplementary Information (Supplementary Notes S1–S4 and Supplementary Tables S2–S9) reports extended model outputs, bootstrap uncertainty quantification, and additional robustness checks that support the stability of the main temporal and cross-regional patterns. Table 8 A. Validity checks (summary) Check Evidence Implication Extractability constraint Stage 2 limited to stable, extractable full text (PDFs); rules documented. Reduces noise in semantic mapping and framing measurement. Representativeness (raking) Stage 2 reweighted to Stage 1 margins (Appendix A4). Long-run patterns are stable under reweighting. Classification validation Archetype anchors and validation diagnostics (Appendix A6). Archetype assignment is not driven by chance. Table 8 B. Robustness checks (summary) Robustness check What changes Result Exclude DAO/DeFi-labelled documents Recompute archetype patterns without DAO/DeFi subset Core transitions remain directionally stable. Bootstrap uncertainty Resample documents within region–period cells Key semantic contrasts remain within the same qualitative range. Friction decomposition Report semantic distance, archetype divergence, and risk-gap components Heatmap patterns are not driven by a single component. Notes: These checks are designed to be non-causal but decision-relevant, aligning with the mapping goal. Detailed diagnostics and extended outputs are reported in Supplementary Notes S1–S4 (Supplementary Tables S2–S9). 5. Discussion Computational mapping can easily read as mere visualization unless it is tied to decision-relevant comparisons and tests. I therefore structure the findings around three research questions, each paired with a minimal inferential check and a practical coordination output. The central policy implication is that cross-border coordination problems are often semantic before they are technical. When jurisdictions attach different governance purposes to blockchain (e.g., financial integrity, national identity infrastructure, or administrative modernization), attempts at wholesale harmonization are likely to stall. The Coordination Friction Index makes this logic actionable by identifying where semantic distance and frame divergence are highest. These coordination frictions are not reducible to vocabulary differences. Work on international standards and transnational regulation shows that alignment failures often reflect clashes between private authority, state regulation, and market power in the production of global rules (Mattli & Büthe, 2003 ; Büthe & Mattli, 2011 ; Abbott & Snidal, 2009 ; Drezner, 2007 ). In networked economies, regulatory choices can be amplified through control of infrastructural nodes and interdependence, making semantic divergence a channel of geopolitical leverage rather than mere misunderstanding (Farrell & Newman, 2019 ). The CFI operationalizes this institutional-logics dimension by treating cross-region semantic distance as an early warning signal of where harmonization, mutual recognition, or interoperability arrangements will face persistent friction. CFI can also be read as a diagnostic of friction at the science–policy interface: when documents translate technical uncertainty into policy-relevant categories, disagreements emerge over what counts as credible evidence, salient risk, and legitimate authority (Cash et al., 2003 ; Jasanoff, 2004 ; Pielke, 2007 ). By quantifying systematic mismatches in evaluative framing and governance purpose, the index complements qualitative accounts of co-production and boundary work, which show that policy meaning is produced through iterative negotiation rather than linear knowledge transfer (Maas, 2022 ). For practitioners, the results imply three concrete coordination artifacts. First, a Coordination Heatmap ranks region pairs and time windows by expected alignment costs; this can be used to prioritize bilateral working groups, mutual-recognition pilots, or translation efforts. Second, a layered coordination playbook sequences negotiation in the order that minimizes deadlock: start with minimum viable alignment on consumer protection and anti-money-laundering interfaces, then move to interoperability and standards, and only then to deeper identity and infrastructure integration. Third, a translation table documents region- and issuer-specific vocabulary for similar governance functions (e.g., how 'identity' is institutionalized as KYC, digital ID, or infrastructure security), reducing the risk that coordination fails because equivalent functions are labeled differently. Methodologically, the dual-corpus design addresses a practical reality of big policy repositories: coverage and extractability do not coincide. Rather than treating missing full text as nuisance, the paper treats extractability as a measurement constraint that can be audited and partially corrected. This design is generalizable to other emerging technologies where policy attention arrives faster than stable document formats and where the governance field spans heterogeneous institutions. Limitations follow from the same design choices. Stage 2 inference is constrained to extractable PDFs, and embedding and framing outputs remain proxies that must be interpreted comparatively rather than as direct measures of latent intent. Future work can extend the approach by (i) integrating additional languages with multilingual models, (ii) linking discourse shifts to measurable regulatory outcomes (e.g., rule adoption, enforcement intensity), and (iii) tracing how policy discourse interacts with media and market narratives in real time. 6. Conclusion By combining broad metadata coverage with full-text inference, this study maps how blockchain is redefined in policy discourse from a crypto-finance experiment into a contested digital-governance infrastructure. The evidence shows (i) a measurable shift in governance archetypes over time, (ii) persistent regional clustering in semantic meaning that signals where coordination frictions are likely to be highest, and (iii) systematic differences in evaluative tone that cannot be reduced to archetype composition alone. Beyond describing the field, the paper offers decision-oriented outputs - a friction index, a coordination heatmap, and a translation table - that help policymakers anticipate where standard-setting and mutual recognition are most likely to stall and how to stage coordination in practice. Declarations Data availability This study draws metadata and document identifiers from Overton under license and therefore cannot redistribute raw Overton exports or source PDFs. To support transparency and reuse, the replication package includes (i) the list of Overton document IDs used to construct the two-stage corpus, (ii) all derived data tables used to generate figures and tables (e.g., archetype assignments, descriptor shares, semantic centroids, and CFI matrices), and (iii) scripts that reproduce the full processing and analysis pipeline for users with licensed access to Overton. Code availability Upon acceptance, all analysis code (Python) and configuration files necessary to reproduce the results will be made available as part of the replication package. The data used in this study are subject to access restrictions and therefore cannot be publicly shared. Competing interests The author declares no competing interests. Ethics approval and consent to participate : This study does not involve human participants or identifiable personal data. Ethical approval and consent were not required. Author Contribution H.K. conceived and designed the study, curated the data, developed the methodology, performed the formal analysis, and wrote the analysis code. H.K. prepared all figures and tables and wrote the original draft of the manuscript. H.K. reviewed and edited the manuscript and takes responsibility for the integrity of the work as a whole. References Abbott, K. W., & Snidal, D. (2009). Strengthening international regulation through transnational new governance: Overcoming the orchestration deficit. Vanderbilt Journal of Transnational Law, 42, 501–578. Allen, D. W. E., Berg, C., Lane, A. M., & Potts, J. (2019). Blockchains and the economic institutions of capitalism. Journal of Institutional Economics, 15(4), 569–589. Al‑Saqaf, W., & Seidler, N. (2017). Blockchain technology for social impact: Opportunities and challenges ahead. Journal of Cyber Policy, 2(3), 338–354. Anagnostopoulos, I. (2018). Fintech and regtech: Impact on regulators and banks. Journal of Economics and Business, 100, 7–25. Angelis, J., & da Silva, E. R. (2019). Blockchain adoption: A value driver perspective. Business Horizons, 62(3), 307–314. Araci, D. (2019). FinBERT: Financial sentiment analysis with pre-trained language models. arXiv preprint arXiv:1908.10063. Araos, M., Berrang‑Ford, L., Ford, J. D., Austin, S. E., Biesbroek, R., & Lesnikowski, A. (2016). Climate change adaptation planning in large cities: A systematic global assessment. Urban Climate, 17, 176–197. Arner, D. W., Barberis, J., & Buckley, R. P. (2017). FinTech, RegTech, and the reconceptualization of financial regulation. Northwestern Journal of International Law & Business, 37(3), 371–414. Ash, E., & Hansen, S. (2023). Text algorithms in economics. Annual Review of Economics, 15(1), 659–688. Baker, S. R., Bloom, N., & Davis, S. J. (2016). Measuring economic policy uncertainty. Quarterly Journal of Economics, 131(4), 1593–1636. Berrang‑Ford, L., Biesbroek, R., Ford, J. D., Lesnikowski, A., Tanabe, A., & Wang, F. M. (2021). A systematic global stocktake of evidence on human adaptation to climate change. Nature Climate Change, 11(11), 989–1000. Berryhill, J., Bourgery, T., & Hanson, A. (2018). Blockchains unchained: Blockchain technology and its use in the public sector. OECD Working Papers on Public Governance, No. 28. Biesbroek, R., Dupuis, J., Jordan, A., Wellstead, A., Howlett, M., Cairney, P., … Rayner, T. (2015). Opening up the black box of adaptation decision‑making. Nature Climate Change, 5(6), 493–494. Bornmann, L., Haunschild, R., & Marx, W. (2022). How relevant is climate change research for climate change policy? An analysis of policy citations of climate change publications. PLOS ONE, 17(9), e0274693. Büthe, T., & Mattli, W. (2011). The New Global Rulers: The Privatization of Regulation in the World Economy. Princeton University Press. Caldara, D., & Iacoviello, M. (2022). Measuring geopolitical risk. American Economic Review, 112(4), 1194–1225. Campbell‑Verduyn, M. (Ed.). (2017). Bitcoin and beyond: Cryptocurrencies, blockchains, and global governance. Routledge. Cash, D. W., Clark, W. C., Alcock, F., Dickson, N. M., Eckley, N., & Jäger, J. (2003). Knowledge systems for sustainable development. Proceedings of the National Academy of Sciences, 100(14), 8086–8091. Catalini, C., & Gans, J. S. (2020). Some simple economics of the blockchain. Communications of the ACM, 63(7), 80–90. Ceron, A., & Negri, F. (2016). Using sentiment analysis to support the action of policy‑makers: A case study on the Italian labour market reform. Policy & Internet, 8(3), 258–278. Cong, L. W., & He, Z. (2019). Blockchain disruption and smart contracts. Review of Financial Studies, 32(5), 1754–1797. De Filippi, P., & Wright, A. (2018). Blockchain and the law: The rule of code. Harvard University Press. Denny, M., & Spirling, A. (2018). Text preprocessing for unsupervised learning: Why it matters, when it misleads, and what to do about it. Political Analysis, 26(2), 168–189. Devlin, J., Chang, M. W., Lee, K., & Toutanova, K. (2019). BERT: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of NAACL-HLT 2019 (pp. 4171–4186). Drezner, D. W. (2007). All Politics Is Global: Explaining International Regulatory Regimes. Princeton University Press. El‑Haddadeh, R., Osmani, M., Hindi, N., Fadlalla, A., & Darem, A. (2021). Examining the role of blockchain technology in improving public sector services: A systematic literature review. Information Systems Frontiers, 23(6), 1405–1423. Farrell, H., & Newman, A. L. (2019). Weaponized interdependence: How global economic networks shape state coercion. International Security, 44(1), 42–79. Finck, M. (2018). Blockchain regulation and governance in Europe. Cambridge University Press. Garrod, B., & Samuel, J. (2021). Blockchain applications in government: A systematic review. Government Information Quarterly, 38(4), 101619. Gikay, A. A. (2018). Bitcoin and the blockchain as possible corporate governance tools. The Journal of Business, Entrepreneurship & the Law, 11(2), 281–324. Gomber, P., Koch, J.‑A., & Siering, M. (2017). Digital finance and FinTech: Current research and future research directions. Journal of Business Economics, 87(5), 537–580. Grimmer, J., & Stewart, B. M. (2013). Text as data: The promise and pitfalls of automatic content analysis methods for political texts. Political Analysis, 21(3), 267–297. Grimmer, J., Roberts, M. E., & Stewart, B. M. (2022). Text as Data: A New Framework for Machine Learning and the Social Sciences. Princeton University Press. Hassan, T. A., Hollander, S., van Lent, L., & Schwedeler, M. (2019). Firm-level political risk: Measurement and effects. Quarterly Journal of Economics, 134(4), 2135–2202. Hawlitschek, F., Notheisen, B., & Teubner, T. (2018). The limits of trust‑free systems: A literature review on blockchain technology and trust in the sharing economy. Electronic Commerce Research and Applications, 29, 50–63. Hileman, G., & Rauchs, M. (2017). Global cryptocurrency benchmarking study. Cambridge Centre for Alternative Finance. Hsieh, Y. Y., Vergne, J. P. J., & Wang, S. (2017). The internal and external governance of blockchain-based organizations: Evidence from cryptocurrencies. In Bitcoin and beyond (pp. 48-68). Routledge. Hughes, L., Dwivedi, Y. K., Misra, S. K., Rana, N. P., Raghavan, V., & Akella, V. (2019). Blockchain research, practice and policy: Applications, benefits, limitations, emerging research themes and research agenda. International Journal of Information Management, 49, 114–129. Iman, N. (2020). The rise and rise of financial technology: The good, the bad, and the verdict. Cogent Business & Management, 7(1), 1725309. Jasanoff, S. (Ed.). (2004). States of Knowledge: The Co-production of Science and Social Order. Routledge. Jia, W., Liang, R., & Xu, S. (2022). The public sentiment analysis of “Double Reduction Policy” on Weibo platform. Frontiers in Psychology, 13, 947756. Kozlowski, A. C., Taddy, M., & Evans, J. A. (2019). The geometry of culture: Analyzing the meanings of class through word embeddings. American Sociological Review, 84(5), 905–949. Kozlowski, A. C., Taddy, M., & Evans, J. A. (2019). The geometry of culture: Analyzing meaning through word embeddings. American Sociological Review, 84(5), 905–949. Le, Q., & Mikolov, T. (2014). Distributed representations of sentences and documents. In Proceedings of the 31st International Conference on Machine Learning (ICML) (pp. 1188–1196). Lee, I., & Shin, Y. J. (2018). Fintech: Ecosystem, business models, investment decisions, and challenges. Business Horizons, 61(1), 35–46. Lesnikowski, A., Belfer, E., Rodman, E., Smith, J., Biesbroek, R., Wilkerson, J. D., … Berrang‑Ford, L. (2019). Frontiers in data analytics for adaptation research: Topic modeling. Wiley Interdisciplinary Reviews: Climate Change, 10(3), e576. Loughran, T., & McDonald, B. (2011). When is a liability not a liability? Textual analysis, dictionaries, and 10-Ks. Journal of Finance, 66(1), 35–65. Lumineau, F., Wang, W., & Schilke, O. (2021). Blockchain governance—A new way of organizing collaborations?. Organization Science, 32(2), 500-521. Li, J., Zhang, Y., Chen, H., & Luo, Y. (2023). Semantics of policy discourse in digital government. Government Information Quarterly, 40(1), 101761. Liu, Y., Lu, Q., Yu, G., Paik, H. Y., & Zhu, L. (2022). Defining blockchain governance principles: A comprehensive framework. Information systems, 109, 102090. Ma, D., Zhu, X., Wang, Y., & Li, J. (2024). Same same but different: How policies frame societal‑level digital transformation. Government Information Quarterly, 41(1), 101877. Maas, T. Y. (2022). Co-producing the science–policy interface: Towards common but differentiated responsibilities. Humanities and Social Sciences Communications, 9, 94. Mandell, C. (2022). Theoretical foundations and limits of word embeddings. Journal of Cultural Analytics, 7(1), 1–28. Mattli, W., & Büthe, T. (2003). Setting international standards: Technological rationality or primacy of power? World Politics, 56(1), 1–42. Mikolov, T., Chen, K., Corrado, G., & Dean, J. (2013). Efficient estimation of word representations in vector space. arXiv preprint arXiv:1301.3781. Morkunas, V. J., Paschen, J., & Boon, E. (2019). How blockchain technologies impact your business model. Business Horizons, 62(3), 295–306. Nakamoto, S. (2008). Bitcoin: A peer‑to‑peer electronic cash system. https://bitcoin.org/bitcoin.pdf Nitzbon, J., Wang, Y., & Lu, H. (2023). Measuring policy debate in a regrowing city by sentiment analysis of online media. Cities, 141, 104469. Rodriguez, P. L., & Spirling, A. (2022). Word embeddings: What works, what doesn't, and how to tell the difference for applied research. Journal of Politics, 84(1), 101–115. Ølnes, S., Ubacht, J., & Janssen, M. (2017). Blockchain in government: Benefits and implications of distributed ledger technology for information sharing. Government Information Quarterly, 34(3), 355–364. Panagiotopoulos, P., Barnett, J., Bigdeli, A. Z., & Sams, S. (2017). Social media in public policy making: A case in e‑participation. Government Information Quarterly, 34(1), 54–62. Pielke, R. A. Jr. (2007). The Honest Broker: Making Sense of Science in Policy and Politics. Cambridge University Press. Rauchs, M., Glidden, A., Gordon, B., Pieters, G. C., Recanatini, M., & Zhang, B. (2018). Distributed ledger technology systems: A conceptual framework. Cambridge Centre for Alternative Finance. Schlierkamp, J., Balsiger, J., & Lörcher, T. (2025). Uncovering the science‑policy interface: Applying bibliometric approaches to the wildfire risk management domain. Humanities and Social Sciences Communications, 12, 77. Schmidt, V. A. (2008). Discursive institutionalism: The explanatory power of ideas and discourse. Annual Review of Political Science, 11, 303–326. Shore, C., & Wright, S. (2011). Conceptualising policy: Technologies of governance and the politics of visibility. Ethnography, 12(1), 1–26. Sucupira, G., & de Albuquerque, J. P. (2022). Sentiment analysis of public services for smart society: Literature review and future research directions. Government Information Quarterly, 39(5), 101749. Szomszor, M., & Adie, E. (2022). Overton: A bibliometric database of policy document citations. Quantitative Science Studies, 3(3), 624–650. Steinhardt, H. C., & Göbel, C. (2026). Protecting society from itself: How the social credit system is justified on Chinese social media. Big Data & Society, 1–15. https://doi.org/10.1177/20539517251403951 Tan, B. S., & Low, K. Y. (2019). Blockchain as a confidence machine: The problem of trust and challenges of governance. Technology in Society, 59, 101151. Tan, E., Mahula, S., & Crompvoets, J. (2022). Blockchain governance in the public sector: A conceptual framework for public management. Government Information Quarterly, 39(1), 101625. Tang, H., Ou, M., & Zheng, H. (2025). Between promise and peril: Users’ risk–benefit trade-offs in their generative AI usage. Big Data & Society, 1–14. https://doi.org/10.1177/20539517251410046 Treleaven, P., Brown, R. G., & Yang, D. (2017). Blockchain technology in finance. Computer, 50(9), 14–17. van der Maaten, L., & Hinton, G. (2008). Visualizing data using t-SNE. Journal of Machine Learning Research, 9, 2579–2605. Van der Meulen, B., & Rip, A. (2000). Evaluation of societal quality of public sector research in the Netherlands. Research Evaluation, 9(1), 11–25. Weber, R. H. (2018). Corporate governance and blockchains: New paradigms for capital markets? Journal of Banking Regulation, 19(3), 188–195. Werbach, K. (2018). The blockchain and the new architecture of trust. MIT Press. Wilsdon, J., Allen, L., Belfiore, E., Campbell, P., Curry, S., Hill, S., … Johnson, B. (2015). The metric tide: Report of the independent review of the role of metrics in research assessment and management. HEFCE. Xu, J., Chen, M., Kou, G., & Peng, Y. (2019). Understanding the development of blockchain technology: A patent analysis. Technological Forecasting and Social Change, 148, 119735. Yermack, D. (2017). Corporate governance and blockchains. Review of Finance, 21(1), 7–31. Yeung, K. (2019). Regulation by blockchain: The emerging battle for supremacy between the code of law and code as law. Modern Law Review, 82(2), 207–239. Zetzsche, D. A., Buckley, R. P., Barberis, J. N., & Arner, D. W. (2017). Regulating a revolution: from regulatory sandboxes to smart regulation. Fordham J. Corp. & Fin. L., 23, 31. Zwitter, A., & Hazenberg, J. (2020). Decentralized network governance: blockchain technology and the future of regulation. Frontiers in Blockchain, 3, 12. Additional Declarations No competing interests reported. Supplementary Files HSSCommsmainSFINAL.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8702396","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":593092293,"identity":"3ca9d78a-53b2-4503-9630-62dda15b7459","order_by":0,"name":"hana kim","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAwUlEQVRIiWNgGAWjYJCCAwkVNjB2AmHlPAwMjAcenEkjTQvzwYdth0nQYs/e/OBAAtt5e3OJBMYPPxjS8gnbwnPM4EACz+3EnTMSmCV7GHIsGwhqkUgAapG4nWBwI4FBmoGhwoCwLfLPPxxIMDhnD9TC/Js4LRI8QFsSDjBuuJHABrQlhwgtZ3IKDiQcSE7ccOZhm2WPQRphLeztxzd//PnPzt7gePLhGz8qkglrQQKMDQwMJGkYBaNgFIyCUYATAABjcDzt9Z7tEgAAAABJRU5ErkJggg==","orcid":"","institution":"Korea Advanced Institute of Science and Technology","correspondingAuthor":true,"prefix":"","firstName":"hana","middleName":"","lastName":"kim","suffix":""}],"badges":[],"createdAt":"2026-01-26 16:38:18","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8702396/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8702396/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":103052875,"identity":"9d6f2595-f0b0-493d-9e20-06e4f6ee6b08","added_by":"auto","created_at":"2026-02-20 08:07:55","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":16582,"visible":true,"origin":"","legend":"\u003cp\u003eAnalytical pipeline\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8702396/v1/248da578dea5bccd27ffb007.png"},{"id":103052877,"identity":"9891c12c-e9e6-4065-a5fd-e03eac4c6269","added_by":"auto","created_at":"2026-02-20 08:07:55","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":269669,"visible":true,"origin":"","legend":"\u003cp\u003eDescriptor shifts and actor participation in the full metadata corpus, 1994–2022.\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8702396/v1/b6204800af1406b057b3b4f5.png"},{"id":103056486,"identity":"08bca9af-b45b-446b-b55d-022cb2d08f6d","added_by":"auto","created_at":"2026-02-20 09:11:43","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":163251,"visible":true,"origin":"","legend":"\u003cp\u003eEvolution of archetype shares by region (Stage 2; annual series shown for 2016–2022).\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8702396/v1/17f20daf0528b64b041c9352.png"},{"id":103052881,"identity":"618e5e60-c13e-4a3b-b0de-355647987dc6","added_by":"auto","created_at":"2026-02-20 08:07:55","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":467548,"visible":true,"origin":"","legend":"\u003cp\u003eDocument embedding map\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8702396/v1/ca669806bc7dee2b4088b2a2.png"},{"id":103052878,"identity":"805dcb6c-6bdc-465c-8b7e-99da46701bbe","added_by":"auto","created_at":"2026-02-20 08:07:55","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":195413,"visible":true,"origin":"","legend":"\u003cp\u003eCoordination Friction Heatmap (region pairs x time windows)\u003c/p\u003e","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-8702396/v1/c63f1b6a5cd969f30e18251e.png"},{"id":106724219,"identity":"03e9284e-9a7c-4dc4-bce8-687f0de8d233","added_by":"auto","created_at":"2026-04-12 18:26:46","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2420747,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8702396/v1/57fe052f-0193-42c4-8562-858313b1b59d.pdf"},{"id":103052876,"identity":"dee0b773-741a-4b5e-b684-5ef2c83ceae4","added_by":"auto","created_at":"2026-02-20 08:07:55","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":44147,"visible":true,"origin":"","legend":"","description":"","filename":"HSSCommsmainSFINAL.docx","url":"https://assets-eu.researchsquare.com/files/rs-8702396/v1/66941f93f9ecb8a2260483b3.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"From Crypto Trust to Digital Governance: Mapping Coordination Frictions in Global Policy Discourse","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eBlockchain has moved from a niche protocol debate to a recurring object of governmental problem definition - a site where institutions negotiate compliance, identity, and cross-border coordination in data-driven economies. Yet existing scholarship typically observes this transformation through case studies, national regulatory histories, or sector-specific episodes, which makes it difficult to compare how policy meaning travels, fragments, or converges across jurisdictions over long horizons.\u003c/p\u003e \u003cp\u003eThis paper treats policy discourse itself as a governance signal that can be measured at scale. Using Overton, I assemble a large global corpus of blockchain-related policy documents and map how the field evolves in (i) institutional participation, (ii) topical scaffolding, and (iii) semantic and evaluative framing. The key challenge is that broad coverage and reliable full-text inference rarely coincide: metadata repositories provide breadth, while computational text inference depends on extractable, linguistically stable PDFs.\u003c/p\u003e \u003cp\u003eTo address this coverage-inference trade-off, I use a dual-corpus, two-stage design. Stage 1 uses Overton metadata (N\u0026thinsp;=\u0026thinsp;17,019; 1994\u0026ndash;2022) to map the long-run field and motivate historically grounded periods. Stage 2 constructs a full-text inference corpus (N\u0026thinsp;=\u0026thinsp;393; 2014\u0026ndash;2022) that supports three complementary modules: transparent governance-archetype measurement, document-embedding maps of semantic proximity, and evaluative risk framing via a transparent lexicon (with a transformer diagnostic reported as a robustness check). Importantly, selection into extractable full text is diagnosed and, where possible, stress-tested rather than assumed away.\u003c/p\u003e \u003cp\u003eThe analysis is organized around three research questions that convert field mapping into decision-relevant comparisons. RQ1 asks: how does the functional center of blockchain governance discourse shift over time - from regulatory control toward coordination/standards and identity/infrastructure - and when do inflection points occur? RQ2 asks: where do cross-jurisdiction coordination frictions concentrate, as measured by semantic distance and archetype divergence between region pairs? RQ3 asks: is evaluative tone (risk sensitivity) primarily a region effect, or is it better explained by governance archetypes and institutional source composition?\u003c/p\u003e \u003cp\u003eThe paper contributes to a interdisciplinary scholarship on social implications of emerging technologies in three ways. Substantively, it offers a longitudinal, comparative map of how an emerging technology is repurposed into digital-governance infrastructure through policy discourse. Methodologically, it provides a reproducible dual-corpus template that separates field coverage from text inference while making extractability an explicit measurement constraint. Practically, it translates empirical outputs into coordination tools: a friction heatmap that identifies where semantic misalignment is highest, a layered playbook that sequences minimum viable alignment before deeper harmonization, and a translation table that documents how similar governance functions are named differently across regions and institutions.\u003c/p\u003e \u003cp\u003eThe remainder proceeds as follows. Section \u003cspan refid=\"Sec2\" class=\"InternalRef\"\u003e2\u003c/span\u003e situates the study in scholarship on technology governance, policy discourse, and computational mapping. Section \u003cspan refid=\"Sec6\" class=\"InternalRef\"\u003e3\u003c/span\u003e describes the dual-corpus design and inference modules. Section \u003cspan refid=\"Sec23\" class=\"InternalRef\"\u003e4\u003c/span\u003e reports results as answers to RQ1-RQ3, with uncertainty and robustness diagnostics. Section \u003cspan refid=\"Sec28\" class=\"InternalRef\"\u003e5\u003c/span\u003e discusses implications for cross-border coordination and the study's limitations.\u003c/p\u003e"},{"header":"2. Literature Review","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1 Blockchain as an Institutional and Governance Technology\u003c/h2\u003e \u003cp\u003eBlockchain emerged from its origins as a cryptographic protocol for peer-to-peer currency exchange (Nakamoto, \u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e2008\u003c/span\u003e) to a broader institutional technology that reconfigures how coordination, verification, and rule enforcement operate in digital environments. Rather than functioning solely as a decentralized transaction mechanism, blockchain introduces architectures that reassign authority and compliance responsibilities from intermediaries to protocol-based consensus systems (Catalini \u0026amp; Gans, \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Allen et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Rauchs et al., \u003cspan citationid=\"CR63\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). This shift has positioned blockchain as a governance-relevant infrastructure whose implications extend well beyond cryptocurrency markets, including adoption logics and organizational value drivers (Angelis \u0026amp; da Silva, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2019\u003c/span\u003e) as well as broader fintech ecosystem transformations (Lee \u0026amp; Shin, \u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Iman, \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2020\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eStudies in financial economics demonstrate that programmable logic and decentralized validation reshape transaction costs, corporate governance mechanisms, and contractual enforcement processes (Yermack, \u003cspan citationid=\"CR80\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Cong \u0026amp; He, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Related work in finance and computing highlights blockchain\u0026rsquo;s expanding role in financial infrastructures and market operations (Treleaven et al., \u003cspan citationid=\"CR73\" class=\"CitationRef\"\u003e2017\u003c/span\u003e), while governance-oriented legal and management research discusses how blockchains may function as corporate governance tools and reshape capital-market paradigms (Gikay, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Weber, \u003cspan citationid=\"CR76\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Complementing these perspectives, patent-based evidence charts the technological development trajectory of blockchain and its diffusion across domains (Xu et al., \u003cspan citationid=\"CR79\" class=\"CitationRef\"\u003e2019\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eIn the public sector, governments have experimented with blockchain applications across identity systems, procurement, land registries, and record-management infrastructures, often motivated by transparency and auditability goals (\u0026Oslash;lnes et al., \u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Hileman \u0026amp; Rauchs, \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Hughes et al., \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Berryhill et al., \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Systematic reviews and conceptual work further consolidate this domain by mapping how blockchain is operationalized in government and what governance challenges recur across cases (Garrod \u0026amp; Samuel, \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; El-Haddadeh et al., 2021; Tan et al., \u003cspan citationid=\"CR71\" class=\"CitationRef\"\u003e2022\u003c/span\u003e), including more recent efforts to translate these insights into practical governance principles and a comprehensive framework for blockchain systems (Liu et al., \u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). These developments underscore blockchain\u0026rsquo;s emerging institutional importance as states explore alternative models of trust, information sharing, and accountability.\u003c/p\u003e \u003cp\u003eGovernance research highlights that blockchain redistributes\u0026mdash;rather than eliminates\u0026mdash;trust by shifting reliance from institutional intermediaries to protocol designers, validators, and platform operators (Hawlitschek et al., \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Tan \u0026amp; Low, \u003cspan citationid=\"CR70\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Legal scholarship examines the tensions between immutable code and adaptive regulatory regimes, pointing to unresolved questions around liability, compliance, and the interaction between on-chain and off-chain governance (De Filippi \u0026amp; Wright, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Werbach, \u003cspan citationid=\"CR77\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Yeung, \u003cspan citationid=\"CR81\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Wright \u0026amp; De Filippi, 2015). Political-economy perspectives similarly highlight the rise of \u0026ldquo;shadow governance,\u0026rdquo; where private consortia and standards bodies set de facto rules governing blockchain infrastructures (Campbell-Verduyn, 2017). In parallel, fintech and regtech scholarship emphasizes how regulators and financial institutions respond to these shifts through new supervisory technologies and regulatory design, including sandbox experimentation and \u0026ldquo;smart regulation\u0026rdquo; approaches (Anagnostopoulos, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Zetzsche et al., \u003cspan citationid=\"CR82\" class=\"CitationRef\"\u003e2017\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThese streams portray blockchain as a contested socio-technical institution characterized by distributed authority and heterogeneous governance imaginaries. Yet scholarship remains fragmented across sectors and regions, resulting in limited understanding of how governments collectively frame blockchain\u0026rsquo;s institutional significance or how this framing evolves over time.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2 Policy Documents as Arenas of Technology Framing and Public-Sector Interpretation\u003c/h2\u003e \u003cp\u003ePolicy documents\u0026mdash;national strategies, white papers, regulatory consultations, guidelines, and expert reports\u0026mdash;serve as arenas where governments interpret technological change and articulate expectations for its governance. Discursive institutionalism conceptualizes such documents as sites where ideas, norms, and governance logics are circulated, stabilized, and contested (Schmidt, \u003cspan citationid=\"CR65\" class=\"CitationRef\"\u003e2008\u003c/span\u003e; Shore \u0026amp; Wright, \u003cspan citationid=\"CR66\" class=\"CitationRef\"\u003e2011\u003c/span\u003e). These texts reveal how institutions define opportunities and risks, legitimizing certain technological pathways while delimiting others. They also intersect with longer-standing discussions on how public-sector research is evaluated for societal relevance and how policy-facing knowledge infrastructures are organized (Van der Meulen \u0026amp; Rip, \u003cspan citationid=\"CR75\" class=\"CitationRef\"\u003e2000\u003c/span\u003e), as well as debates over the use of metrics in research assessment and management (Wilsdon et al., \u003cspan citationid=\"CR78\" class=\"CitationRef\"\u003e2015\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe growth of large-scale infrastructures like the Overton database has enabled systematic mapping of science\u0026ndash;policy linkages and cross-national policy agenda patterns (Szomszor \u0026amp; Adie, \u003cspan citationid=\"CR68\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Bornmann et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Such corpora provide visibility into which institutions contribute to policy discourse, the scientific evidence they cite, and how their priorities shift over time. Recent bibliometric work illustrates how these approaches can be operationalized to uncover science\u0026ndash;policy interfaces in concrete policy domains (Schlierkamp et al., \u003cspan citationid=\"CR64\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eComputational text-analytic methods have deepened the capacity to analyze these documents. Topic models and clustering techniques have revealed thematic structures in climate adaptation, environmental governance, public health, and digital government policy (Lesnikowski et al., \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Ma et al., \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Araos et al., \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Biesbroek et al., \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). Embedding-based semantic models illuminate conceptual proximity, discursive alignment, and narrative shifts across political and technological domains (Kozlowski et al., \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Li et al., \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Sentiment and evaluative-language models further capture how institutions express support, caution, or concern when articulating governance positions around emerging technologies (Ceron \u0026amp; Negri, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2016\u003c/span\u003e, Wang et al., 2023), and adjacent digital-government research extends these tools to map public sentiment and policy debate dynamics across platforms and media environments (Panagiotopoulos et al., \u003cspan citationid=\"CR61\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Jia et al., \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Nitzbon et al., \u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Sucupira \u0026amp; de Albuquerque, \u003cspan citationid=\"CR67\" class=\"CitationRef\"\u003e2022\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eRecent scholarship in Big Data \u0026amp; Society shows how large-scale analysis of policy and social-media discourse can produce decision-relevant governance insights\u0026mdash;for example, by quantifying justification frames around surveillance infrastructures (Steinhardt \u0026amp; G\u0026ouml;bel, \u003cspan citationid=\"CR69\" class=\"CitationRef\"\u003e2026\u003c/span\u003e) or by unpacking how users negotiate risk\u0026ndash;benefit trade-offs around generative AI (Tang et al., \u003cspan citationid=\"CR72\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). This paper extends this orientation to blockchain policy documents, pairing field mapping with explicit comparative tests that are directly interpretable for cross-jurisdiction coordination.\u003c/p\u003e \u003cp\u003eDespite these methodological advances, blockchain remains underexamined as a policy-discourse domain. Existing studies focus largely on parliamentary debates, regulatory announcements, or specific national strategies (De Filippi \u0026amp; Wright, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Yeung, \u003cspan citationid=\"CR81\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). While recent scholarship has increasingly conceptualized blockchain as a novel mode of organizational collaboration (Lumineau et al., \u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e2021\u003c/span\u003e) and analyzed the distinct dynamics of its internal and external governance (Hsieh \u0026amp; Vergne, 2023), these conceptual advancements have yet to be fully integrated into a comprehensive policy-discourse framework.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e2.3 Analytical Perspectives for Mapping Global Technology Policy Discourse\u003c/h2\u003e \u003cp\u003eSynthesizing insights from blockchain governance research and policy-discourse analytics reveals several major empirical gaps. First, although blockchain\u0026rsquo;s institutional relevance has expanded rapidly, comprehensive evidence remains limited regarding which actors\u0026mdash;national governments, international organizations, regulatory agencies, NGOs, and expert institutions\u0026mdash;have contributed to the production of blockchain policy documents over time. Existing analyses focus primarily on isolated regulatory episodes or single-country contexts (De Filippi \u0026amp; Wright, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Yeung, \u003cspan citationid=\"CR81\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Finck, 2019), limiting visibility into the global architecture of policy discourse. This fragmentation persists even as studies on decentralized network governance propose models of regulated self-regulation (Zwitter \u0026amp; Hazenberg, \u003cspan citationid=\"CR83\" class=\"CitationRef\"\u003e2020\u003c/span\u003e) and public-sector research begins to consolidate actor\u0026ndash;governance frameworks and implementation guidance (Garrod \u0026amp; Samuel, \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2021\u003c/span\u003e; Tan et al., \u003cspan citationid=\"CR71\" class=\"CitationRef\"\u003e2022\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eSecond, the evolution of policy themes has not been systematically documented. Blockchain\u0026rsquo;s policy salience has shifted from cryptographic innovation to financial regulation, public-sector modernization, data governance, and systemic-risk management (Hileman \u0026amp; Rauchs, \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Campbell-Verduyn, 2017; Hughes et al., \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Closely related fintech and regtech trajectories\u0026mdash;covering ecosystem change, supervisory technology, and new regulatory design\u0026mdash;suggest additional pathways through which blockchain becomes policy-salient, but these have rarely been integrated into longitudinal discourse mapping (Lee \u0026amp; Shin, \u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Anagnostopoulos, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Iman, \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2020\u003c/span\u003e; Zetzsche et al., \u003cspan citationid=\"CR82\" class=\"CitationRef\"\u003e2017\u003c/span\u003e). Moreover, although technology-development studies document blockchain\u0026rsquo;s expanding innovation base (Xu et al., \u003cspan citationid=\"CR79\" class=\"CitationRef\"\u003e2019\u003c/span\u003e) and organizational adoption rationales (Angelis \u0026amp; da Silva, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2019\u003c/span\u003e), the timing, magnitude, and stability of discourse transitions have not been mapped across a multi-decade global corpus.\u003c/p\u003e \u003cp\u003eThird, semantic and evaluative framing remains underexplored. While prior work identifies divergent narratives surrounding blockchain\u0026mdash;innovation enabler, transparency mechanism, surveillance risk, systemic-risk factor\u0026mdash;empirical evidence on framing variation across institutional sectors and world regions is extremely limited (Ceron \u0026amp; Negri, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Li et al., \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e2023\u003c/span\u003e; Wang et al., 2023). Embedding-based methods now make this analysis feasible, but they have not been applied to blockchain policy at scale, despite accumulating foundations in public-sector blockchain governance and systematic syntheses of blockchain\u0026rsquo;s role in public services (El-Haddadeh et al., 2021; Berryhill et al., \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2018\u003c/span\u003e).\u003c/p\u003e \u003cp\u003ePrior literature offers conceptual clarity about blockchain\u0026rsquo;s institutional and governance characteristics and provides robust methodological tools for analyzing large policy corpora, but it lacks systematic, multi-actor, longitudinal evidence on how blockchain is framed, interpreted, and prioritized across the global policy landscape. This study addresses these gaps by constructing a large-scale corpus of blockchain policy documents and examining its structural, thematic, and semantic dimensions over nearly three decades.\u003c/p\u003e \u003c/div\u003e"},{"header":"3. Data and methods","content":"\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003e3.1. Two-stage analytical design\u003c/h2\u003e \u003cp\u003ePolicy documents record institutional choices and the policy logics that justify them, making them primary materials for measuring policy meaning at scale. Text-as-data approaches therefore offer a principled route to recover latent positions, frames, and implied governance priorities from large collections of political and policy texts, while requiring explicit attention to assumptions and validation (Grimmer \u0026amp; Stewart, \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; Grimmer et al., \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Ash \u0026amp; Hansen, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThis study uses a dual-corpus, two-stage analytical design that separates broad field mapping from full-text semantic and evaluative inference. Stage 1 constructs a wide-coverage metadata corpus from Overton to characterize the emergence of the blockchain policy field, its institutional composition, and the distribution of Overton topical descriptors across historically grounded periods and regions. Stage 2 then draws a smaller full-text analytic corpus consisting of policy documents whose PDF text can be reliably extracted and cleaned for modeling. Using this inference corpus, the analysis (i) represents policy meaning with document embeddings (Doc2Vec, building on distributed representations; Mikolov et al., \u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; Le \u0026amp; Mikolov, \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e2014\u003c/span\u003e), (ii) estimates evaluative risk framing using a transparent lexicon (FinBERT is used only as a diagnostic robustness check; Devlin et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Araci, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2019\u003c/span\u003e), and (iii) assesses convergence and divergence across regions and institutional source types via centroid-based cosine similarity. This separation aligns coverage and interpretability by preserving breadth for field mapping while prioritizing text quality for semantic and stance inference.\u003c/p\u003e \u003cp\u003eStage 2 covers 2014\u0026ndash;2022 (N\u0026thinsp;=\u0026thinsp;393) and supports embedding-based semantic mapping and lexicon-based evaluative risk framing; a transformer sentiment model (FinBERT) is computed only as a diagnostic robustness check (Devlin et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Araci, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2019\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eTo make the coverage\u0026ndash;inference tradeoff transparent, I report the sample construction in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e and summarize the end-to-end analytical pipeline in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eSample construction\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"2\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMetric\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eValue\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStage 1 metadata corpus (N)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e17019\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStage 1 years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1994\u0026ndash;2022\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStage 2 full-text corpus (N)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e393\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eStage 2 full-text years\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2014\u0026ndash;2022\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e3.2. Stage1. Field mapping and scaffolding\u003c/h2\u003e \u003cdiv id=\"Sec9\" class=\"Section3\"\u003e \u003ch2\u003e3.2.1 Data sources and metadata corpus construction\u003c/h2\u003e \u003cp\u003eThe dataset is built from Overton, an international database of policy and government documents. A dual-corpus structure is used to align field coverage with text quality. The metadata corpus includes all documents tagged with \u0026ldquo;blockchain\u0026rdquo; or assigned blockchain-related topical descriptors between 1994 and 2022 September 9th, retaining publication year, source country, institutional source type, and Overton topic labels. This corpus supports long-run field mapping and participation patterns in the policy space.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section3\"\u003e \u003ch2\u003e3.2.2 Temporal harmonization and historically grounded periodization\u003c/h2\u003e \u003cp\u003ePublication years are validated and harmonized into numeric form. Because blockchain policy evolved through punctuated shifts associated with widely recognized technological and regulatory transitions, time is modeled as historically staged rather than purely continuous. Documents are assigned to four periods: Early Cryptography and Trust (1994\u0026ndash;2010), Financialization and ICO Emergence (2011\u0026ndash;2017), Regulatory Consolidation (2018\u0026ndash;2020), and Digital Government Integration (2021\u0026ndash;2022). Periodization provides the temporal scaffold needed to interpret topic shifts, semantic movement, and changes in evaluative language as transformations in policy meaning rather than as noise in annual production (Schmidt, \u003cspan citationid=\"CR65\" class=\"CitationRef\"\u003e2008\u003c/span\u003e). The period boundaries are additionally motivated by governance scholarship emphasizing transitions in the architecture of trust, the rise of code-mediated rule systems, and the intensification of regulatory contestation around blockchain (Werbach, \u003cspan citationid=\"CR77\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; De Filippi \u0026amp; Wright, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Yeung, \u003cspan citationid=\"CR81\" class=\"CitationRef\"\u003e2019\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section3\"\u003e \u003ch2\u003e3.2.3 Regional and institutional classification\u003c/h2\u003e \u003cp\u003eTo compare governance logics across contexts, documents are grouped by macro-region and institutional source type. Countries are aggregated into US/CAN, EU/UK, Asia (China, South Korea, Japan, Singapore, India), IGOs, and Global South/Other. Institutional types follow Overton\u0026rsquo;s source-type labels (e.g., government, IGO, think tanks). This classification operationalizes the study\u0026rsquo;s core comparative claim: that policy meaning is produced within institutional settings and is therefore expected to differ across regulatory traditions and governance models rather than converge mechanically across jurisdictions (\u0026Oslash;lnes et al., \u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Campbell-Verduyn, 2017). It also supports comparative assessment of how trust and regulatory philosophies are articulated across contexts in blockchain governance (Werbach, \u003cspan citationid=\"CR77\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Yeung, \u003cspan citationid=\"CR81\" class=\"CitationRef\"\u003e2019\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section3\"\u003e \u003ch2\u003e3.2.4 Archetype operationalization: keyword rules and anchor concepts\u003c/h2\u003e \u003cp\u003eTo aid interpretation, Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e reports the operational definitions and representative keyword anchors for each governance archetype. Additional implementation details and extended outputs are provided in the Supplementary Information.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eArchetype operationalization\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eArchetype\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGovernance logic\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eRepresentative keywords (illustrative)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAnchor terms (seed concepts)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRegulatory Control\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGovernance through legal and compliance mechanisms\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ecompliance; regulation; licensing; enforcement; AML/KYC; sanctions\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003ecompliance; security; securities; AML/KYC\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDigital Identity \u0026amp; Infrastructure\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGovernance through public digital infrastructure and identity systems\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003edigital identity; authentication; credentials; registry; infrastructure; public services\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eidentity; trust; registry; public services\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCoordination \u0026amp; Standards\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGovernance through coordination, interoperability, and international standards\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003einteroperability; standards; cross-border; harmonization; international cooperation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003estandards; interoperability; cross-border; ISO\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUnclassified\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNo dominant governance logic detected\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"4\"\u003eNotes:\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eThis table reports archetype definitions with representative keywords and anchor concepts for transparency.\u003c/p\u003e \u003cp\u003eThe full regex rule set and complete anchor lists used for classification are available upon request.\u003c/p\u003e \u003cp\u003eDocuments with no keyword match and low semantic proximity to all anchor sets are labeled Unclassified.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003e3.3 Stage2. Semantic and evaluative inference\u003c/h2\u003e \u003cp\u003eAccordingly, Stage 2 prioritizes extractability and linguistic stability, and all semantic, stance, and alignment estimates are computed exclusively on this inference corpus, while Stage 1 indicators are computed on the full metadata corpus.\u003c/p\u003e \u003cdiv id=\"Sec14\" class=\"Section3\"\u003e \u003ch2\u003e3.3.1 Inference corpus and preprocessing (semantic structure \u0026rarr; evaluative stance \u0026rarr; alignment)\u003c/h2\u003e \u003cp\u003eA full-text corpus is derived from the metadata corpus by retaining only those documents whose PDF text can be extracted without structural corruption. Extracted text is cleaned to remove encoding artefacts and linked back to the metadata; documents with unusable or extremely short text are excluded. Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e summarizes the Stage 2 inclusion and exclusion rules that operationalize this design choice, yielding a corpus (393 documents) that is tailored for inference tasks requiring stable linguistic inputs rather than broad archival coverage. This construction reflects the trade-off commonly encountered in large-scale policy repositories (Szomszor \u0026amp; Adie, \u003cspan citationid=\"CR68\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). The approach also aligns with best practice in policy-text analytics, where the credibility of semantic and stance estimates depends on prioritizing extractability and interpretability over maximal inclusion (Lesnikowski et al., \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Ma et al., \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e \u003cp\u003ePreprocessing is intentionally conservative. Full texts are standardized through lowercasing, removal of disruptive formatting artefacts introduced by PDF extraction, whitespace normalization, and minimum-length filtering. As indicated in Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e, these quality checks are applied uniformly to reduce artefactual meaning and sentiment estimates driven by OCR noise, boilerplate, or layout corruption. This yields inputs that are comparable across heterogeneous document templates and minimizes the risk that model outputs are driven by layout noise rather than policy framing and evaluative language (Kozlowski et al., \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Emphasizing format heterogeneity and analytic cleanliness is consistent with established guidance from adaptation and policy text-analytics research, which repeatedly notes that extraction artefacts can otherwise dominate measured thematic and evaluative signals (Araos et al., \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2016\u003c/span\u003e; Lesnikowski et al., \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Ma et al., \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eStage 2 inclusion and exclusion rules for the inference corpus (2014\u0026ndash;2022; N\u0026thinsp;=\u0026thinsp;393).\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRule type\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCriterion\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eRationale / design purpose\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eNotes (where applied)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInclusion\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eIn-scope blockchain-related policy document in the Stage 1 Overton metadata corpus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eFixes topical scope and preserves field-wide coverage for representativeness diagnostics\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eStage 1 \u0026rarr; Stage 2 linkage\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInclusion\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eFull-text PDF retrievable and machine-extractable (no structural corruption)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eEnsures stable linguistic inputs for embeddings and stance inference\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eStage 2 construction\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInclusion\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eText passes conservative preprocessing quality checks (e.g., minimum-length and noise filters)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eReduces artefactual sentiment/meaning estimates driven by OCR noise or boilerplate\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eStage 2 preprocessing\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eExclusion\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eImage-only / scanned PDFs or extraction failures\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAvoids unreliable semantic and sentiment inference from corrupted inputs\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eDropped at extraction step\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eExclusion\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDuplicates or near-duplicates (mirrored reports, re-hosted PDFs, minor versioning)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePrevents overweighting a single document family in time-series and clustering summaries\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eDe-duplication prior to inference\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eExclusion\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eExtremely short or non-substantive texts after extraction (e.g., stubs, announcements)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eInsufficient semantic signal for archetype assignment and model-based stance estimation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eDropped by minimum-length rule\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFlag (robustness)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDAO/DeFi-labeled documents\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePotentially distinct protocol-governance discourse; retained in baseline but excluded in a robustness replication\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eSee Section 4.5 and Appendix A3\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section3\"\u003e \u003ch2\u003e3.3.2 Inference Module I: Semantic structure via document embeddings\u003c/h2\u003e \u003cp\u003eThe first inference module recovers the latent semantic structure of policy discourse using document embeddings. I train a Doc2Vec (paragraph vector) model on the cleaned full-text corpus to represent each document as a fixed-length vector; Doc2Vec extends the distributed-representation logic of Word2Vec to sentences and documents (Mikolov et al., \u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; Le \u0026amp; Mikolov, \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e2014\u003c/span\u003e). These vectors encode differences in problem framing, governance priorities, and regulatory orientation that are often expressed through contextual phrasing rather than explicit keywords. Embedding-based representations are increasingly used to map meaning structures and discursive proximity among actors (Kozlowski et al., \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2019\u003c/span\u003e), but I interpret them cautiously given conceptual and measurement limits when embeddings are used as proxies for meaning and culture (Mandell, \u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). For visualization, I project document vectors to two dimensions with t‑SNE (van der Maaten \u0026amp; Hinton, \u003cspan citationid=\"CR74\" class=\"CitationRef\"\u003e2008\u003c/span\u003e) to produce embedding maps, and I compute group-level semantic centroids to quantify convergence and divergence across regions and institutional types.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec16\" class=\"Section3\"\u003e \u003ch2\u003e3.3.3 Inference Module II: Evaluative framing via a transparent risk lexicon (with transformer diagnostic)\u003c/h2\u003e \u003cp\u003eTo characterize evaluative framing in blockchain policy discourse, I operationalize a transparent risk-framing index based on a domain-specific lexicon of risk and compliance terms (see Supplementary Note S4). The index counts lexicon hits per 1,000 tokens in the cleaned full text, supporting direct interpretability and sensitivity checks to alternative dictionaries. As a diagnostic, I also apply FinBERT, a transformer model pretrained on financial text, to the same documents (Araci, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). In this corpus, however, the resulting FinBERT-based signed sentiment series is constant (0), consistent with domain mismatch between policy-legal governance language and financial sentiment benchmarks. Accordingly, the main analyses use the lexicon-based risk-framing index, while FinBERT is retained only as a diagnostic robustness check. This approach follows established text-based risk and uncertainty measurement traditions\u0026mdash;from policy-uncertainty indices (Baker et al., \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2016\u003c/span\u003e) to domain-specific risk dictionaries (Loughran \u0026amp; McDonald, \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e2011\u003c/span\u003e)\u0026mdash;and aligns with embedding-enabled political risk measurement that captures contextual risk beyond raw counts (Hassan et al., \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). For external benchmarking, I compare the resulting series to a widely used news-based geopolitical risk index (Caldara \u0026amp; Iacoviello, \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2022\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eFor transparency, I convert FinBERT labels (Devlin et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Araci, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2019\u003c/span\u003e) into a signed score s_d by mapping the positive and negative classes to their posterior probabilities with opposite signs and assigning 0 to the neutral class. Because the resulting series is constant (0) in this dataset, it is not used as an outcome variable in the main text; it is reported only as a diagnostic in Supplementary Note S4.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec17\" class=\"Section3\"\u003e \u003ch2\u003e3.3.4 Inference Module III: Institutional and regional semantic proximity (convergence/divergence)\u003c/h2\u003e \u003cp\u003eTo quantify convergence and divergence in policy meaning, group-level semantic centroids are computed by averaging document embeddings within each region and each institutional type. Pairwise cosine similarity between group centroids yields similarity matrices that summarize whether the discourse field exhibits shared semantic \u0026ldquo;centers of gravity\u0026rdquo; or persistent separation across governance contexts. This aligns with prior work on institutional differentiation in technology governance, where similar technologies can be stabilized under distinct policy imaginaries and regulatory philosophies (Werbach, \u003cspan citationid=\"CR77\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; De Filippi \u0026amp; Wright, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Yeung, \u003cspan citationid=\"CR81\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). In interpretive terms, proximity patterns are read as evidence about whether the architecture of trust and the code\u0026ndash;law interface are framed similarly across actors, or whether they remain regionally and institutionally segmented (Werbach, \u003cspan citationid=\"CR77\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; De Filippi \u0026amp; Wright, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Yeung, \u003cspan citationid=\"CR81\" class=\"CitationRef\"\u003e2019\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec18\" class=\"Section2\"\u003e \u003ch2\u003e3.4 Aggregation and reporting strategy\u003c/h2\u003e \u003cp\u003eResults are reported using aggregation choices that preserve interpretability while aligning with the dual-corpus design. Topic prevalence is summarized using relative frequencies derived from the reconstructed document\u0026ndash;topic matrix; sentiment is summarized using arithmetic means of the continuous index; semantic alignment is summarized with cosine similarity among group-level centroids. Where document volumes differ substantially across groups, reporting emphasizes structural patterns and robustness-consistent rank ordering rather than raw magnitude comparisons, consistent with best practice in large-scale policy-document analyses (Szomszor \u0026amp; Adie, \u003cspan citationid=\"CR68\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). This reporting strategy also mirrors cross-domain policy analytics where thematic mapping and evaluative inference are interpreted primarily as comparative signals rather than exact measures of \u0026ldquo;true\u0026rdquo; sentiment or meaning (Lesnikowski et al., \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Sucupira \u0026amp; de Albuquerque, \u003cspan citationid=\"CR67\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Mandell, \u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e2022\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eIn addition to reporting raw aggregates, I attach uncertainty or contrast-based evidence to the two comparisons that underpin RQ2 and RQ3. For semantic proximity, centroid similarities are complemented with bootstrap confidence intervals and selected difference tests; for evaluative tone, group contrasts are supported by regression models that condition on archetype and issuer composition. Finally, to translate mappings into a coordination metric, I construct a Coordination Friction Index for each region pair and time window that combines (i) semantic distance (1 - cosine similarity), (ii) divergence in archetype composition, and (iii) differences in evaluative tone. This index is non-causal but decision-oriented: it is designed to identify where mutual recognition and standard-setting are likely to face the highest semantic and institutional alignment costs.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec19\" class=\"Section2\"\u003e \u003ch2\u003e3.5 Robustness and validation\u003c/h2\u003e \u003cp\u003eRobustness checks evaluate whether the main conclusions are sensitive to reasonable perturbations in corpus construction and modeling choices. First, extraction-quality and minimum-length thresholds are varied to ensure that results are not driven by a small number of unusually short or noisy documents. Second, embedding specifications are perturbed within defensible ranges and similarity matrices are compared for rank-order stability, while interpretation remains attentive to known limits in embedding-based meaning inference (Mandell, \u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Kozlowski et al., \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). Third, sentiment results are re-estimated under alternative truncation rules and score constructions to confirm that regional differences are not artefacts of document length or input constraints (Araci, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2019\u003c/span\u003e; Sucupira \u0026amp; de Albuquerque, \u003cspan citationid=\"CR67\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Finally, period boundary sensitivity is assessed by shifting cut points around major transitions to verify that observed turning points and regional contrasts persist under plausible alternative temporal partitions, consistent with historically grounded periodization in governance and regulatory-change studies (Schmidt, \u003cspan citationid=\"CR65\" class=\"CitationRef\"\u003e2008\u003c/span\u003e; Werbach, \u003cspan citationid=\"CR77\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Yeung, \u003cspan citationid=\"CR81\" class=\"CitationRef\"\u003e2019\u003c/span\u003e).\u003c/p\u003e \u003cdiv id=\"Sec20\" class=\"Section3\"\u003e \u003ch2\u003e3.5.1 Validating archetype assignment (keyword rules).\u003c/h2\u003e \u003cp\u003eArchetype labels are validated by (i) conducting face-validity audits on stratified random samples across regions and years to confirm that rule hits map to the intended anchor concepts, (ii) perturbing keyword lists and minimum-hit thresholds to verify that dominant-archetype shares and turning points are stable, and (iii) checking that substantive conclusions persist under the representativeness reweighting and the DAO/DeFi exclusion exercise reported later. These checks treat the archetypes as transparent measurement rules and evaluate whether substantive patterns survive reasonable rule variations rather than relying on a single fixed dictionary.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec21\" class=\"Section3\"\u003e \u003ch2\u003e3.5.2 Validating embedding-based semantic structure.\u003c/h2\u003e \u003cp\u003eThe embedding module is validated by (i) testing rank-order stability of region-by-region cosine similarity matrices under defensible perturbations to embedding specifications and preprocessing, (ii) verifying that the main cluster contrasts are not artefacts of a particular projection method by comparing 2D visualizations to the underlying high-dimensional similarity structure, and (iii) bootstrapping centroid estimates to confirm that cross-regional distance patterns are not driven by a small number of influential documents. Interpretation remains comparative and structure-focused, consistent with known limits of embedding-based meaning inference. Following best-practice guidance on measurement validity in automated text analysis (Grimmer \u0026amp; Stewart, \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2013\u003c/span\u003e), I treat embedding-based meaning measures as potentially noisy constructs and therefore report sensitivity checks for preprocessing and hyperparameter choices (Denny \u0026amp; Spirling, \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e2018\u003c/span\u003e) and for embedding stability and interpretability in applied research (Rodriguez \u0026amp; Spirling, \u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). I further interpret distances and projections in embedding space as measurable shifts in cultural and policy meaning in line with the geometry-of-meaning framework (Kozlowski et al., \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2019\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec22\" class=\"Section3\"\u003e \u003ch2\u003e3.5.3 Validating evaluative framing (risk lexicon) and transformer diagnostic.\u003c/h2\u003e \u003cp\u003eEvaluative framing is validated in three ways. First, I re-estimate the risk-framing index using alternative lexicon variants (narrow vs. broad dictionaries and exclusions of highly polysemous terms) to verify that cross-region contrasts are not an artifact of a single word list. Second, I replicate key patterns after excluding DAO/DeFi-labeled documents and other protocol-governance subcorpora to ensure that regional contrasts are not mechanically driven by a distinct discourse segment. Third, I compute FinBERT-based signed sentiment as an external diagnostic (Araci, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2019\u003c/span\u003e); in this corpus the score is constant (0), reinforcing the decision to rely on transparent lexicon-based framing rather than interpret transformer sentiment as psychological positivity/negativity. Accordingly, I emphasize relative cross-group differences in risk framing and interpret them as systematic variation in governance logics.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e"},{"header":"4. Results","content":"\u003cp\u003eThis section reports empirical patterns from the dual-corpus design as answers to RQ1-RQ3. Stage 1 documents long-run field participation and topic scaffolding (1994\u0026ndash;2022) (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Stage 2 (2014\u0026ndash;2022) supports full-text inference of governance archetypes, semantic proximity, and evaluative tone. Where possible, key comparisons are accompanied by uncertainty diagnostics (bootstrap confidence intervals or permutation tests) to convert visual contrasts into interpretable evidence.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cdiv id=\"Sec24\" class=\"Section2\"\u003e \u003ch2\u003e4.1 RQ1: How does the functional center of blockchain governance discourse shift over time?\u003c/h2\u003e \u003cp\u003eThe governance-archetype measurement reveals a long-run transition in what policy documents treat blockchain as 'for'. In the early period, blockchain is frequently framed through regulatory control and financial integrity, consistent with a problem definition centered on compliance, fraud, and market stability. Over time, the center of gravity moves toward coordination/standard-setting and, later, toward identity and public-infrastructure integration. Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e summarizes archetype shares by period and region, and Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e traces annual evolution in the inference corpus (2016\u0026ndash;2022 shown).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eTo test whether these shifts are systematic rather than visual artefacts, Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e estimates multinomial (or fractional-logit) models in which archetype prevalence is explained by flexible time trends (year splines) and context (region and institutional source type). The estimated time profiles confirm that the governance center of gravity moves away from pure regulatory control toward coordination and infrastructure frames in the later period, even after conditioning on region and issuer composition.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eArchetype shift models (time trends, region, and issuer composition)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePost-period contrast\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCoordination \u0026amp; standards\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eIdentity \u0026amp; infrastructure\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eRegulatory control\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2020\u0026ndash;2022 vs 2017\u0026ndash;2019\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2.900*** [1.430, 5.880]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.260** [0.070, 0.920]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.59 [0.32, 1.11]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"4\"\u003eNotes: Odds ratios with 95% confidence intervals. The post-period contrast is 2020\u0026ndash;2022 vs 2017\u0026ndash;2019 (A option).\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec25\" class=\"Section2\"\u003e \u003ch2\u003e4.2 RQ2: Where do coordination frictions concentrate?\u003c/h2\u003e \u003cp\u003eSemantic proximity patterns indicate that blockchain policy meanings do not converge into a single global center, even when archetype anchors are shared. Figure\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e visualizes the embedding space and highlights clustering by region and institutional source type. Table\u0026nbsp;\u003cspan refid=\"Tab6\" class=\"InternalRef\"\u003e6\u003c/span\u003e reports centroid-based cosine similarity across regions, which can be interpreted as a coarse measure of how costly translation and mutual recognition may be for cross-border coordination.\u003c/p\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e5\u003c/span\u003e and Table\u0026nbsp;\u003cspan refid=\"Tab6\" class=\"InternalRef\"\u003e6\u003c/span\u003e indicate that cross-region semantic alignment tightened in 2020\u0026ndash;2022 for some pairs, while others diverged, foreshadowing where policy coordination costs are likely to be highest.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab5\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eRegional semantic similarity (2017\u0026ndash;2019; cosine similarity between region centroids)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRegion\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eEU/UK\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eUS/CAN\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAsia\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eIGO\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eGlobal South/Other\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEU/UK\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.600\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.830\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.830\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUS/CAN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.600\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.660\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.610\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAsia\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.830\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.660\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.660\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIGO\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGlobal South/Other\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.830\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.610\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.660\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"6\"\u003eNotes: Entries are cosine similarities between region-level centroids in the TF\u0026ndash;IDF\u0026thinsp;+\u0026thinsp;SVD semantic space (higher\u0026thinsp;=\u0026thinsp;more semantically aligned).\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eTo quantify uncertainty and enable sharp comparisons (e.g., whether EU-Asia similarity is lower than EU-US similarity), Supplementary Table S4 reports bootstrap confidence intervals for centroid similarities, and Supplementary Table S5 reports selected difference contrasts. These uncertainty-aware comparisons support interpretation of coordination frictions without requiring causal identification.\u003c/p\u003e \u003cp\u003eUncertainty for the similarity estimates and selected pairwise contrasts are reported in Supplementary Tables S4\u0026ndash;S5.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab6\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 6\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eRegional semantic similarity (2020\u0026ndash;2022; cosine similarity between region centroids)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"6\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRegion\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eEU/UK\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eUS/CAN\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAsia\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eIGO\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eGlobal South/Other\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEU/UK\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.660\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.560\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.570\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.540\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUS/CAN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.660\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.540\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.680\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.260\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAsia\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.560\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.540\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.470\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.240\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIGO\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.570\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.680\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.470\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.240\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGlobal South/Other\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.540\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.260\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.240\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.240\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e1.000\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"6\"\u003eNotes: Same embedding pipeline as Table\u0026nbsp;\u003cspan refid=\"Tab6\" class=\"InternalRef\"\u003e6\u003c/span\u003eA; differences reflect period-specific document sets.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eBuilding on these similarities, Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e reports a Coordination Friction Heatmap for region pairs and time windows. Higher friction indicates greater expected translation and alignment costs (semantic distance, archetype divergence, and risk gaps). The heatmap is intended as a decision aid for standard-setting and mutual-recognition efforts: it identifies where coordination is most likely to stall because jurisdictions are discussing the same technology in functionally different ways.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec26\" class=\"Section2\"\u003e \u003ch2\u003e4.3 RQ3: Is risk sensitivity a region effect or a framing effect?\u003c/h2\u003e \u003cp\u003eEvaluative tone varies systematically across contexts. Table\u0026nbsp;\u003cspan refid=\"Tab7\" class=\"InternalRef\"\u003e7\u003c/span\u003e reports average lexicon-based risk index (per 1,000 tokens) by region and archetype, showing that some regions maintain more cautious, risk-sensitive framing even when discussing similar governance functions. This pattern is consistent with the idea that institutions differ not only in what they prioritize (archetypes) but also in how they normatively evaluate opportunities and risks.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab7\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 7\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eEvaluative risk framing by region and archetype\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRegion\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCoordination \u0026amp; standards\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eIdentity \u0026amp; infrastructure\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eRegulatory control\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eUnclassified\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEU/UK\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2.81 (N\u0026thinsp;=\u0026thinsp;24)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e9.43 (N\u0026thinsp;=\u0026thinsp;4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5.78 (N\u0026thinsp;=\u0026thinsp;85)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2.64 (N\u0026thinsp;=\u0026thinsp;26)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eUS/CAN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e5.40 (N\u0026thinsp;=\u0026thinsp;4)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.96 (N\u0026thinsp;=\u0026thinsp;12)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e9.85 (N\u0026thinsp;=\u0026thinsp;104)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2.46 (N\u0026thinsp;=\u0026thinsp;32)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAsia\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e4.50 (N\u0026thinsp;=\u0026thinsp;8)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3.00 (N\u0026thinsp;=\u0026thinsp;6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5.31 (N\u0026thinsp;=\u0026thinsp;6)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIGO\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1.77 (N\u0026thinsp;=\u0026thinsp;3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e7.94 (N\u0026thinsp;=\u0026thinsp;17)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGlobal South/Other\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e2.41 (N\u0026thinsp;=\u0026thinsp;13)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e4.04 (N\u0026thinsp;=\u0026thinsp;3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2.79 (N\u0026thinsp;=\u0026thinsp;30)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2.39 (N\u0026thinsp;=\u0026thinsp;18)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"5\"\u003eNotes: Lexicon-based risk index per 1,000 tokens. FinBERT sentiment is constant in this corpus (0), so I use a transparent risk lexicon for evaluative framing.\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eTo distinguish region effects from framing and issuer effects, Supplementary Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e regresses document-level risk index on region indicators, archetype indicators, and institutional source-type controls (with year fixed effects or flexible trends as appropriate). The results show whether regional caution/optimism persists net of governance frame composition, thereby clarifying whether region proxies for deeper differences in how governance problems are framed or simply reflects different mixes of document types.\u003c/p\u003e \u003cp\u003eNotes: Linear model with year fixed effects. Coefficients reported with robust standard errors in parentheses where available. Alternative specifications and full outputs are reported in Supplementary Note S3 (Supplementary Tables S6\u0026ndash;S7).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec27\" class=\"Section2\"\u003e \u003ch2\u003e4.4 Validity and robustness\u003c/h2\u003e \u003cp\u003eValidity checks for the two-stage design are summarized in Table\u0026nbsp;\u003cspan refid=\"Tab9\" class=\"InternalRef\"\u003e8\u003c/span\u003eA, and Table\u0026nbsp;\u003cspan refid=\"Tab9\" class=\"InternalRef\"\u003e8\u003c/span\u003eB reports robustness checks for key descriptive patterns.\u003c/p\u003e \u003cp\u003eSupplementary Information (Supplementary Notes S1\u0026ndash;S4 and Supplementary Tables S2\u0026ndash;S9) reports extended model outputs, bootstrap uncertainty quantification, and additional robustness checks that support the stability of the main temporal and cross-regional patterns.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab8\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 8\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eA. Validity checks (summary)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCheck\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eEvidence\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eImplication\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eExtractability constraint\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eStage 2 limited to stable, extractable full text (PDFs); rules documented.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eReduces noise in semantic mapping and framing measurement.\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRepresentativeness (raking)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eStage 2 reweighted to Stage 1 margins (Appendix A4).\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLong-run patterns are stable under reweighting.\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eClassification validation\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eArchetype anchors and validation diagnostics (Appendix A6).\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eArchetype assignment is not driven by chance.\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab9\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 8\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eB. Robustness checks (summary)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRobustness check\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eWhat changes\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eResult\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eExclude DAO/DeFi-labelled documents\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eRecompute archetype patterns without DAO/DeFi subset\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCore transitions remain directionally stable.\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eBootstrap uncertainty\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eResample documents within region\u0026ndash;period cells\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eKey semantic contrasts remain within the same qualitative range.\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFriction decomposition\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eReport semantic distance, archetype divergence, and risk-gap components\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHeatmap patterns are not driven by a single component.\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003ctfoot\u003e \u003ctr\u003e\u003ctd colspan=\"3\"\u003eNotes: These checks are designed to be non-causal but decision-relevant, aligning with the mapping goal. Detailed diagnostics and extended outputs are reported in Supplementary Notes S1\u0026ndash;S4 (Supplementary Tables S2\u0026ndash;S9).\u003c/td\u003e\u003c/tr\u003e \u003c/tfoot\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"5. Discussion","content":"\u003cp\u003eComputational mapping can easily read as mere visualization unless it is tied to decision-relevant comparisons and tests. I therefore structure the findings around three research questions, each paired with a minimal inferential check and a practical coordination output.\u003c/p\u003e \u003cp\u003eThe central policy implication is that cross-border coordination problems are often semantic before they are technical. When jurisdictions attach different governance purposes to blockchain (e.g., financial integrity, national identity infrastructure, or administrative modernization), attempts at wholesale harmonization are likely to stall. The Coordination Friction Index makes this logic actionable by identifying where semantic distance and frame divergence are highest.\u003c/p\u003e \u003cp\u003eThese coordination frictions are not reducible to vocabulary differences. Work on international standards and transnational regulation shows that alignment failures often reflect clashes between private authority, state regulation, and market power in the production of global rules (Mattli \u0026amp; B\u0026uuml;the, \u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e2003\u003c/span\u003e; B\u0026uuml;the \u0026amp; Mattli, \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2011\u003c/span\u003e; Abbott \u0026amp; Snidal, \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2009\u003c/span\u003e; Drezner, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2007\u003c/span\u003e). In networked economies, regulatory choices can be amplified through control of infrastructural nodes and interdependence, making semantic divergence a channel of geopolitical leverage rather than mere misunderstanding (Farrell \u0026amp; Newman, \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2019\u003c/span\u003e). The CFI operationalizes this institutional-logics dimension by treating cross-region semantic distance as an early warning signal of where harmonization, mutual recognition, or interoperability arrangements will face persistent friction.\u003c/p\u003e \u003cp\u003eCFI can also be read as a diagnostic of friction at the science\u0026ndash;policy interface: when documents translate technical uncertainty into policy-relevant categories, disagreements emerge over what counts as credible evidence, salient risk, and legitimate authority (Cash et al., \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2003\u003c/span\u003e; Jasanoff, \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2004\u003c/span\u003e; Pielke, \u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e2007\u003c/span\u003e). By quantifying systematic mismatches in evaluative framing and governance purpose, the index complements qualitative accounts of co-production and boundary work, which show that policy meaning is produced through iterative negotiation rather than linear knowledge transfer (Maas, \u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e2022\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eFor practitioners, the results imply three concrete coordination artifacts. First, a Coordination Heatmap ranks region pairs and time windows by expected alignment costs; this can be used to prioritize bilateral working groups, mutual-recognition pilots, or translation efforts. Second, a layered coordination playbook sequences negotiation in the order that minimizes deadlock: start with minimum viable alignment on consumer protection and anti-money-laundering interfaces, then move to interoperability and standards, and only then to deeper identity and infrastructure integration. Third, a translation table documents region- and issuer-specific vocabulary for similar governance functions (e.g., how 'identity' is institutionalized as KYC, digital ID, or infrastructure security), reducing the risk that coordination fails because equivalent functions are labeled differently.\u003c/p\u003e \u003cp\u003eMethodologically, the dual-corpus design addresses a practical reality of big policy repositories: coverage and extractability do not coincide. Rather than treating missing full text as nuisance, the paper treats extractability as a measurement constraint that can be audited and partially corrected. This design is generalizable to other emerging technologies where policy attention arrives faster than stable document formats and where the governance field spans heterogeneous institutions.\u003c/p\u003e \u003cp\u003eLimitations follow from the same design choices. Stage 2 inference is constrained to extractable PDFs, and embedding and framing outputs remain proxies that must be interpreted comparatively rather than as direct measures of latent intent. Future work can extend the approach by (i) integrating additional languages with multilingual models, (ii) linking discourse shifts to measurable regulatory outcomes (e.g., rule adoption, enforcement intensity), and (iii) tracing how policy discourse interacts with media and market narratives in real time.\u003c/p\u003e"},{"header":"6. Conclusion","content":"\u003cp\u003eBy combining broad metadata coverage with full-text inference, this study maps how blockchain is redefined in policy discourse from a crypto-finance experiment into a contested digital-governance infrastructure. The evidence shows (i) a measurable shift in governance archetypes over time, (ii) persistent regional clustering in semantic meaning that signals where coordination frictions are likely to be highest, and (iii) systematic differences in evaluative tone that cannot be reduced to archetype composition alone. Beyond describing the field, the paper offers decision-oriented outputs - a friction index, a coordination heatmap, and a translation table - that help policymakers anticipate where standard-setting and mutual recognition are most likely to stall and how to stage coordination in practice.\u003c/p\u003e "},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eData availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study draws metadata and document identifiers from Overton under license and therefore cannot redistribute raw Overton exports or source PDFs. To support transparency and reuse, the replication package includes (i) the list of Overton document IDs used to construct the two-stage corpus, (ii) all derived data tables used to generate figures and tables (e.g., archetype assignments, descriptor shares, semantic centroids, and CFI matrices), and (iii) scripts that reproduce the full processing and analysis pipeline for users with licensed access to Overton.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCode availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eUpon acceptance, all analysis code (Python) and configuration files necessary to reproduce the results will be made available as part of the replication package.\u003cbr\u003e\u0026nbsp;The data used in this study are subject to access restrictions and therefore cannot be publicly shared.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe author declares no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEthics approval and consent to participate\u003c/strong\u003e:\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThis study does not involve human participants or identifiable personal data. Ethical approval and consent were not required.\u003c/p\u003e\n\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eH.K. conceived and designed the study, curated the data, developed the methodology, performed the formal analysis, and wrote the analysis code. H.K. prepared all figures and tables and wrote the original draft of the manuscript. H.K. reviewed and edited the manuscript and takes responsibility for the integrity of the work as a whole.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eAbbott, K. W., \u0026amp; Snidal, D. (2009). Strengthening international regulation through transnational new governance: Overcoming the orchestration deficit. Vanderbilt Journal of Transnational Law, 42, 501\u0026ndash;578.\u003c/li\u003e\n\u003cli\u003eAllen, D. W. E., Berg, C., Lane, A. M., \u0026amp; Potts, J. (2019). Blockchains and the economic institutions of capitalism. Journal of Institutional Economics, 15(4), 569\u0026ndash;589.\u003c/li\u003e\n\u003cli\u003eAl‑Saqaf, W., \u0026amp; Seidler, N. (2017). Blockchain technology for social impact: Opportunities and challenges ahead. Journal of Cyber Policy, 2(3), 338\u0026ndash;354.\u003c/li\u003e\n\u003cli\u003eAnagnostopoulos, I. (2018). Fintech and regtech: Impact on regulators and banks. Journal of Economics and Business, 100, 7\u0026ndash;25.\u003c/li\u003e\n\u003cli\u003eAngelis, J., \u0026amp; da Silva, E. R. (2019). Blockchain adoption: A value driver perspective. Business Horizons, 62(3), 307\u0026ndash;314.\u003c/li\u003e\n\u003cli\u003eAraci, D. (2019). FinBERT: Financial sentiment analysis with pre-trained language models. arXiv preprint arXiv:1908.10063.\u003c/li\u003e\n\u003cli\u003eAraos, M., Berrang‑Ford, L., Ford, J. D., Austin, S. E., Biesbroek, R., \u0026amp; Lesnikowski, A. (2016). Climate change adaptation planning in large cities: A systematic global assessment. Urban Climate, 17, 176\u0026ndash;197.\u003c/li\u003e\n\u003cli\u003eArner, D. W., Barberis, J., \u0026amp; Buckley, R. P. (2017). FinTech, RegTech, and the reconceptualization of financial regulation. Northwestern Journal of International Law \u0026amp; Business, 37(3), 371\u0026ndash;414.\u003c/li\u003e\n\u003cli\u003eAsh, E., \u0026amp; Hansen, S. (2023). Text algorithms in economics. Annual Review of Economics, 15(1), 659\u0026ndash;688.\u003c/li\u003e\n\u003cli\u003eBaker, S. R., Bloom, N., \u0026amp; Davis, S. J. (2016). Measuring economic policy uncertainty. Quarterly Journal of Economics, 131(4), 1593\u0026ndash;1636.\u003c/li\u003e\n\u003cli\u003eBerrang‑Ford, L., Biesbroek, R., Ford, J. D., Lesnikowski, A., Tanabe, A., \u0026amp; Wang, F. M. (2021). A systematic global stocktake of evidence on human adaptation to climate change. Nature Climate Change, 11(11), 989\u0026ndash;1000.\u003c/li\u003e\n\u003cli\u003eBerryhill, J., Bourgery, T., \u0026amp; Hanson, A. (2018). Blockchains unchained: Blockchain technology and its use in the public sector. OECD Working Papers on Public Governance, No. 28.\u003c/li\u003e\n\u003cli\u003eBiesbroek, R., Dupuis, J., Jordan, A., Wellstead, A., Howlett, M., Cairney, P., \u0026hellip; Rayner, T. (2015). Opening up the black box of adaptation decision‑making. Nature Climate Change, 5(6), 493\u0026ndash;494.\u003c/li\u003e\n\u003cli\u003eBornmann, L., Haunschild, R., \u0026amp; Marx, W. (2022). How relevant is climate change research for climate change policy? An analysis of policy citations of climate change publications. PLOS ONE, 17(9), e0274693.\u003c/li\u003e\n\u003cli\u003eB\u0026uuml;the, T., \u0026amp; Mattli, W. (2011). The New Global Rulers: The Privatization of Regulation in the World Economy. Princeton University Press.\u003c/li\u003e\n\u003cli\u003eCaldara, D., \u0026amp; Iacoviello, M. (2022). Measuring geopolitical risk. American Economic Review, 112(4), 1194\u0026ndash;1225.\u003c/li\u003e\n\u003cli\u003eCampbell‑Verduyn, M. (Ed.). (2017). Bitcoin and beyond: Cryptocurrencies, blockchains, and global governance. Routledge.\u003c/li\u003e\n\u003cli\u003eCash, D. W., Clark, W. C., Alcock, F., Dickson, N. M., Eckley, N., \u0026amp; J\u0026auml;ger, J. (2003). Knowledge systems for sustainable development. Proceedings of the National Academy of Sciences, 100(14), 8086\u0026ndash;8091.\u003c/li\u003e\n\u003cli\u003eCatalini, C., \u0026amp; Gans, J. S. (2020). Some simple economics of the blockchain. Communications of the ACM, 63(7), 80\u0026ndash;90.\u003c/li\u003e\n\u003cli\u003eCeron, A., \u0026amp; Negri, F. (2016). Using sentiment analysis to support the action of policy‑makers: A case study on the Italian labour market reform. Policy \u0026amp; Internet, 8(3), 258\u0026ndash;278.\u003c/li\u003e\n\u003cli\u003eCong, L. W., \u0026amp; He, Z. (2019). Blockchain disruption and smart contracts. Review of Financial Studies, 32(5), 1754\u0026ndash;1797.\u003c/li\u003e\n\u003cli\u003eDe Filippi, P., \u0026amp; Wright, A. (2018). Blockchain and the law: The rule of code. Harvard University Press.\u003c/li\u003e\n\u003cli\u003eDenny, M., \u0026amp; Spirling, A. (2018). Text preprocessing for unsupervised learning: Why it matters, when it misleads, and what to do about it. Political Analysis, 26(2), 168\u0026ndash;189.\u003c/li\u003e\n\u003cli\u003eDevlin, J., Chang, M. W., Lee, K., \u0026amp; Toutanova, K. (2019). BERT: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of NAACL-HLT 2019 (pp. 4171\u0026ndash;4186).\u003c/li\u003e\n\u003cli\u003eDrezner, D. W. (2007). All Politics Is Global: Explaining International Regulatory Regimes. Princeton University Press.\u003c/li\u003e\n\u003cli\u003eEl‑Haddadeh, R., Osmani, M., Hindi, N., Fadlalla, A., \u0026amp; Darem, A. (2021). Examining the role of blockchain technology in improving public sector services: A systematic literature review. Information Systems Frontiers, 23(6), 1405\u0026ndash;1423.\u003c/li\u003e\n\u003cli\u003eFarrell, H., \u0026amp; Newman, A. L. (2019). Weaponized interdependence: How global economic networks shape state coercion. International Security, 44(1), 42\u0026ndash;79.\u003c/li\u003e\n\u003cli\u003eFinck, M. (2018). Blockchain regulation and governance in Europe. Cambridge University Press.\u003c/li\u003e\n\u003cli\u003eGarrod, B., \u0026amp; Samuel, J. (2021). Blockchain applications in government: A systematic review. Government Information Quarterly, 38(4), 101619.\u003c/li\u003e\n\u003cli\u003eGikay, A. A. (2018). Bitcoin and the blockchain as possible corporate governance tools. The Journal of Business, Entrepreneurship \u0026amp; the Law, 11(2), 281\u0026ndash;324.\u003c/li\u003e\n\u003cli\u003eGomber, P., Koch, J.‑A., \u0026amp; Siering, M. (2017). Digital finance and FinTech: Current research and future research directions. Journal of Business Economics, 87(5), 537\u0026ndash;580.\u003c/li\u003e\n\u003cli\u003eGrimmer, J., \u0026amp; Stewart, B. M. (2013). Text as data: The promise and pitfalls of automatic content analysis methods for political texts. Political Analysis, 21(3), 267\u0026ndash;297.\u003c/li\u003e\n\u003cli\u003eGrimmer, J., Roberts, M. E., \u0026amp; Stewart, B. M. (2022). Text as Data: A New Framework for Machine Learning and the Social Sciences. Princeton University Press.\u003c/li\u003e\n\u003cli\u003eHassan, T. A., Hollander, S., van Lent, L., \u0026amp; Schwedeler, M. (2019). Firm-level political risk: Measurement and effects. Quarterly Journal of Economics, 134(4), 2135\u0026ndash;2202.\u003c/li\u003e\n\u003cli\u003eHawlitschek, F., Notheisen, B., \u0026amp; Teubner, T. (2018). The limits of trust‑free systems: A literature review on blockchain technology and trust in the sharing economy. Electronic Commerce Research and Applications, 29, 50\u0026ndash;63.\u003c/li\u003e\n\u003cli\u003eHileman, G., \u0026amp; Rauchs, M. (2017). Global cryptocurrency benchmarking study. Cambridge Centre for Alternative Finance.\u003c/li\u003e\n\u003cli\u003eHsieh, Y. Y., Vergne, J. P. J., \u0026amp; Wang, S. (2017). The internal and external governance of blockchain-based organizations: Evidence from cryptocurrencies. In Bitcoin and beyond (pp. 48-68). Routledge.\u003c/li\u003e\n\u003cli\u003eHughes, L., Dwivedi, Y. K., Misra, S. K., Rana, N. P., Raghavan, V., \u0026amp; Akella, V. (2019). Blockchain research, practice and policy: Applications, benefits, limitations, emerging research themes and research agenda. International Journal of Information Management, 49, 114\u0026ndash;129.\u003c/li\u003e\n\u003cli\u003eIman, N. (2020). The rise and rise of financial technology: The good, the bad, and the verdict. Cogent Business \u0026amp; Management, 7(1), 1725309.\u003c/li\u003e\n\u003cli\u003eJasanoff, S. (Ed.). (2004). States of Knowledge: The Co-production of Science and Social Order. Routledge.\u003c/li\u003e\n\u003cli\u003eJia, W., Liang, R., \u0026amp; Xu, S. (2022). The public sentiment analysis of \u0026ldquo;Double Reduction Policy\u0026rdquo; on Weibo platform. Frontiers in Psychology, 13, 947756.\u003c/li\u003e\n\u003cli\u003eKozlowski, A. C., Taddy, M., \u0026amp; Evans, J. A. (2019). The geometry of culture: Analyzing the meanings of class through word embeddings. American Sociological Review, 84(5), 905\u0026ndash;949.\u003c/li\u003e\n\u003cli\u003eKozlowski, A. C., Taddy, M., \u0026amp; Evans, J. A. (2019). The geometry of culture: Analyzing meaning through word embeddings. American Sociological Review, 84(5), 905\u0026ndash;949.\u003c/li\u003e\n\u003cli\u003eLe, Q., \u0026amp; Mikolov, T. (2014). Distributed representations of sentences and documents. In Proceedings of the 31st International Conference on Machine Learning (ICML) (pp. 1188\u0026ndash;1196).\u003c/li\u003e\n\u003cli\u003eLee, I., \u0026amp; Shin, Y. J. (2018). Fintech: Ecosystem, business models, investment decisions, and challenges. Business Horizons, 61(1), 35\u0026ndash;46.\u003c/li\u003e\n\u003cli\u003eLesnikowski, A., Belfer, E., Rodman, E., Smith, J., Biesbroek, R., Wilkerson, J. D., \u0026hellip; Berrang‑Ford, L. (2019). Frontiers in data analytics for adaptation research: Topic modeling. Wiley Interdisciplinary Reviews: Climate Change, 10(3), e576.\u003c/li\u003e\n\u003cli\u003eLoughran, T., \u0026amp; McDonald, B. (2011). When is a liability not a liability? Textual analysis, dictionaries, and 10-Ks. Journal of Finance, 66(1), 35\u0026ndash;65.\u003c/li\u003e\n\u003cli\u003eLumineau, F., Wang, W., \u0026amp; Schilke, O. (2021). Blockchain governance\u0026mdash;A new way of organizing collaborations?. Organization Science, 32(2), 500-521.\u003c/li\u003e\n\u003cli\u003eLi, J., Zhang, Y., Chen, H., \u0026amp; Luo, Y. (2023). Semantics of policy discourse in digital government. Government Information Quarterly, 40(1), 101761.\u003c/li\u003e\n\u003cli\u003eLiu, Y., Lu, Q., Yu, G., Paik, H. Y., \u0026amp; Zhu, L. (2022). Defining blockchain governance principles: A comprehensive framework. Information systems, 109, 102090.\u003c/li\u003e\n\u003cli\u003eMa, D., Zhu, X., Wang, Y., \u0026amp; Li, J. (2024). Same same but different: How policies frame societal‑level digital transformation. Government Information Quarterly, 41(1), 101877.\u003c/li\u003e\n\u003cli\u003eMaas, T. Y. (2022). Co-producing the science\u0026ndash;policy interface: Towards common but differentiated responsibilities. Humanities and Social Sciences Communications, 9, 94.\u003c/li\u003e\n\u003cli\u003eMandell, C. (2022). Theoretical foundations and limits of word embeddings. Journal of Cultural Analytics, 7(1), 1\u0026ndash;28.\u003c/li\u003e\n\u003cli\u003eMattli, W., \u0026amp; B\u0026uuml;the, T. (2003). Setting international standards: Technological rationality or primacy of power? World Politics, 56(1), 1\u0026ndash;42.\u003c/li\u003e\n\u003cli\u003eMikolov, T., Chen, K., Corrado, G., \u0026amp; Dean, J. (2013). Efficient estimation of word representations in vector space. arXiv preprint arXiv:1301.3781.\u003c/li\u003e\n\u003cli\u003eMorkunas, V. J., Paschen, J., \u0026amp; Boon, E. (2019). How blockchain technologies impact your business model. Business Horizons, 62(3), 295\u0026ndash;306.\u003c/li\u003e\n\u003cli\u003eNakamoto, S. (2008). Bitcoin: A peer‑to‑peer electronic cash system. https://bitcoin.org/bitcoin.pdf\u003c/li\u003e\n\u003cli\u003eNitzbon, J., Wang, Y., \u0026amp; Lu, H. (2023). Measuring policy debate in a regrowing city by sentiment analysis of online media. Cities, 141, 104469.\u003c/li\u003e\n\u003cli\u003eRodriguez, P. L., \u0026amp; Spirling, A. (2022). Word embeddings: What works, what doesn\u0026apos;t, and how to tell the difference for applied research. Journal of Politics, 84(1), 101\u0026ndash;115.\u003c/li\u003e\n\u003cli\u003e\u0026Oslash;lnes, S., Ubacht, J., \u0026amp; Janssen, M. (2017). Blockchain in government: Benefits and implications of distributed ledger technology for information sharing. Government Information Quarterly, 34(3), 355\u0026ndash;364.\u003c/li\u003e\n\u003cli\u003ePanagiotopoulos, P., Barnett, J., Bigdeli, A. Z., \u0026amp; Sams, S. (2017). Social media in public policy making: A case in e‑participation. Government Information Quarterly, 34(1), 54\u0026ndash;62.\u003c/li\u003e\n\u003cli\u003ePielke, R. A. Jr. (2007). The Honest Broker: Making Sense of Science in Policy and Politics. Cambridge University Press.\u003c/li\u003e\n\u003cli\u003eRauchs, M., Glidden, A., Gordon, B., Pieters, G. C., Recanatini, M., \u0026amp; Zhang, B. (2018). Distributed ledger technology systems: A conceptual framework. Cambridge Centre for Alternative Finance.\u003c/li\u003e\n\u003cli\u003eSchlierkamp, J., Balsiger, J., \u0026amp; L\u0026ouml;rcher, T. (2025). Uncovering the science‑policy interface: Applying bibliometric approaches to the wildfire risk management domain. Humanities and Social Sciences Communications, 12, 77.\u003c/li\u003e\n\u003cli\u003eSchmidt, V. A. (2008). Discursive institutionalism: The explanatory power of ideas and discourse. Annual Review of Political Science, 11, 303\u0026ndash;326.\u003c/li\u003e\n\u003cli\u003eShore, C., \u0026amp; Wright, S. (2011). Conceptualising policy: Technologies of governance and the politics of visibility. Ethnography, 12(1), 1\u0026ndash;26.\u003c/li\u003e\n\u003cli\u003eSucupira, G., \u0026amp; de Albuquerque, J. P. (2022). Sentiment analysis of public services for smart society: Literature review and future research directions. Government Information Quarterly, 39(5), 101749.\u003c/li\u003e\n\u003cli\u003eSzomszor, M., \u0026amp; Adie, E. (2022). Overton: A bibliometric database of policy document citations. Quantitative Science Studies, 3(3), 624\u0026ndash;650.\u003c/li\u003e\n\u003cli\u003eSteinhardt, H. C., \u0026amp; G\u0026ouml;bel, C. (2026). Protecting society from itself: How the social credit system is justified on Chinese social media. Big Data \u0026amp; Society, 1\u0026ndash;15. https://doi.org/10.1177/20539517251403951\u003c/li\u003e\n\u003cli\u003eTan, B. S., \u0026amp; Low, K. Y. (2019). Blockchain as a confidence machine: The problem of trust and challenges of governance. Technology in Society, 59, 101151.\u003c/li\u003e\n\u003cli\u003eTan, E., Mahula, S., \u0026amp; Crompvoets, J. (2022). Blockchain governance in the public sector: A conceptual framework for public management. Government Information Quarterly, 39(1), 101625.\u003c/li\u003e\n\u003cli\u003eTang, H., Ou, M., \u0026amp; Zheng, H. (2025). Between promise and peril: Users\u0026rsquo; risk\u0026ndash;benefit trade-offs in their generative AI usage. Big Data \u0026amp; Society, 1\u0026ndash;14. https://doi.org/10.1177/20539517251410046\u003c/li\u003e\n\u003cli\u003eTreleaven, P., Brown, R. G., \u0026amp; Yang, D. (2017). Blockchain technology in finance. Computer, 50(9), 14\u0026ndash;17.\u003c/li\u003e\n\u003cli\u003evan der Maaten, L., \u0026amp; Hinton, G. (2008). Visualizing data using t-SNE. Journal of Machine Learning Research, 9, 2579\u0026ndash;2605.\u003c/li\u003e\n\u003cli\u003eVan der Meulen, B., \u0026amp; Rip, A. (2000). Evaluation of societal quality of public sector research in the Netherlands. Research Evaluation, 9(1), 11\u0026ndash;25.\u003c/li\u003e\n\u003cli\u003eWeber, R. H. (2018). Corporate governance and blockchains: New paradigms for capital markets? Journal of Banking Regulation, 19(3), 188\u0026ndash;195.\u003c/li\u003e\n\u003cli\u003eWerbach, K. (2018). The blockchain and the new architecture of trust. MIT Press.\u003c/li\u003e\n\u003cli\u003eWilsdon, J., Allen, L., Belfiore, E., Campbell, P., Curry, S., Hill, S., \u0026hellip; Johnson, B. (2015). The metric tide: Report of the independent review of the role of metrics in research assessment and management. HEFCE.\u003c/li\u003e\n\u003cli\u003eXu, J., Chen, M., Kou, G., \u0026amp; Peng, Y. (2019). Understanding the development of blockchain technology: A patent analysis. Technological Forecasting and Social Change, 148, 119735.\u003c/li\u003e\n\u003cli\u003eYermack, D. (2017). Corporate governance and blockchains. Review of Finance, 21(1), 7\u0026ndash;31.\u003c/li\u003e\n\u003cli\u003eYeung, K. (2019). Regulation by blockchain: The emerging battle for supremacy between the code of law and code as law. Modern Law Review, 82(2), 207\u0026ndash;239.\u003c/li\u003e\n\u003cli\u003eZetzsche, D. A., Buckley, R. P., Barberis, J. N., \u0026amp; Arner, D. W. (2017). Regulating a revolution: from regulatory sandboxes to smart regulation. Fordham J. Corp. \u0026amp; Fin. L., 23, 31.\u003c/li\u003e\n\u003cli\u003eZwitter, A., \u0026amp; Hazenberg, J. (2020). Decentralized network governance: blockchain technology and the future of regulation. Frontiers in Blockchain, 3, 12.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-8702396/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8702396/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eBlockchain is no longer only a crypto-finance experiment; it is a contested digital infrastructure shaping financial regulation, data governance, and digital sovereignty. Yet policy debates lack systematic evidence on how blockchain is framed as a governable object across regions and institutions\u0026mdash;and where such framings create coordination frictions for cross-border regulation and standard-setting. This study maps global blockchain policy discourse using a dual-corpus, two-stage design that links broad metadata coverage with full-text inference. We classify documents into governance archetypes and quantify cross-group alignment using a low-dimensional semantic space derived from text, alongside a policy-risk index capturing evaluative stance. Three findings emerge. First, the discourse shifts from monetary framings toward regulatory consolidation and then toward coordination/standard-setting and public-infrastructure integration. Second, meanings cluster by region and issuer type despite shared anchors, indicating persistent interpretive divergence. Third, evaluative stance varies systematically across groups, revealing heterogeneous risk sensitivity. By separating governance logic from evaluative stance, the paper provides a scalable measurement framework and a diagnostic toolkit (e.g., coordination heatmaps and translation tables) to identify where regulatory alignment is most likely to break down.\u003c/p\u003e","manuscriptTitle":"From Crypto Trust to Digital Governance: Mapping Coordination Frictions in Global Policy Discourse","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-20 08:07:50","doi":"10.21203/rs.3.rs-8702396/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"d6d3f0cc-5b24-408e-a789-2e8aa615fe1b","owner":[],"postedDate":"February 20th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":63108839,"name":"Humanities/Complex networks"},{"id":63108840,"name":"Social science/Complex networks"},{"id":63108841,"name":"Physical sciences/Mathematics and computing"}],"tags":[],"updatedAt":"2026-04-09T04:24:11+00:00","versionOfRecord":[],"versionCreatedAt":"2026-02-20 08:07:50","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8702396","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8702396","identity":"rs-8702396","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}