Digital resources in the Social Sciences and Humanities OpenEdition Our platforms OpenEdition Books OpenEdition Journals Hypotheses Calenda Libraries OpenEdition Freemium Follow us

Behind the Counter: Networks of Visibility in the Shenbao

Introduction

Network analysis as employed in this study is not intended to reconstruct a social network of shopkeepers in Shanghai, nor to infer stable interpersonal or organizational ties among historical actors. Instead, it is used deliberately as a heuristic device to detect structured patterns of interaction and institutional encounter as they emerge from the reporting practices of the Shenbao. The analytical object is therefore not the social organization of shopkeepers, but the institutional environments through which shopkeepers entered the public record.

The central historical question animating this analysis is how shopkeeping in Shanghai became progressively governable, that is, how everyday commercial activity was rendered legible, classifiable, and actionable within expanding systems of institutional oversight. By “governable commerce,” we mean not that shopkeepers were successfully controlled or that they organized collectively, but rather that commercial life became increasingly processed through institutional routines: inspected by police, adjudicated by courts, registered by municipal offices, and framed within legal and administrative categories. The making of governable commerce is thus a story not of merchant power or state repression, but of the gradual thickening of an institutional envelope within which commercial activity became publicly visible and bureaucratically intelligible. Network analysis allows us to trace this transformation through the changing patterns of institutional co-occurrence in Shenbao reporting.

Our analysis here is based on a co-occurrence network of named entities built from a corpus of 69,851 Shenbao articles in which at least one lexical form referring to “shopkeepers” appears. Unlike our earlier word co-occurrence analyses, which capture associations among terms and themes in discourse, this approach reconstructs a relational structure of actors and institutions as it surfaces in the press by linking entities that are co-mentioned within the same article. Within this constrained dataset, we extracted and disambiguated names of individuals (PERSON) and institutions (ORG), and connected them through article-level co-mention. The resulting affiliation networks are bipartite, periodized across the five historical intervals defined earlier, and analyzed through their main connected components.[1]

These networks do not represent social relations in the conventional sense. Co-presence in a newspaper article does not imply membership, cooperation, mutual awareness, or enduring association. It reflects editorial selection, narrative framing, and the newsworthiness criteria of the press. As such, the networks capture patterns of discursive association rather than lived social networks. This distinction is not a limitation to be corrected, but a property to be exploited analytically. By treating press co-occurrence as a historical object in its own right, network analysis becomes a tool for identifying regularities in how shopkeepers were narrated, classified, and situated within institutional contexts over time.

From Network Metrics to Discursive and Institutional Interpretation

Within this framework, standard network metrics such as degree, betweenness, and eigenvector centrality are explicitly reinterpreted. They function primarily as indicators of structured presence at the intersection of commerce and institutional intervention.[2] These measures are therefore not ends in themselves. They are tools for detecting how different kinds of institutional encounters—judicial, municipal, policing, financial, charitable, or administrative—structured the presence of shopkeepers in the press over time. Because these graphs are bipartite (PERSON–ORG), centrality values are not directly comparable across node types, and some “standard” measures (especially eigenvector-style indices) behave differently than in one-mode networks. We therefore interpret centralities within mode (PERSON vs. ORG) and rely primarily on weighted and normalized variants to reduce artifacts created by unequal node populations and by changes in network size across periods

A key hypothesis guiding this analysis is that the strong and persistent centrality of certain institutions, such as the Mixed Court (英界會審公廨), the Shanghai Municipal Council (工部局), the various police bodies, or later Republican courts and ministries, does not indicate organizational integration with shopkeepers. Rather, it reflects repeated forms of interaction that brought shopkeepers into the news: litigation, fines, policing, licensing, taxation, disputes, inspections, or relief. In this sense, highly central institutions should be understood as sites of encounter and friction, where everyday commercial activity intersected with systems of governance and control. Their prominence in the network marks the expansion and transformation of the institutional envelope surrounding shopkeeping, not the formation of cooperative or associative networks. The analysis of collocations already gave us a hint as to the overarching presence of certain topics and institutions in relation to shopkeepers. Network analysis will bring us closer to these actors.

This interpretation applies equally to early judicial bodies and to later state organs. The Mixed Court’s centrality in the late nineteenth century signals the judicialization of commercial disputes in a treaty-port setting; the dominance of Republican courts and executive bodies in the 1930s and 1940s reflects the growing role of law and bureaucracy in rendering commerce legible to the state. Similarly, the nodes for individuals with high centrality, whether early intermediaries or later elite businessmen, are not treated as representatives of shopkeepers at large. Their prominence reflects their exceptional visibility in the press: sometimes as repeat litigants, but more frequently as institutional intermediaries, board members, donors, or public figures whose actions repeatedly triggered news coverage. Early figures with high betweenness centrality often function as narrative bridges across institutional domains, linking courts, county offices, and municipal authorities within a sparse news environment. Later elite figures, by contrast, appear as structurally embedded actors within a dense institutional core, inseparable from formal organizations and governance structures. In neither case should individual centrality be conflated with collective organization or leadership among shopkeepers. What the network captures is not who spoke for shopkeepers, but who repeatedly appeared when shopkeepers became newsworthy.

Applied longitudinally, this approach makes it possible to trace the making of governable commerce as seen through the press. The analytical trajectory moves from asking “who were the shopkeepers?” to asking “through which institutions, under what circumstances, and according to what logic did shopkeepers enter the public record?. Across the five periods analyzed in this study, network structures reveal a clear trajectory:

  • from episodic, broker-mediated encounters with courts and local administration,
  • to hub-dominated regimes of policing and regulation,
  • to diversified institutional engagement and elite prominence,
  • to dense institution–elite fusion,
  • and finally to a state-centered, judicialized regime of commercial representation.

Each of these structural configurations corresponds to a different mode through which commerce became governable—that is, legible and actionable to institutions of authority. By explicitly treating network analysis as a heuristic rather than a reconstructive tool, this study leverages the strengths of network methods (pattern detection, comparison, and visualization) to make visible historical regularities that would be difficult to discern through close reading alone.

2. Network Construction, Centrality Profiles, and Visual Strategy

In this section, we explain how the affiliation networks were constructed from the Shenbao articles, how centrality measures were operationalized and interpreted, and how visualization was used as an analytical instrument. Together, these choices define a coherent analytical pipeline designed to support our reading of press-mediated institutional interaction.

All networks analyzed in this study are affiliation networks linking two types of nodes:

  • PERSON nodes: individuals named in the Shenbao articles that mention shopkeepers;
  • ORG nodes: organizations, institutions, or administrative bodies co-mentioned in those same articles.

Edges are constructed through article-level co-occurrence: an individual and an institution are connected if they appear together in a single article. No direct individual–individual or institutions–institutions ties are inferred; any such connections emerge indirectly through shared affiliations. This design reflects the heuristic stance outlined in Section 1. The unit of association is not membership or collaboration, but co-presence in news discourse. Each edge thus represents a reported institutional encounter, broadly defined, rather than a social relationship. Edges are weighted by within-period recurrence: repeated co-mentions of the same individual–institution pair across multiple articles increase the edge weight. This allows differentiation between episodic and recurrent forms of institutional contact.

In the analysis of centrality measures, we proceeded in two steps. First, we produced centrality analyses for the whole dataset for each period. This included the full range of centrality measures as produced using computational methods.[3] The processing of the centrality measures for each period allowed to establish the basis of our analysis and interpretation of network data. Yet, as this produced a voluminous amount of data unfit for useful visualization, we filtered the data to select only the nodes with a degree above 4. We consider that they represent only peripheral mentions and, at any rate, would have been eliminated at the next stage of network analysis. We used this filtered data to build a new set of network visualization for each period using Cytoscape. The analysis and visualizations reported in this study are based on the main connected component (MC) of each period-specific network. Peripheral components were excluded for analytical clarity, as they typically represent isolated or one-off mentions that do not participate in the structured institutional field revealed by the press.

Furthermore, rather than relying on single centrality rankings, this study adopts a centrality profile approach, examining how degree, betweenness, and eigenvector centrality combine within individual nodes. This allows the identification of recurring structural roles that are analytically meaningful within a press-mediated network. In all five periods, the distributions of centrality measures are highly skewed. A small number of nodes concentrate a disproportionate share of connectivity and routing, while the majority remain marginal. Simple “top-N” lists are therefore insufficient to capture the diversity of structural positions.

This profile-based approach avoids conflating different modes of visibility. By examining combinations of centrality measures, it becomes possible to distinguish:

  • nodes that are frequently mentioned but structurally redundant,
  • nodes that appear infrequently but connect otherwise separate institutional domains,
  • nodes embedded within a dominant institutional core.

The analysis focuses on three centrality dimensions: Degree for discursive exposure (frequency of appearance in shopkeeper-related articles); Betweenness for narrative routing (connecting multiple institutional contexts); Eigenvector for core embeddedness (co-occurrence with other highly visible institutions). Rather than treating these metrics as competing indicators, they are read jointly to identify structural roles, such as institutional hubs (high degree, high eigenvector), narrative bridges (low–moderate degree, high betweenness), core-embedded elites (high eigenvector with broad institutional reach), peripheral but strategic nodes (low degree, non-trivial betweenness). These roles are not interpreted sociologically, but discursively: they describe how actors and institutions function within Shenbao’s reporting logic. The various dimensions of these centrality profiles, with concrete examples, are presented in table 1 in the Appendix.

Network visualization in this study is treated as an analytical device, used to corroborate, refine, and sometimes challenge interpretations derived from numerical centrality measures. All graphs were constructed in Cytoscape using consistent visual encodings:

  • Node size scaled by degree;
  • Node shape indicating type (PERSON vs. ORG);
  • Node color scaled by betweenness
  • Force-directed layouts to expose clustering and centralization.

To complement centrality rankings, additional filtered networks were constructed using degree thresholds. Only nodes exceeding a given degree level (e.g., ≥20, ≥150, ≥200, ≥800, depending on period size) were retained, and the resulting subgraphs were visualized (see Figures 1-5 below). These pruned graphs serve three analytical purposes: making the institutional backbone visually explicit; suppressing one-off mentions that obscure structural patterns; enabling visual comparison across periods with vastly different network sizes. Importantly, pruning is not used to redefine the network, but to reveal structure already present in the centrality distributions.

A central methodological principle of this study is cross-validation between metrics and visualization: highly centralized degree distributions should correspond to visually dominant hubs; convergent centrality measures should produce compact cores; declining brokerage should appear as collapsing bridge structures in filtered graphs. Where visual patterns and numerical results diverge, this divergence becomes analytically productive, prompting closer inspection of narrative or institutional context.

The methodological framework outlined above—periodized affiliation networks, centrality profile analysis, and targeted visualization—provides the analytical infrastructure for the empirical investigation that follows. Section 3 applies this framework systematically across five historical periods, examining how each period’s network structure reveals a distinct centrality regime. These regimes capture successive reconfigurations of institutional exposure, from sparse broker-mediated encounters to dense state-centered judicialization. The analysis proceeds chronologically, treating each period’s network as a cross-sectional snapshot whose structure can be compared longitudinally to trace institutional transformation over time.

3. Periodized Network Structures and Centrality Regimes

This analysis of the five period-specific affiliation networks aims at identifying historically distinct centrality regimes. Each regime corresponds to a particular configuration of institutional presence, narrative mediation, and governance logic as reflected in the newspaper’s reporting. These regimes capture how shopkeeping activity became progressively entangled with different institutional frameworks over time.

Period 1 (1872–1884): Broker-Mediated Institutional Encounters

The earliest network is characterized by sparsity, fragmentation, and limited institutional differentiation. The main component contains relatively few nodes and edges, low density, and minimal clustering, reflecting the highly selective circumstances under which shopkeepers entered the newspaper’s field of reporting.

The most central organizational nodes are judicial and basic administrative bodies, notably the Mixed Court of the International Settlement (英界會審公廨), the Shanghai county administration (上海縣署), and the Shanghai Municipal Council (工部局). Their prominence in degree and betweenness centrality indicates that shopkeepers appeared in Shenbao almost exclusively in connection with litigation, complaints, or administrative procedures. The Mixed Court of the French Concession appears relegated to a secondary role. Within the framework adopted here, these institutions should be understood as gateways of visibility. Their centrality reflects the fact that shopkeepers became newsworthy primarily when commercial activity intersected with authority through disputes, infractions, or appeals. A small number of individuals exhibit relatively high betweenness centrality despite modest degree. Figures such as 陳司馬 and 莫邑尊, both Shanghai county magistrate, function as narrative intermediaries, repeatedly appearing in articles that link shopkeepers to multiple institutional arenas. These individuals are not brokers in a sociological sense. Instead, their structural position reflects not just a narrative function: in their capacity as the main local Chinese official, they bridge otherwise separate institutional encounters within the newspaper’s reporting.

The network in Period 1 exhibits a clearly centralized yet weakly aggregated structure. Centrality does not derive from large-scale accumulation or dense clustering, but from repeated acts of mediation across a limited institutional field. Judicial and administrative bodies form a small and stable institutional nucleus, but connectivity between them depends heavily on a handful of recurrent individuals. The network remains sparse and fragmented, with multiple small components and no oligarchic core. What emerges is an incipient institutional environment in which shopkeepers enter the press chiefly through broker-mediated encounters with authority, rather than through sustained organizational integration. Shopkeepers appear as exceptional cases rather than as a recognizable social or economic group, and institutional encounters remain sporadic rather than routinized. In terms of governability, Period 1 represents a regime in which commerce becomes visible to authority only intermittently and exceptionally, through judicial breakdown rather than everyday administration. Shopkeepers are not yet systematically governed; they are episodically adjudicated.

Figure 1 visualizes the Period 1 network backbone, retaining only nodes with degree ≥10. The sparse structure is immediately apparent: a small cluster of institutional nodes (squares) centered on the Mixed Court and county magistrate, with relatively few connecting edges. Individual nodes (circles) appear scattered rather than concentrated, reflecting the episodic rather than systematic nature of shopkeeper-institution encounters. The visual sparseness corroborates the low-density measures: connectivity depends on a handful of repeated co-mentions rather than dense institutional interlocking. Crucially, no dominant hub emerges; instead, the graph reveals a fragmented field in which legibility arises from case-by-case mediation rather than institutional routines. The Mixed Court’s visual centrality corresponds to its degree dominance, but the overall structure remains loose and under-aggregated, a visual signature of limited institutional penetration into commercial life.

Figure 1. Period 1 network backbone (degree ≥10): sparse structure and broker-mediated visibility.

Period 2 (1885–1902): Hub-Dominated Regulation and Policing

The network structure of Period 2 marks a decisive shift. Compared to Period 1, the network becomes more centralized, with a small number of organizational hubs dominating degree, betweenness, and eigenvector centrality simultaneously.

The most prominent nodes in this period are police and order-maintenance bodies, particularly the Patrol Office (巡防局) and related organizations such as the Baojia General Patrol Office (保甲總巡局). Their exceptionally high centrality indicates that shopkeepers entered Shenbao overwhelmingly through encounters with coercive regulatory institutions (Figure 2). The nature of these encounters is revealed through the types of articles that generate co-mentions: shopkeepers appear in connection with theft investigations, disputes over patrol fees, enforcement of closing-hour regulations, inspections of weights and measures, and conflicts over street encroachments. Unlike Period 1, where shopkeepers entered the press through exceptional disputes requiring magistrate intervention, Period 2 reporting reflects the routinization of regulatory contact. The Patrol Office’s centrality derives not from spectacular cases but from accumulated routine interactions: weekly market inspections, standardized fee collection, and systematic enforcement of municipal ordinances. This shift from exceptional to regular encounter marks the institutionalization of regulatory oversight as a permanent feature of commercial life.

This dominance signals intensified regulatory scrutiny. It should be added that the  representation of police bodies is underestimated as they appear also under a variety of police station names. This dispersion effectively dilutes the weight of the police. Yet, we refrained from artificially merging them under a single umbrella. Shopkeepers are visible as objects of inspection, enforcement, and punishment, rather than as participants in broader civic or economic narratives. Judicial and municipal bodies such as the Mixed Court, the county administration, and the Shanghai Municipal Council remain structurally important, but their role shifts. Centrality measures converge strongly at the top of the network, meaning that the same institutions simultaneously function as attachment points, routing hubs, and core nodes. This convergence reflects a hub-dominated regime in which institutions themselves define the pathways through which shopkeepers enter the public record. Individual actors remain visible, but their prominence is increasingly derivative of their repeated association with specific institutional hubs.

The Period 2 network is markedly more centralized than its predecessor, with the clear emergence of very large institutional hubs alongside a long tail of peripheral nodes. Centrality is highly concentrated at the top, where the same organizations dominate degree, betweenness, and eigenvector measures, forming an unmistakable oligarchic core. Brokerage has not disappeared, but it has been absorbed into hub dominance: institutions that attract the most connections also control the principal narrative pathways through which shopkeepers appear in the press. Low-degree bridging actors persist, yet they no longer shape overall connectivity. Unlike the brokerage-dependent structure of Period 1, Period 2 thus represents a shift toward institutional hub consolidation, driven by recurrent interactions with a small set of core regulatory bodies, most notably policing and order-maintenance institutions. This structural transformation corresponds to a shift in the conditions of governability: shopkeepers are no longer exceptional cases requiring judicial mediation, but recurring objects of regulatory routine. Commerce becomes governable through daily inspection and enforcement rather than through crisis intervention. Shopkeepers appear as recurring objects of governance within a relatively stable regulatory apparatus, and institutional practices begin to replace episodic mediation.

Figure 2 applies a substantially higher degree threshold (≥150) to accommodate the much larger network, yet the resulting visualization reveals a strikingly different structure from Period 1. A clear hub-and-spoke architecture emerges, with the Patrol Office (巡防局) and related policing nodes forming visually dominant hubs of exceptional size. The concentration of large institutional nodes at the center  of the graph visually confirms the centrality convergence observed in the metrics: the same institutions dominate degree, betweenness, and routing. Individual nodes remain more peripheral and smaller, indicating their derivative presence. They appear through institutional hubs rather than as independent actors. Compared to the scattered topology in Figure 1, the dense central clustering and radiating periphery in Figure 2 visualizes the consolidation of a regulatory core. The visual hierarchy—massive institutional hubs, thinner connecting edges, peripheral individual nodes—makes the power-law distribution of centrality immediately legible: a small number of institutions structure the recording of the vast majority of shopkeepers.

Figure 2. Period 2 network backbone (degree ≥150): hub consolidation and regulatory core formation.

Period 3 (1903–1916): Institutional Diversification and Elite Emergence

The third period introduces a more complex and differentiated network structure. While centralization remains strong, the institutional field broadens considerably, and new types of organizations enter the core of shopkeeper-related reporting.

Infrastructure agencies, financial institutions, courts, and charitable organizations all achieve high centrality in Period 3. The Shanghai Waterworks Company (自來水公司), the Commercial Press (商務印書館), various native banks (錢莊), and charitable halls (善堂) now appear alongside courts and police as significant nodes (Figure 3). This diversification reflects a fundamental shift in reporting patterns: shopkeepers now enter Shenbao not only through conflict and enforcement, but through participation in subscription drives for public infrastructure, as depositors or creditors in financial disputes, as beneficiaries or donors in relief campaigns, and as ratepayers in utility expansion projects. For instance, shopkeeper names appear in published subscription lists for road paving, in reports of charitable grain distribution during floods, and in notices regarding water or electricity connection disputes. The institutional field through which commerce becomes visible has thus expanded from a narrow judicial-regulatory corridor into a multi-dimensional urban public sphere encompassing welfare, infrastructure, and finance.

This shift reflects the expansion of the institutional envelope surrounding commerce. Shopkeepers are no longer visible solely through policing or adjudication, but through multiple domains of urban governance. Period 3 marks the emergence of a small group of highly visible merchant figures who achieve structural centrality comparable to that of major institutions. Most notably, Zhu Baosan 朱葆三 exhibits exceptionally high degree and eigenvector centrality, indicating sustained co-occurrence with multiple central institutions across the period.[1] His prominence in the network does not reflect leadership of a shopkeeping constituency, but rather his exceptional position as a merchant-philanthropist whose activities intersected systematically with finance, charitable relief, and municipal administration. Zhu’s centrality represents a new phenomenon: the selective elevation of elite shopkeepers into multi-domain institutional prominence. This marks a departure from pattern in Period 2, where individual visibility derived primarily from regulatory or judicial encounters. In Period 3, elite businessmen begin to appear as institutional participants rather than merely as objects of governance. Crucially, however, this public exposure remains exceptional. The vast majority of shopkeepers continue to enter the press only through policing, disputes, or infractions, while figures like Zhu occupy a structurally distinct position at the intersection of commerce, philanthropy, and urban governance.

The diversification of institutional domains through which shopkeepers appear signals an expansion in the scope of governability. Commerce is now legible across multiple registers—financial, infrastructural, charitable, municipal—indicating that governance operates through coordination across specialized bureaucratic functions rather than through a single coercive apparatus. The network of Period 3 maintains a high degree of centralization, but its structure is no longer dominated by a single institutional hub as in Period 2. Instead, it coalesces around a multi-polar oligarchic core spanning several institutional domains. Compared longitudinally, Period 3 represents a phase of core consolidation coupled with institutional diversification: the sparse, broker-dependent structure of Period 1 gives way to the hub-dominated regime of Period 2, which in turn evolves into a more articulated institutional ecology in which elite businessmen and specialized organizations jointly structure press-mediated commercial visibility.

Figure 3 (degree ≥100) captures the structural diversification characteristic of Period 3. Unlike Period 2 with a single dominant hub, this visualization reveals multiple large nodes distributed across different regions of the graph, representing the emergence of financial institutions, infrastructure agencies, and charitable organizations alongside judicial and policing bodies. The topology is more evenly distributed: no single hub dominates to the exclusion of others. Edges now connect not just individuals to institutional hubs, but institutions to each other, indicating their co-appearance in the same articles, a sign of more complex institutional coordination in reporting. Notably, several large circular nodes (individuals) now approach the size of institutional nodes, a visual signature of elite businessmen emergence. The graph thus displays a multi-polar rather than monopolar structure, visually confirming the centrality profile analysis: diversified institutional engagement produces a broader, more differentiated core rather than a single regulatory apex. The transition from radial hierarchy in Figure 2 to networked field structure in Figure 3 visualizes the thickening of the institutional environment surrounding commerce.

Figure 3. Period 3 network backbone (degree ≥100): institutional diversification and elite emergence.

Period 4 (1917–1932): Institutional Saturation and Elite–Institution Fusion

The fourth period represents the apex of institutional density and centrality convergence. The network is dominated by a compact core in which municipal, judicial, financial, and welfare institutions are tightly interconnected. Degree, betweenness, and eigenvector centrality converge almost completely among the top nodes. This indicates that shopkeepers entered the press through a highly routinized and interlocking set of institutional interfaces, rather than through isolated encounters.

This period represents the point of maximum centralization in the network, with a very small fraction of nodes accounting for the overwhelming majority of connectivity and routing (Figure 4). Centrality measures converge almost completely at the top, producing a fully consolidated oligarchic core that integrates municipal infrastructure, judicial and financial organs, medical and welfare institutions, and a narrow stratum of embedded merchant elites. Brokerage as a distinct structural role effectively disappears: connectivity is generated by dense and persistent elite–institution interlocks rather than by individual mediation. The resulting architecture is hierarchical, resilient, and highly cohesive, such that the removal of any single individual would not significantly disrupt overall structure.

Period 4 also consolidates and extends the pattern observed in Period 3, but with a critical structural difference: elite business figures now occupy positions not merely comparable to major institutions, but integrated within the institutional core itself. Yu Qiaqing 虞洽卿 and Wang Yiting 王一亭, alongside their contemporaries, exhibit centrality profiles virtually indistinguishable from those of municipal councils, courts, and financial organs.[2] Their high degree, betweenness, and eigenvector measures all converge, indicating systematic participation across multiple institutional domains simultaneously: chambers of commerce, charitable boards, municipal committees, financial institutions, and public relief organizations. This configuration represents elite–institution fusion: merchant prominence is no longer derived from exceptional wealth or philanthropic activity alone, but from sustained structural embeddedness within the interlocking governance apparatus of Republican Shanghai. Compared to Zhu Baosan’s emergence in Period 3, Yu and Wang’s centrality is denser, more multifaceted, and more tightly coupled to formal organizational roles. They function as nodes through which institutional connections are routinely articulated in press coverage. Yet this structural position remains categorically distinct from that of ordinary shopkeepers, who continue to appear overwhelmingly through regulatory, judicial, or crisis-driven encounters. Elite–institution fusion thus creates a bifurcated regime of commercial visibility: a densely connected elite core that participates in urban governance, and a periphery of ordinary shopkeepers who remain primarily legible through enforcement and adjudication.

Period 4 thus exhibits an oligarchic centrality regime, characterized by dense institutional interlocks and the absorption of elite individuals into the institutional core that integrates municipal infrastructure, judicial and financial organs, medical and welfare institutions, and a narrow stratum of embedded elite businessmen. Brokerage as a distinct role largely disappears, replaced by structural redundancy within a saturated core. As represented in Shenbao reporting, the network takes on the appearance of a corporatist urban governance complex in which commerce is fully embedded within institutional routines. Period 4 thus marks the apex of institutional–elite fusion and provides a critical baseline against which the reconfiguration observed in Period 5 can be meaningfully assessed.

Figure 4 requires a dramatically higher threshold (degree ≥800) due to the massive network size, yet even at this extreme cutoff, a dense core structure persists. What emerges is a tightly clustered oligarchic center in which institutional and elite nodes are visually indistinguishable in size and interconnection. The graph’s compact, nearly circular topology reflects the saturation observed in centrality measures: nodes are not arranged hierarchically around a single hub but form a densely interconnected core where nearly every node connects to every other node through thick, overlapping edges. This structure visualizes elite-institution fusion: prominent merchants (circular nodes) are embedded within rather than peripheral to the institutional cluster. The visual density—multiple large nodes, thick intersecting edges, minimal empty space in the core—makes the network’s resilience apparent: removing any single node would not disrupt overall connectivity. In contrast to hub-dependent or diversified structures in the earlier periods, the compact, saturated core in Figure 4 visualizes the apex of institutional-elite interlocking in Republican Shanghai.

Figure 4. Period 4 network backbone (degree ≥800): dense oligarchic core and elite-institution fusion.

Period 5 (1933–1949): State-Centered and Judicialized Visibility

The final period marks a profound reorientation of the network around state and judicial institutions. Central government organs and courts dominate all centrality measures, eclipsing both municipal bodies and elite commercial figures (Figure 5).

Institutions such as the National Government (國民政府), the Executive Yuan (行政院), and the Shanghai First and Second Special District Courts (上海第一、第二特區地方法院) occupy the top positions in degree, betweenness, and eigenvector centrality. Shopkeepers now appear predominantly in reports of commercial dispute adjudication under the new Civil Code, tax assessment appeals processed through Finance Ministry channels, licensing conflicts mediated by municipal bureaus under Executive Yuan authority, debt litigation handled by the Special District Courts, and bankruptcy proceedings formalized under Republican commercial law. Unlike earlier periods where mentions of shopkeeper arose from diverse institutional encounters, Period 5 reporting is overwhelmingly structured by formal legal categories and administrative procedures. Articles now regularly cite specific legal articles, court case numbers, and administrative regulation codes when mentioning shopkeepers, a discourse of legibility that reflects the state’s expanded capacity to classify and process commercial activity through standardized bureaucratic-legal frameworks.

This reflects a renewed judicialization of commercial activity. Shopkeepers appear as legally categorized subjects rather than as participants in urban society. While a new cohort of elite individuals remains visible in Period 5, notably Wang Xiaolai 王曉籟 and Lin Kanghou 林康侯, their structural position differs markedly from the elite–institution fusion characteristic of Period 4.[3]  These figures exhibit high centrality, but their prominence is now more narrowly concentrated within state administrative and judicial spheres rather than distributed across multiple civic and commercial domains. Wang and Lin appear primarily in connection with formal government positions, legal proceedings, or direct state-mediated commercial regulation, rather than through the philanthropic, financial, and municipal networks that characterized Yu Qiaqing’s and Wang Yiting’s public image.

This shift reflects a broader structural transformation: as courts and central state organs come to dominate the network, elite businessmen lose their multi-domain brokerage capacity and are increasingly subsumed within hierarchical state-commercial relations. The diversified institutional reach that defined elite prominence in Periods 3 and 4, spanning charity, finance, municipal governance, and commerce, contracts into a more narrowly state-centered form of visibility. Elite businessmen in Period 5 thus appear less as interface actors bridging multiple institutional spheres and more as state-adjacent figures whose commercial prominence is legible primarily through their integration into Republican legal and administrative structures. Unlike earlier business elite figures, their presence is less multi-domain and more tightly coupled to state authority. The judicialization of commercial activity in Period 5 marks a shift toward formalized governability. Shopkeepers are rendered governable not through personalistic elite mediation or municipal coordination, but through legal classification and bureaucratic processing by central state organs. Commerce becomes an administrative category rather than a domain of urban negotiation.

Figure 5 (degree ≥500) reveals a structural reconfiguration distinct from the saturated core in Period 4. The graph displays a more hierarchical architecture, with several very large institutional nodes, notably the Special District Courts and central state organs, forming a clearly dominant apex, while a broader field of medium-sized nodes surrounds them. Figure 5 shows a more stratified topology: a small judicial-state pinnacle, a substantial middle tier of municipal and administrative bodies, and a periphery of individuals and smaller institutions. The visual hierarchy corresponds to the metrics: state-judicial institutions achieve exceptional centrality while other nodes, though still visible, occupy subordinate positions. Crucially, individual nodes (elite businessmen) are now noticeably smaller relative to institutional nodes than in Period 4, visually confirming their reduced multi-domain prominence. The structure  of the graph—clear apex, thicker middle layer, dispersed periphery—visualizes the shift from elite-institution fusion to state-centered hierarchy. Where Figure 4 showed a resilient, redundant core, Figure 5 displays a pyramidal structure in which centrality flows through a narrow judicial-administrative summit.

Figure 5. Period 5 network backbone (degree ≥500): state-centered hierarchy and judicial apex.

The five period-specific networks analyzed above reveal a clear structural trajectory: from sparse, broker-dependent encounters (Period 1), through hub-dominated regulation (Period 2) and diversified institutional engagement (Period 3), to dense elite-institution fusion (Period 4), and finally to state-centered judicialization (Period 5). The analysis charts the gradual thickening of an institutional environment that progressively framed, regulated, and classified everyday commerce.  What emerges is not a story of increasing institutional saturation, in which commercial life is steadily absorbed into the expanding infrastructures of governance, adjudication, and public order. Yet these descriptive characterizations require interpretation. What do these centrality regimes signify historically? How should changes in network topology be translated into claims about governance, state capacity, and commercial legibility?

4. From Centrality Regimes and the Making of Governable Commerce

This section shifts from description to interpretation, examining three analytically distinct dimensions of the longitudinal transformation: (1) the relationship between centrality regimes and evolving state capacity, (2) the progressive judicialization of commercial activity, and (3) the rise and contraction of elite merchant mediation as a transitional governance mechanism. Together, these interpretive moves ground the network analysis in broader historiographical debates about state-building, legalism, and commercial regulation in modern China.

The periodized network structures identified in the preceding section trace the progressive transformation of the institutional conditions under which shopkeeping became visible, legible, and actionable within public discourse. The succession of centrality regimes reveals a long-term process through which commerce in Shanghai was rendered increasingly governable, first through episodic judicial encounter, then through regulatory aggregation, elite mediation, and finally through state-centered legal rationalization. Across the five periods examined, changes in network centrality register changes in the density, reach, and differentiation of institutional intervention. In this sense, centrality regimes can be read as proxies for evolving state capacity, not in terms of fiscal extraction or coercive power alone, but in terms of the ability to classify, monitor, and process commercial activity through institutional routines.

In Period 1, state capacity appears limited and uneven. Judicial and administrative institutions are present but operate as narrow gateways of visibility. Shopkeepers enter the public record sporadically, typically through disputes or infractions that escalate to courts or county offices. The network’s sparse structure and reliance on narrative intermediaries reflect a governance environment in which commercial activity remains largely outside systematic oversight. By Period 2, the emergence of policing bodies as dominant hubs signals a qualitative shift. The concentration of degree and betweenness in institutions such as the Patrol Office indicates not merely intensified enforcement, but the routinization of regulatory contact. Shopkeepers are no longer exceptional cases; they are recurrent objects of inspection and discipline. In network terms, the consolidation of institutional hubs corresponds to an expansion of the state’s day-to-day regulatory reach into commercial life.

Periods 3 and 4 further demonstrate that increasing state capacity does not necessarily imply centralization alone. The diversification and subsequent saturation of the institutional core suggest a capacity to govern commerce through multiple, overlapping domains: infrastructure, finance, welfare, adjudication, and municipal administration. The dense core of the network in Period 4 reflects not coercive penetration alone, but organizational integration, a hallmark of more mature urban governance. Finally, Period 5 reveals a different dimension of state capacity: the ability to reframe commerce primarily through legal and bureaucratic categories. The dominance of courts and central state organs indicates that shopkeeping is now processed through formalized legal procedures rather than through localized administrative negotiation..

One of the most striking long-term trends revealed by the network analysis is the growing prominence of legal institutions, first courts, then specialized judicial bodies, as central nodes in shopkeeper-related reporting. This pattern points to a gradual judicialization of commerce, in which disputes, infractions, and regulatory issues are increasingly framed in legal terms. In early periods, legal institutions function primarily as episodic arenas for conflict resolution. Their centrality reflects moments of breakdown rather than routine governance. Over time, however, courts become stable and dominant sites of commercial exposure. By Period 5, they are no longer exceptional interfaces but the primary institutional lens through which shopkeeping is represented in the press. The network analysis shows that shopkeepers did not merely encounter the law more frequently; they increasingly appeared as legal subjects in the public record. Importantly, this judicialization and the concentration of centrality in courts and state organs suggests that commercial actors were rendered legible to the state primarily as objects of adjudication and administration.

The judicialization of commercial activity proceeded alongside, and sometimes through, a parallel but ultimately distinct phenomenon: the temporary prominence of elite businessmen as structurally central nodes. Understanding how elite centrality relates to state-centered judicialization requires tracing the elite merchant trajectory as a discrete analytical object. Figures such as Zhu Baosan, Yu Qiaqing, Wang Yiting, Wang Xiaolai or Lin Kanhou appear not as collective representatives, but as interface actors whose activities intersect with multiple institutional domains. From a governance perspective, these elite businessmen can be understood as facilitating a mode of rule that operates through co-optation and participation rather than pure coercion. Their prominence reflects a moment in which commercial prominence could translate into sustained engagement with urban institutions through boards, committees, philanthropy, and public initiatives. Yet the network analysis also reveals the fragility of this arrangement. Elite mediation does not persist into Period 5. As state and judicial institutions come to dominate the network, elite individuals lose their multi-domain centrality and are subsumed within a state-centered hierarchy. This suggests that elite participation functioned as a transitional mechanism, enabling institutional expansion without fundamentally decentralizing authority.

The longitudinal trajectory of elite merchant centrality reveals a transitional mode of commercial governance operative between the 1900s and early 1930s, unfolding in three distinct phases. In Perid 3, elite businessmen like Zhu Baosan first achieve sustained centrality through multi-domain activity spanning commerce, philanthropy, and municipal affairs. Their prominence reflects a moment when commercial wealth could translate into civic visibility across charitable boards, financial institutions, and public relief initiatives. Crucially, this presence remains exceptional rather than representative: these figures operate in a structurally distinct sphere of urban governance, not as mediators between institutions and ordinary shopkeepers. In Period 4 elite centrality intensifies and becomes embedded within the institutional core. Yu Qiaqing and Wang Yiting exhibit centrality profiles comparable to major governmental institutions themselves, participating systematically in chambers of commerce, charitable organizations, and municipal committees.

Period 4 thus marks a moment of elite–institution interlocking in which prominent businessmen become constitutive of the governance apparatus rather than external to it. This configuration enabled governance through co-optation and structural integration rather than pure regulatory coercion, yet remained fundamentally oligarchic: elite access did not extend to ordinary shopkeepers, who continued to appear primarily through policing and litigation. Finally, as central state institutions like courts, the Executive Yuan, the National Government, come to dominate the network in Period 5, elite businessmen lose their multi-domain reach. Wang Xiaolai and Lin Kanghou remain visible, but their prominence is now narrowly tied to formal state positions and legal proceedings rather than distributed across civic and philanthropic domains. Elite businessmen are no longer interface actors bridging institutional spheres, but state-adjacent figures whose visibility operates primarily through hierarchical administrative structures.

This trajectory suggests that elite-mediated governance functioned as a transitional mechanism specific to a conjuncture when state capacity was expanding but institutional infrastructure remained incomplete. Elite businessmen filled structural gaps, providing coordination and legitimacy that formal institutions could not yet systematically supply. However, as state legal and administrative capacity intensified in the 1930s-1940s, elite participation was absorbed into, rather than displaced by, state-centered governance. The rise and contraction of elite merchant centrality thus indexes shifting modalities of state-commercial articulation, from episodic encounter, through co-optive participation, to hierarchical integration.

The trajectories of judicialization and elite mediation intersect in analytically productive ways. Both phenomena reflect expanding state capacity, yet they represent different modalities through which commerce became institutionally legible. Judicialization operated through formalized legal categories and bureaucratic processing; elite mediation operated through personal prominence and multi-domain participation. The first proved durable and intensifying; the second proved transitional and contingent. Their interaction illuminates a fundamental dynamic: as state institutions developed the capacity to classify and process commerce systematically through law and administration, the structural space for elite-mediated governance contracted. Taken together, these transformations—hub consolidation, institutional diversification, judicialization, elite mediation, and state absorption—reveal a long-term process through which commerce in Shanghai was rendered increasingly governable. Shopkeeping becomes governable not when it is socially organized, but when it is regularly inspected, legally classified, and administratively processed. The Shenbao functions not merely as a passive observer, but as a discursive infrastructure that reflects and reinforces the institutionalization of governance.

Finally, it is important to underscore what this approach contributes to the historiography of commerce and what it deliberately leaves aside. This study offers a structural reading of institutional encounter, allowing historians to trace how governance, law, and administration progressively enveloped everyday commercial life. The contribution of network analysis lies in its capacity to detect patterned regularities across thousands of articles, to compare institutional configurations across long time spans, and to render visible transformations that are difficult to apprehend through close reading alone. At the same time, the approach places clear limits on interpretation. Centrality does not equal power; co-occurrence does not equal connection; and presence in the press does not equal voice. These limits are not shortcomings, but safeguards against over-interpretation. Read within these bounds, network analysis becomes a powerful ingredient of historical methods. It enables a shift in perspective, from asking who shopkeepers were, to asking through which institutions, and under what conditions, shopkeepers became visible, regulated, and governable in modern China. By treating Shenbao reporting practices as the object of analysis rather than as transparent evidence of social organization, network analysis has made it possible to trace how shopkeeping was progressively rendered visible, classifiable, and actionable within public discourse.

Conclusion

This study has used network analysis to examine how shopkeeping was progressively rendered visible and governable through press-mediated institutional interaction. By treating Shenbao reporting practices as the object of analysis, the paper has shown that affiliation networks derived from co-mention map the expanding institutional envelope within which everyday commerce was narrated and regulated. Across five historical periods, the analysis identifies a clear structural trajectory. What begins in the late nineteenth century as episodic, broker-mediated encounters with courts and local administration develops into hub-dominated regimes of policing and regulation, followed by a phase of institutional diversification and elite–institutional interlock. By the Republican period, this configuration gives way to a dense, state-centered and judicialized regime of commercial exposure, in which shopkeepers appear primarily as legal and administrative subjects. Network centrality thus tracks not merchant organization, but the growing capacity of institutions to classify, monitor, and process commercial activity.

Centrality measures function here as indicators of discursive exposure, narrative routing, and institutional embedding rather than social power or cohesion. This avoids reifying press co-occurrence as social structure while retaining network methods’ analytical advantages: pattern detection, comparison across time, and the ability to synthesize large textual corpora. More broadly, these findings contribute to debates on state capacity and legalism in modern China by shifting attention from formal policy or elite discourse to the everyday interfaces through which commerce encountered authority. Shopkeepers emerge not as a collective actor, but as a social category increasingly processed through institutional routines and legal frameworks. In this sense, the making of governable commerce appears less as a top-down imposition than as a cumulative outcome of repeated, routinized encounters, encounters that Shenbao both recorded and helped to normalize.

Finally, this study demonstrates the value of reflexive computational history. Network analysis, applied to large-scale datasets like the 69,851 articles examined here, opens new ways of reading historical sources that reveal long-term transformations in governance and public visibility that might otherwise remain diffuse or anecdotal. For historians of modern China, such an approach offers a means of connecting micro-level encounters to macro-level processes without collapsing one into the other.


Appendix

 


 

Statement on the Use of AI

Artificial intelligence tools — ChatGPT 5.2 and Claude Sonner 4.5—were used in this research as analytical and writing support instruments.

Specifically, ChatGPT was employed to:

  • help structure and systematize the analysis of network-derived metrics (degree, betweenness, eigenvector centrality) across multiple periods;
  • assist in the analytical framing of network results, particularly by reformulating centrality measures as heuristic indicators of press-mediated institutional interaction rather than representations of social networks;
  • support the drafting, revising, and harmonizing of analytical prose, including the synthesis of period-specific findings, comparative interpretations, tables, and figures;
  • suggest conceptual typologies, visual strategies, and narrative organization consistent with historical methodology and the epistemological limits of press-based data.

Claude Sonnet provided comprehensive editorial support for the final draft included: (1) complete review of the text to spot typos, errors, mismatches, and contradictions; (2) adding concrete empirical texture to period descriptions by identifying specific types of institutional encounters (theft investigations, subscription lists, Civil Code citations) (3) enhancing figure analysis by writing detailed analytical readings of five network visualizations; (4) strengthening section transitions to improve argumentative flow from methodology through empirical analysis to interpretation; (5) creating two visualizations—a conceptual diagram (Figure A1) mapping eight profile types in three-dimensional centrality space and a color-coded table of centrality profiles (Figure A2)—delivered as both HTML and high-resolution PNG files.

All data selection, preprocessing, network construction, metric computation, visualization, and historical interpretation were designed, executed, and validated by the author in R-Studio. The AI tools did not access external archives, generate empirical results, or introduce historical claims independently. Final arguments, interpretations, and wording remain the sole responsibility of the author.


[1] On Zhu Baosan and his role as a businessman, philanthropist and political actor, see Christian Henriot, “A Tale of Three Merchants: Social Worlds, Political Networks, and Public Sphere in Modern Shanghai” in Armand, Henriot, Lien (eds.), Performing Power: Elites and the Making of Modern China. For a full biography of Zhu, Qian Maowei 钱茂伟 and Ying Fangzhou 应芳舟, Yi nuo jiu ding: Zhu Baosan zhuan 诺九鼎: 朱葆三传 (A Promise of Nine Cauldrons: The Biography of Zhu Baosan)  (Beijing: Zhongguo shehui kexue chubanshe, 2008).


[2] On Yu Qiaqing and Wang Yiting, see Christian Henriot, “A Tale of Three Merchants: Social Worlds, Political Networks, and Public Sphere in Modern Shanghai” in Armand, Henriot, Lien (eds.), Performing Power: Elites and the Making of Modern China. For a full biography of Yu, see Feng, Xiaocai 馮筱才. Zhengshang zhongguo. Yu Qiaqing yu ta de shidai 政商中國:虞洽卿與他的時代 (Political Merchants China: Yu Qiaqing and his Time). Beijing: Shehui kexue wenxian chubanshe, 2013.


[3] On Wang Xiaolai and Lin Kanghou, see Christian Henriot, “A Tale of Three Merchants: Social Worlds, Political Networks, and Public Sphere in Modern Shanghai” in Cécile Armand, Christian Henriot, Ling-ling Lien (eds.), Performing Power: Elites and the Making of Modern China (Berlin: De Gruyter-Brill, 2026), forthcoming. For a full biography of Wang, see Wong, Katrine Victoria. “Wang Xiaolai 王曉籟 (1887-1967) : A Biography.” Doctoral dissertation, Lyon, 2019.


[1] The five periods analyzed are: Period 1 (1872–1884), Period 2 (1885–1902), Period 3 (1903–1916), Period 4 (1917–1932), and Period 5 (1933–1949). These periodization choices reflect the temporal distribution of articles mentioning shopkeepers in the Shenbao.

[2] Degree centrality measures how frequently a person or organization appears in articles involving shopkeepers. High degree thus signals repeated exposure within Shenbao’s coverage of commercial life. For organizations, it points to recurrent institutional contact with shopkeepers; for individuals, it reflects repeated narrative appearance rather than representativeness. Betweenness centrality highlights entities that recurrently connect otherwise distinct institutional contexts in reporting. In this press-mediated setting, betweenness captures narrative routing rather than brokerage: courts, police offices, municipal bodies, and financial institutions are linked through the way stories are told, not through social mediation. Individuals with high betweenness are best understood as recurring narrative figures—litigants, officials, intermediaries, or elite actors—whose cases or actions bridge multiple institutional arenas in the news. Eigenvector centrality identifies nodes embedded in the newspaper’s dominant institutional core. High eigenvector values do not indicate prestige in a sociological sense, but consistent co-appearance with other highly visible institutions. It thus captures the consolidation of an institutional center of gravity within the press narrative.

[3] To prepare the data (cleaning, normalization) and prepare the datasets to be used for network analysis, we operated in R using the igraph, tidygraph, and tydiverse libraries. We also used tydiverse to prepare the datasets for Cytoscape, the network application we used for the filtered data. All the scripts are available on our GitHub repository.

Christian Henriot
Christian Henriot

The text only may be used under licence Creative Commons Attribution Non Commercial 4.0 International. All other elements (illustrations, imported files) are “All rights reserved”, unless otherwise stated.


OpenEdition suggests that you cite this post as follows:
Christian Henriot (February 8, 2026). Behind the Counter: Networks of Visibility in the Shenbao. Elites, Networks and Power in modern China. Retrieved June 18, 2026 from https://doi.org/10.58079/15nd8


You may also like...