Measuring China’s Open Sharing for scientific instruments: an ascendency analysis with benchmark use-time data

Abstract

Over the past decade, China has significantly improved the retention and utilization rates of scientific instruments. However, systemic inefficiencies persist due to fragmented governance structures, human capital mismatches, and uneven regional development. This study evaluates the sustainability of 15 regional organizations managing large-scale scientific instruments from 2013 to 2022 by applying ecological network analysis (ENA) to model use-time as flow-based networks. Based on the concept of ascendency, this work introduces the degree of order (α) as a dynamic indicator of ecological efficiency. Empirical results reveal that while total instrument use-time (TST) has more than doubled, efficiency gains have been primarily driven by scale expansion rather than improved coordination. Technical staffing in typical regions such as Lanzhou increased by only 112% over ten years, lagging behind a > 200% growth in total use-time, leading to a decline in average mutual information (AMI) and suboptimal α performance. Furthermore, a statistically significant inverse relationship is found between average sharing time and ecological efficiency, indicating that quota-based sharing policies may inadvertently reduce systemic resilience. This paradox reflects a critical trade-off between quantity-driven incentives and the structural integrity of sharing networks. Regression analysis identifies GDP per capita and policy timeliness as significant predictors of organizational efficiency, yet accounts for only 65% of its variance, underscoring the need to incorporate operational and structural variables. To address these challenges, this study proposes an “efficiency window” framework that defines optimal α ranges, enabling more adaptive evaluation across development stages. The findings offer a transferable model for assessing sustainability in resource-sharing infrastructures, while emphasizing the contextual limits of applying centralized governance approaches to decentralized systems. For policymakers, this study underscores the need to align economic incentives, staffing structures, and network flexibility to foster resilient, inclusive, and efficient scientific innovation ecosystems.

Introduction

Scientific instruments, such as high-resolution microscopes, mass spectrometers, DNA sequencers, and electron accelerators, are essential tools for advancing research across disciplines (Coulter, 1978; Lu, 2008; An and Holme, 2021). These complex and costly machines are typically acquired and maintained through public investment and are hosted by universities, research institutes, or government laboratories. In modern science systems, they constitute a critical infrastructure component and are simultaneously a scarce public resource. Therefore, the construction and operation quality of scientific instruments represent a strategic imperative for emerging nations committed to closing the science and technology gap with the developed world (Yang et al. 2024). Their availability and usage patterns, particularly in university and government research systems, critically affect a nation’s capacity for innovation, knowledge production, and policy responsiveness (Champenois and Etzkowitz, 2018).

China provides an illustrative case. Over the past two decades, it has made unprecedented investments in scientific infrastructure. By 2020, more than 120,000 large-scale instruments (defined as single unit valued over RMB 500,000 or ~70,000 USD) were operating within central-level institutions, with a total asset value exceeding RMB 180 billion and an annual growth rate over 20% (Xu et al. 2022). These instruments are heavily concentrated in national labs, elite universities, and government institutes. Despite this physical abundance, under-utilization remains a persistent concern (Wang et al. 2023). Idle time, redundant procurement, and regionally uneven distribution have triggered public scrutiny and called into question the effectiveness of investment-driven innovation (Liu et al. 2019; Williams, 2021).

In response, the Chinese government launched the Open Sharing of large-scale scientific instruments in 2014, which referred to a national-level policy and institutional framework designed to promote co-construction, shared access, and cross-institutional use of major scientific research infrastructures. From 2018 onwards, institutions were mandated to annually report use-time data, i.e., the number of effective machine-hours each instrument was in operation (Wang, 2020). This policy aimed to improve efficiency and promote inter-institutional collaboration (You et al. 2024). The impact has been significant: from 2014 to 2022, average annual use-time per instrument increased from ~500 to over 1,300 h, while average external (shared) service hours rose from under 50 to more than 200 h (Yu, 2022). The institutional sharing rate reportedly surpassed 90% in many cases. These changes not only reflect technical improvements but also signal a policy-driven transformation of how public research resources are managed.

Yet fundamental questions remain. Is the rise in sharing time genuinely contributing to system efficiency, or is it merely a statistical artifact of compliance incentives? What constitutes a reasonable standard of utilization, considering disciplinary differences and institutional capacities? Can machine-hour metrics alone capture the dynamics of resource fairness, resilience, and long-term system viability? While traditional evaluations focus on throughput (e.g., total use-time or external service rate), they often neglect the systemic interactions, interdependencies, and vulnerabilities within instrument-sharing networks. These are not merely operational concerns, but governance challenges in managing scientific instruments as common-pool resources.

Sustainability, especially when assessed under policy-driven framework, is a clearly defined yet complex and inter-disciplinary issue requiring a balance between productivity (efficiency), adaptability (resilience), and accessibility (equity) (Shmelev and Shmeleva, 2025). At the same time, it is necessary to adopt an outside-in perspective that enables stakeholders to move beyond self-centered views and reach a consensus on the supporting strategies for sustainability management (Derksen and Mithofer, 2022). An efficient system delivers high output, while a resilient system can absorb shocks (e.g., COVID-19 disruptions). Moreover, an equitable system allows small institutions and emerging researchers to access core facilities. However, current evaluation methods, which rely on linear metrics and a single-institution perspective rather than on a holistic system-level perspective (Ostrom, 2009), poorly equipped to capture these dynamic, interrelated dimensions.

To address this gap, this paper introduces an ecological network analysis (ENA) framework based on the concept of ascendency, a system ecology construct that integrates total activity with organizational structure (Ulanowicz, 2009; Fath and Scharler, 2019). In this framework, the scientific instrument use-time is not treated as an isolated statistic but as a flow across nodes (i.e., management units), whose structure reveals the system’s capacity to coordinate, buffer, and evolve (Fath and Scharler, 2019; Ulanowicz, 2020; Zisopoulos et al. 2023). Applying this model with benchmark data from 15 regional scientific instrument organizations in China, this work evaluates how economic conditions, governance models, and policy timing affect the ecological efficiency of shared scientific instrument resources. The findings reveal unexpected trade-offs: for instance, increasing shared use-time may reduce systemic efficiency, particularly when monopolization or over-concentration occurs. By extending sustainability analysis to scientific instrument systems, this study offers a new diagnostic approach for assessing shared infrastructure in innovation policy.

Conceptual framework and hypotheses

Rethinking efficiency in instrument sharing systems

Efficiency is a common measure of how well an organizational unit utilizes its resources to produce benefits. It is often evaluated using either parametric approaches, such as regression analysis, or nonparametric methods like data envelopment analysis (DEA) (Park and Cho, 2011; Nguyen and O’Donnell, 2023). The relationship between allocative and Pareto efficiency is frequently used interchangeably due to the notion that a perfectly competitive market yields optimal societal outcomes (Pirgmaier, 2017). However, these traditional metrics focus on individual units or linear input-output ratios, which may inadequately capture systemic interactions in complex resource networks. Mathematical methods like DEA can handle efficiency calculations involving multiple inputs and outputs, but the resulting efficiency frontiers are often neither Pareto-optimal nor aligned with ecological efficiency principles (Kannan et al. 2013). Therefore, there is a need for a more accepted, systematic, and ecological concept of efficiency.

Eco-efficiency, rooted in ecological network analysis (ENA), diverges by emphasizing system-level interactions, resilience, and sustainability (Ulanowicz, 2009). In the context of scientific instrument management, this means evaluating how regional organizations balance resource allocation, sharing behavior, and their capacity to adapt to shocks. These are factors that are overlooked by DEA or Pareto-based assessments. A complex efficiency problem arises from considering the adequate weighting of common development in defining efficiency, both for organizational and policy purposes (Xu et al. 2018). If left unchecked, increasing efficiency in the operation and management of scientific instruments can result in full-load operation or even the collapse of the entire technical support system. Unlike conventional input-output efficiency metrics, ecological efficiency in this study places greater emphasis on systemic adaptability (such as the ability to withstand or respond to shocks, i.e., resilience) and equitable access (such as accepting small users), rather than focusing solely on throughout maximization.

There is evidence that the operation of instrument platforms often aims at supporting scientific and technological outputs to obtain returns of knowledge and resources, forming a closed-loop sustainable system (Wang et al. 2021b). In this transformation process, even with linear utilization of scientific instruments, there are synergistic forces and antagonistic tendencies driving evolutionary trends of interweaving competition and cooperation (Ulanowicz et al. 2009). The cost-effectiveness over eco-efficiency is actually a contentious topic, having multiple impact on almost all entrepreneurial activities of linear economy (Kyriakopoulos, 2021). As an essential part of the value chain of scientific and technological innovation, instrumental efficiency should not be addressed as merely a numerical issue, nor should it be viewed as a straightforward linear practice. Operating instruments is a linear business model of providing testing time services by instrument technicians; however, it is neither entirely linear for their management units nor driven solely by cost-benefit logic for innovation activities.

In the current context of scientific resource utilization, the problem of instrument efficiency emphasized by Chinese government departments likely has little to do with Pareto efficiency, as the construction costs of large-scale scientific instruments are annually covered by the central budget, leading to significant imbalances in their distribution (Liu et al. 2019). Therefore, it is suggested that instrument utilization should be evaluated through the lens of eco-efficiency, which measures the effectiveness and operational constraints of multiple nodes within an observed system at the collective level (Ulanowicz, 2009). From this perspective, efficiency in scientific instrument sharing should be redefined through ecological principles. That is, not merely how much is used or how often it is used, but whether usage patterns promote systemic balance, resilience, and sustainability over time. This reconceptualization provides a more comprehensive basis for evaluating performance and designing interventions in complex sharing systems. To guide empirical analysis, the following hypothesis is proposed:

Hypothesis 1 (H1): Increasing diverse sharing time enhances resource throughput but may reduce ecological efficiency by compromising the systemic balance between efficiency and flexibility.

Use-time as a networked flow

Implemented for the first time in 2018, the open-sharing evaluation mechanism that adopts instrument use-time as a key indicator is a continuation of China’s initial Open Sharing management goals (Wang, 2020). Instrument utilization, under the strategy of Open Sharing for scientific instruments, involves balancing customer demand and service supply, where “use-time” (i.e. total machine-hours) establishes interdependent corroboration. Although it is often a mechanically linear activity when focused on one particular technical service for one customer, use-time can be delivered to a diverse customer base to help achieve organizational purposes. Here, linearity refers to a direct input-output relationship (e.g., hours used vs. services delivered). Globally, analogous metrics, such as equipment utilization rates in the U.S. National Science Foundation’s Major Research Instrumentation Program (NSF, 2022), prioritize quantitative outputs and neglect network interdependencies. However, when the time scale or organization size is expanded, instrument use becomes a purposeful, interdependent, and nonlinear process. Nonlinearity arises from feedback loops (e.g., shared scientific instruments attracting more users) and trade-offs (e.g., efficiency vs. sustainability), which are central to ascendency analysis. Additionally, use-time carries a wealth of information about organizational behavior, yet this information has often been regrettably neglected. This represents a valuable opportunity to clarify the open-sharing benefits from a use-time perspective and to enhance the well-being of instrument management units and staff.

Sharing time within use-time is a crucial feature, earning extra points in both internal and external open and sharing evaluations. The notion of sharing time aggregates diversity and choice in practice and can be briefly defined through external services. While it is logical that more sharing services lead to higher instrument utilization, the reverse is not necessarily true. Externally shared machine-hours are essentially rental activities, which may be highly sought after as a means of establishing a monopoly or near-monopoly to control shared technical organizations that facilitate exchange (Webster, 2021). If the Open Sharing framework aims to build a thriving economy for scientific and technological innovation, it should organize use-time and financial resources across all scales with minimal rent-seeking behavior. Moreover, advocating for sharing raises a paradox of scale, where sharing activities fail to deliver as a consequence of scale growth (Andreoni, 2020). By processing use-time flows between any two nodes, the function and significance of these interactions of information can be clarified for node management members and even for the whole organization system.

The closed-loop operation of artificial systems formed between scientific research and instrument utilization provides a foundation for mutual benefit but also risks vulnerability when pursuing maximum efficiency, potentially leading to adverse social side effects (Zisopoulos et al. 2022). As long as there is interaction use-time between nodes in a network, the structural information carried by use-time tells a lot about the relationship between efficiency and resilience. Likewise, the European Strategy Forum on Research Infrastructures (ESFRI) emphasizes cross-border collaboration and centralized platforms for large-scale equipment sharing, yet relies on utilization rates and cost-benefit analyses for evaluation (ESFRI, 2020). These metrics have not been effectively identified in previous efficiency evaluations at the collective level, so this work complements the missing perspective of prior studies and provides a reference for the resource management of scientific instruments.

To summarize, use-time is not just a record of technical service but serves as an active medium of interorganizational coordination and dynamic adaptation. It encodes flow characteristics, both ordered and redundant, that are vital for evaluating system-wide efficiency. Based on this perspective, the following hypothesis is proposed:

Hypothesis 2 (H2): Instrument use-time, when modeled as systemic flows across organizations, exhibits nonlinear trade-offs between utilization intensity or out-put scale and system resilience.

Ecological efficiency and the ascendency analytical framework

Ascendency analysis evaluates the sustainability of flow networks by quantifying two properties: (1) Ascendency (A), the system’s organized, efficient flows, calculated by total system throughout (TST, total annual use-time) and average mutual information (AMI), detailed in the Method section; and (2) Redundancy (φ), which is the system’s unorganized, flexible flows, representing its capacity to absorb shocks (Ulanowicz et al. 2009). In this analytical approach, both ascendancy and redundancy are non-conserved, system-level variables, representing the system’s self-organizing effectiveness and its stability, respectively. Together, they define the system’s overall development capabilities (C) (Zisopoulos et al. 2022). The interaction between A and C, expressed through the degree of order (α = A / C), serve as the sole parameter for calculating the ecological efficiency of the system.

Prior applications of the ascendency analysis approach in areas such as water resource utilization, economic systems, global commodity trade networks, nutrient flows related to food security, and circular urban waste management have demonstrated its effectiveness in quantifying systemic stability and growth (Li and Yang, 2011; Huang and Ulanowicz, 2014; Kharrazi et al. 2017; Liang et al. 2020; Zisopoulos et al. 2022). This approach, grounded in Boltzmann information theory and the concept of Shannon entropy, enables the evaluation of economic system sustainability from the perspectives of efficiency, resilience, and robustness (Huang and Ulanowicz, 2014; Liang et al. 2018). Within this framework, any system can be abstracted as a network composed of interconnected nodes (e.g., industrial sectors or trophic levels), where the connections represent mutually constrained flows of a specific circulating medium (Ludovisi and Scharler, 2017; Ulanowicz, 2020).

On the other hand, multiple frameworks in sustainability science have attempted to model flow-based systems. For example, Emergy analysis offers a method to convert environmental inputs into a unified measure of embodied energy, capturing the hidden cost of sustaining economic activity (Geng et al. 2013; González-Mejía and Ma, 2017). Likewise, Material Flow Analysis (MFA) traces the movement of matter across human and ecological systems from extraction to disposal, emphasizing quantitative mass balance (Huang et al. 2012). In contract, ecological network analysis (ENA) focuses on interaction structure and systemic behavior, revealing network-level robustness through indirect effects and holistic organization (Huang and Ulanowicz, 2014; Fath and Scharler, 2019). As mentioned before, scientific instrument machine-hour and its sharing component have carried unrevealed structural information. By framing use-time as a flow within a network (akin to Emergy flows in ecosystems), ascendency analysis can quantify how efficiently these flows are organized, thereby addressing gaps in linear economy models and modern management frameworks.

Ultimately, the ascendency analysis approach provides a lens to differentiate growth (increasing scale or capacity) from development (enhanced structural efficiency). It offers an indicator of ecological efficiency for measuring Open Sharing status of large-scale scientific instruments. In this abstract flow network representing an actual organization, each node (such as platform, department, or institution) exchanges its scientific instrument use-time as a constrained resource with other nodes. Efficiency describes the degree to which these flows are streamlined, while resilience denotes the diversity and adaptability of the system’s linkages (Fath et al. 2019a). This leads to another core research hypothesis:

Hypothesis 3 (H3): Scientific instrument use-time networks that exhibit higher ecological efficiency (measured by α) are more likely to sustain balanced development, whereas those driven primarily by throughput expansion tend to exhibit structural fragility.

Data and methods

Data collection

The data used in this study were mainly collected from the annual reports of each large-scale regional center organization of Chinese Academy of Sciences (CAS), as well as the websites of Ministry of Science and Technology (MOST) for the national-level data on instrument scale and utilization time, and from regional science and technology investment data in the China Statistical Yearbook. To ensure transparency, the sample selection followed three criteria: (1) scientific instrument organizations must have operated for at least five years (2018–2022); (2) annual use-time data for at least 80% of open-sharing instruments must be reported in detail; and (3) regional coverage must include both eastern and western provinces to mitigate geographical bias. Refer to Appendix A for background information describing the representativeness and reference of the data. Sporadic missing data were supplemented by targeted platform inquiries, and temporal continuity was validated by Spearman’s rank correlation. Statistical surveys on the distribution of use-time were subsequently conducted based on the 2019 annual reports of these 15 organizations, and interviews were carried out in 2023 with five of them. In order to facilitate the retrieval of relevant instrument use-time data, this study was authorized by the Scientific Research Management Committee of the supporting unit. All respondents were proactively informed that their responses would be used solely for research reporting purposes.

For the application of the ecological network analysis (ENA) approach, a separate use-time flow network was constructed for each organization and each year. Co-built by five-member management units, the 2018 and 2022 use-time flow networks of the Lanzhou (LZ) instrument-sharing regional organization can be abstracted into Fig. 1. The dataset for each input and output arrow is presented in Tables A1 and A2 in Appendix A. The use-time structures in 2018 for the remaining 14 regional centers were similarly constructed from their respective data. Given the complexity of the underlying mechanisms governing instrument resource operations, several instrument platforms have been integrated to cooperate and form a regional organization, in order to evaluate the theoretical value of the degree of order (α). Due to the lack of use-time data from external sources beyond the studied system, the impact of instrument use-time input from outside the regional network was not considered in this study. Ten-year statistical data on instrumental use-time (2013–2022) were only available for the LZ regional organization. Fortunately, the LZ regional center not only pioneered open-sharing operations in 2012, but also ranked among first class by the MOST’s Open Sharing assessment system for large-scale instruments at the early period. Based on a realistic comparison with the other of regional organization in 2018, this study therefore considers the LZ organization as a representative case for assessing the overall evolution of Open Sharing practices for large-scale scientific instruments in China.

Fig. 1: Use-time flow network among the LZ regional organization of five-member units in 2018 and 2022 (Unit: hours).

Arrows originating from one box to another indicate instrument use-time flows among members; arrows without endpoints represent outputs to other systems, and those looping back to the same box donate internal utilization.

Ascendency analysis

As mentioned above, the methodology of ascendency analysis has been well documented in previous literature (Ulanowicz et al. 2009; Liang et al. 2018; Zisopoulos et al. 2022). Accordingly, the instrumental use-time data obtained through statistical surveys from different member units of the fifteen regional organizations were abstracted into yearly networks of nodes and links, as illustrated in Fig. 1. Each network information was then encoded into a matrix, where rows and columns represent the use-time provided and demand time of each member unit, respectively. These matrices served as the basis for calculating subsequent indicators in a stepwise manner, applied to each regional instrument organization across multiple years.

In ascendency analysis, the first indicator related to the system’s scale and growth is total system throughput (TST), which is calculated by summing the total activity within the network in terms of its flows. Ascendency metrics are robust to non-linear relationships due to their entropy-based formulation (Ulanowicz, 2020), making them suitable for analyzing complex flow networks. The units of these metrics vary depending on the type of circulating mediums, such as material, energy or information. In this study, the unit of TST is hours per year, and it is computed using Eq. (1).

$${TST}={\sum }_{i,j}{T}_{{ij}}$$

(1)

Where T_ij represents the intermediary flow between nodes i and j.

Development is defined as the improvement of the organization or/and the structural configuration of the system, which is independent of system scale and is reflected in reduced uncertainty. Then, the average mutual information (AMI) is calculated to describe the organized part of the flows, namely organizational efficiency, and is measured in bits using Eq. (2).

$${AMI}={\sum }_{i,j}\left({T}_{i,j}/{TST}\right){\log }_{2}\left(\left({T}_{i,j}* {TST}\right)/\left({T}_{i}* {T}_{j}\right)\right)$$

(2)

Where T_ij /TST is used to capture the mutual information of i and j, which means the probabilities that nodes i and j provide about each other.

In use-time flow networks, the joint event ij representing a flow leaving compartment i and entering compartment j, has a probability associated with it in terms of information theory. This probability yields the diversity (H) of the system, as shown in Eq. (3).

$$H=-{\sum }_{i,j}\left({T}_{i,j}/{TST}\right){\log }_{2}\left({T}_{i,j}/{TST}\right)$$

(3)

H serves as an upper bound for AMI, since the convex nature of the logarithmic function guarantees that H ≥ AMI ≥ 0.

These two indicators (H and AMI) are scaled with TST to assign physical dimensions to the flow network (Ulanowicz et al. 2009). The resulting metrics are:

$$A={TST}* {AMI}={\sum }_{i,j}{T}_{i,j}{\log }_{2}\left(\left({T}_{i,j}* {TST}\right)/\left({T}_{i}* {T}_{j}\right)\right)$$

(4)

$$C={TST}* H=-{\sum }_{i,j}{T}_{i,j}{\log }_{2}\left({T}_{i,j}/{TST}\right)$$

(5)

where A is the ascendency of the system, describing the efficiency with which it organizes and direct flows through its structure, while C is the system’s capacity for development at its current scale. Both indicators are measured in hours • bits per year. The degree of order α is a unitless indicator, defined as the ratio of the organized portion of the system to its total capacity for development. It reflects whether the system is evolving towards greater efficiency (e.g., α > 0.7) or favoring flexibility and redundancy (α < 0.3):

$$\alpha =A/C$$

(6)

The system’s overhead φ (also known as flexibility or redundancy) can then be calculated as:

$$\varphi =C-A=\,-{\sum }_{i,j}{T}_{i,j}{\log }_{2}\left({{T}_{i,j}}^{2}/\left({T}_{i}* {T}_{j}\right)\right)$$

(7)

Finally, the theoretical robustness (R) of the use-time network, also a unitless measure, is computed as a function of the degree of order (α)) using Eq. (8):

$$R=-\alpha * {\mathrm{ln}}\,\alpha$$

(8)

Statistics exploration

Given the limited number of data points across regional organizations for both instrument use-time and degree of order, a Chi-square test was conducted to explore the potential relationship between these two variables. To satisfy the assumption of expected frequencies of at least 5, the continuous variables (average use-time, degree of order) were dichotomized into low and high categories based on their respective median values. A Fisher’s exact test was also employed as a robustness check for small sample sizes. In addition, bootstrapping resampling was performed to further validate the stability of the findings given the scant sample size (N = 28). To account for possible nonlinear associations and ordinal nature of the data, Spearman’s rank correlation was used in place of Pearson’s correlation.

In addition to examining bivariate relationships, this study incorporated regression analysis to explore how organizational or regional factors may influence ecological efficiency, as measured by the degree of order (α). Based on data availability, a multivariate linear regression model was constructed, with α as the dependent variable and per capita GDP, annual R&D expenditure, and a policy support index (i.e., number and issuance year of open-sharing policies) as independent variables. Given the potential multicollinearity among variables, especially between R&D expenditure, policy count, and regional economic indicators, collinearity diagnostics were performed. As a result, only two relatively independent predictors (GDP per capita and policy timeliness) were retained in the final model.

To examine how the degree of order relates to the overall sustainability of the use-time network, this study applies a decomposition model that separates the contribution of system scale (growth) and network organization (development). In this model, any compound variable V (e.g., total system throughput) can be represented as the product of two factors, V = x × y. Based on this framework, the relative contributions of scientific instrument scale and average use-time to total throughput are calculated. In turn, the influence of total system throughput and AMI on ascendency is evaluated. These relationships help to identify whether observed increases in efficiency are driven by structural optimization or simple expansion.

Additionally, the robustness curve of the system was constructed to analyze whether the observed degree of order aligns with theoretical expectations for sustainable operation. A degree of order of approximately 0.4 is the quantized optimum value towards which natural ecosystems have been reported to converge in previous studies, with the corresponding maximum robustness value of 0.368 (Ulanowicz, 2009; Ulanowicz, 2020). Following the method of previous studies, a simulation experiment was designed using Python to randomly generate node-link structures representing regional organizations. In each simulation, 2 to 9 member units were sampled to form a network from the statistical pool of 15 regional organizations, which altogether consist of over 70 instrument management units. Approximately 5% of the total use-time was assigned as internal sharing time, with a 5% probability assigned to each potential sharing connections. This process was repeated 10,000 times, resulting a distribution of AMI and TST values, from which the theoretical bounds of ecological efficiency were estimated. The geometric mean of the maximum and minimum values was taken as the boundary of the efficiency window, which was used to determine the optimal α value (degree of order). The final step involved calculating the most likely values of the parameter β, which was used to calibrate the robustness curve and define the efficiency window of the studied networks, as shown in Eq. (9).

$${R}_{{adj}}=-\left({\alpha }^{\beta }\right)* \mathrm{ln}\left({\alpha }^{\beta }\right)$$

(9)

Results

Interannual trends of use-time in the LZ regional organization

Figure 2 presents the annual dynamics of instrument use-time and flow-network indicators for the LZ regional organization over the period from 2013 to 2022. The quantitative data were derived from official operational reports and transformed into following flow-network metrics: total annual use-time (total system throughput, TST), average use-time (AUT), average sharing time (AST), average mutual information (AMI), ascendency (A), and the degree of order (α).

Fig. 2: Annual statistical use-time data and the flow-network indicators of the LZ regional organization.

a TST equals to annual total use-time; b AUT means annual average use-time, and AST is annual average sharing time; c H indicates system diversity in the sense of information theory; and d A stands for ascendency, while TO represents the theoretical level of degree of order α.

As shown in Fig. 2a, both scale-related metrics of the instrument-sharing platform in LZ increased gradually over the 10-year period. Notably, the total annual use-time (TST) rose steadily, with a marked acceleration after 2017. The upward trend was primarily attributed to the expansion in the number of scientific instrument sets, while average use-time showed noticeable fluctuations (see Fig. 2(b)). According to the decomposition analysis presented in Table 1, the contribution of network scale to TST was more stable and consistently positive (75.5% over 10 years), whereas the contribution of average use-time showed greater volatility and, in some years, had a negative impact (24.5% over 10 years). These findings suggest that improvements in total service output were mainly driven by the expansion of instrument resources, rather than by enhanced utilization efficiency of existing equipment.

Table 1 Contributions of TST and AMI to ascendency levels and the scale and average use-time to total use-time (TST) in LZ regional organization.

In addition to throughput growth, the capacity (measured by staff numbers) of technical personnel is a critical determinant of network effectiveness. Data from the LZ regional organization show that while total use-time more than doubled from 2013 to 2022, the number of technical staff increased only modestly, from 99 to 210, with periods of stagnation or even decline observed between 2015 and 2017. The compound annual growth rate of staff numbers over the decade was approximately 8.4%, considerably lower than the growth rate of total instrument throughput. This widening disparity suggests that labor inputs have not scaled in proportion to the increasing workload, leading to coordination bottlenecks and constraints on network efficiency. This structural imbalance may partly explain the decline in AMI values and the limited gains in the degree of order (α), despite continued investment and scale expansion. Given that the effective operation of shared infrastructure depends not only on equipment availability but also on human coordination, aligning technical human capital with throughput expansion remains a key priority for sustainable development.

On the other hand, annual variations in AMI values closely paralleled changes in the degree of order (α). As shown in Fig. 2c, AMI remained relatively high during the early period (2013–2017), reflecting a stable and coherent use-time structure among units. However, starting in 2019, AMI began to decline, with values in 2021 and 2022 falling below earlier levels. This trend indicates a decrease in the efficiency of information transmission and reduced structural coherence within the use-time network, even as total throughput continued to increase. The decomposition analysis in Table 1 further supports this interpretation. While TST contributed positively to ascendency (117% of the overall contribution), AMI made a negative contribution over the ten-year period (−17.4%), with particularly detrimental effects observed from 2019 to 2021. These findings suggest that growth in scale and throughput was accompanied by a breakdown in structural coordination and mutual interaction, ultimately limiting improvements in ecological efficiency. Meanwhile, the number of flows showed no consistent correlation with AMI trends, fluctuating between a minimum of 6 to a peak of 22 (with 25 being the theoretical maximum for the five-member alliance of LZ). Paradoxically, the highest number of active flows (connections) coincided with the lowest AMI value during peak throughput periods. This implies that increased fragmentation and small-scale shared links may not only fail to enhance information transmission efficiency at the organizational level, but could in fact undermine systemic performance through emergent bottleneck and coordination inefficiencies.

Figure 2d illustrates the temporal evolution of ascendency and the degree of order (α). Overall, ascendency increased steadily over the study period, with the exception of 2020, which experienced a sharp decline. This drop corresponds with the operational disruptions caused by the COVID-19 outbreak. According to reported data, average use-time fell by 6.48% compared to 2019, echoing the national trend of an 11.8% decrease reported in MOST’s Open Sharing assessment (see Appendix Fig. A1). By 2022, ascendency had doubled compared to its 2013 baseline, indicating a substantial enhancement in system organization. However, despite this upward trend in ascendency, the degree of order (α) showed considerable fluctuations, particularly declining after 2019 toward the theoretical optimal threshold, as represented by the green dashed line in Fig. 2d. In most other years, α remained above this optimal line, suggesting that LZ’s system configuration was plausibly efficient. Nevertheless, α dipped below the critical threshold in 2020, which might suggest deeper systemic vulnerabilities that were obscured by surface-level gains in scale efficiency.

In summary, these findings lend support to H1, demonstrating that the increase in ascendency was primarily driven by scale expansion rather than coordination efficiency. The fluctuating α values and declining AMI trends also support H2 and H3, confirming that instrument use-time flows were structurally fragile when network coherence failed to keep pace with throughput growth.

Regional heterogeneity and drivers

In 2018, there were 15 regional scientific instrument organizations under the CAS distributed nationwide, with seven located in tier-1 cities such as Beijing (BJ), Shanghai (SH), and Guangzhou (GZ). Figure 3 displays the cross-sectional distribution of the degree of order (α) and log-transformed total use-time for each organization, enabling visual comparison of network efficiency and operational scale across regions. At first glance, this observed spatial disparities align with broader national trends in economic development and R&D resource allocation, wherein eastern provinces (e.g., BJ, SH) have historically received more scientific funding and attracted a greater concentration of skilled technical personnel (Wang et al. 2021a; Li et al. 2024). However, results from Kruskal–Wallis tests reveal that differences in total use-time across regions are not statistically significant (Chi-square = 3.01, p = 0.083). This suggests a possible moderating effect of the dual-center governance model ((i.e., CAS region-level and institute-level oversight), which may standardize operational protocols and reduce regional disparities in scientific instrument utilization. This finding supports the rationale for further examining whether efficiency gaps are more attributable to organizational or disciplinary factors. In contrast, tier-1 cities exhibit significantly higher α values (F = 5.67, p = 0.028), indicating superior structural coordination and organizational efficiency in managing shared scientific resources.

Fig. 3

Degree of order (α) and log-transformed total use-time across regional organizations.

To investigate the drivers of regional disparities in ecological efficiency, regression analysis was conducted using the degree of order (α) as the dependent variable. Independent variables included per capita GDP, annual R&D expenditure of local institutions, and a policy support index (comprising the number and starting year of open-sharing policies implemented between 2014 and 2022). Details on data sources and variable definitions are provided in Appendix A. The regression results show that economic capacity (b = 0.45, p = 0.034) and policy timeliness (b = 0.50, p = 0.020) have statistically significant positive effects on α. The overall model is significant (F = 11.21, p = 0.002). Other potential predictors, such as annual R&D expenditure, projects counts, and policy issuance frequency, exhibited high multi-collinearity with region and GDP (correlation coefficients ≥0.65) and were therefore excluded from the final specification. Interaction analysis revealed that the effect of region and policy timeliness is more pronounced in tier-1 cities, suggesting heterogeneous responsiveness depending on regional governance structures. By 2022, most organizations, except for Xinjiang (XJ), had seen their α values converge toward 0.5. XJ’s upward trajectory likely reflects later-stage policy uptake combined with increased regional investment. In contrast, the tier-1 cities such as BJ, SH, and GZ showed declining α values despite consistently high use-time, potentially reflecting growing local self-sufficiency, limited cross-regional coordination, and internal competition, trends corroborated by declining sharing rates (see Table A3). Overall, these findings support H1, H2, and H3 by demonstrating that differences in α are not solely explained by total system throughput. Rather, ecological efficiency appears to depend critically on structural coordination, governance maturity, and policy implementation dynamics, beyond mere platform scale.

On the other hand, organizations positioned further to the right along the horizontal axis in Fig. 3 represent those with higher total use-time. These organizations are predominantly affiliated with the geosciences and physical sciences, whereas those associated with the life sciences generally display lower utilization levels. However, substantial internal variation is observed even within the geosciences. Among the selected organizations, despite increases in average use-time, four out of five exhibited declining α values, suggesting potential trade-offs between utilization and systemic efficiency. This finding implies that disciplinary affiliation alone does not fully account for efficiency outcomes. The observed discrepancy between high utilization and low α values may reflect a structural imbalance between the expansion of scientific instrument scale and the capacity for coordinated use, supporting H1. Accordingly, policy frameworks should incorporate such heterogeneity by adopting differentiated support strategies. These may include targeted incentives for cross-institutional sharing, programs to enhance technical coordination capacity, and mechanisms to promote talent mobility within and across disciplines.

Efficiency window and optimal range

To simplify the concept of sustainability, a system can be considered sustainable if it maintains a dynamic balance between efficiency and flexibility (Ulanowicz et al. 2009). This balance is inherently context-dependent. For instance, natural ecosystems tend to prioritize resilience (often clustering around α ≈ 0.4), whereas industrial systems typically emphasize efficiency (α > 0.6) due to centralized governance structures (Fath et al. 2019a). Therefore, systems pursuing sustainability may occupy different positions along the α continuum from 0 to 1 (Ulanowicz, 2020), depending on their structural and functional priorities. In the context of China, the observed dominance of higher efficiency (α ≈ 0.5–0.7) reflects a policy environment shaped by centralized mandates that prioritize quantifiable performance indicators (e.g., total or average use-time) over adaptive capacity or redundancy. Similar trade-offs between efficiency and resilience have also been documented in other decision management systems (Zou et al. 2025). Building on these insights, this study proposes the concept of a theoretical efficiency window, within which the use-time flow networks of regional scientific instrument organizations are expected to operate over time. This window represents a feasible operational range that reflects a system’s capacity to balances growth and resilience, with degree of order (α) serving as the core indicator of ecological efficiency.

To derive the efficiency window, the degree of order (α) was calculated annually for the use-time network of the LZ regional organization. As shown in Fig. 4a, α values are plotted against their corresponding theoretical robustness (R), generating a curve that captures the evolving resilience potential of the system over time. Notably, this curve indicates a temporary disruption of systemic coordination, which may reflect the impact of external shocks; further interpretation is provided in the Discussion section. Drawing upon the ecological boundary framework, the geometric center of the vitality window observed in natural ecosystems is defined by the parameters n = 3.25 and c = 1.25. These correspond to a benchmark α = 0.4596 and a robustness scaling parameter β = 1.288 (Ulanowicz et al. 2009). In the present analysis, all annual α values except that of 2020 fall within or near the high-organization zone. However, not all observed values achieve the theoretically expected level of robustness, suggesting that systemic efficiency and resilience may not always align in practice.

Fig. 4: Corrected robustness curves constructed based on the window of efficiency derived from the 2018 use-time flow networks of members from all fifteen regional organizations.

a LZ represents the use-time data from LZ regional instrument organization. b BJ is one of the four regional organizations on geoscience, SH on physical science, GZ on life science, and LZ and XJ both are also categorized under geosciences.

The structure characteristics of use-time flow networks within scientific instrument-sharing systems differ markedly from those observed in natural ecosystems. Instead, they resemble patterns found in economic or urban water metabolic systems (Huang and Ulanowicz, 2014; Zou et al. 2025). To simulate these dynamics, nodes (n, i.e. member units) were randomly selected in the range of 2 to 9, and sharing links (c, i.e. flows) were assigned probabilistically to estimate real-world sharing patterns of instrument-sharing behavior. The resulting estimated values of n = 6.5632 (max) and n = 1.1944 (min), exceed the typical structural limits of trophic hierarchies in natural ecosystems, underscoring a higher degree of complexity inherent in these human-managed systems, which are driven by performance-based goals. Similarly, link density (c) ranged from 1.0243 to 1.7622, indicating that even randomly generated organizational networks tend to exhibit loose connectivity. This structural configuration implies a fragmented and heterogeneous system, where individual sharing behaviors fail to coalesce into a tightly integrated or highly coordinated networks.

Figure 4b presents the updated robustness curve for the chosen CAS system, constructed using benchmark data from all 15 regional organizations in 2018. The estimated parameter β = 2.2057 yields a new optimal degree of order, α = 0.6355, which corresponds to the most robust system configuration under current operating conditions. Compared to the ecological benchmark (α ≈ 0.40), this upward shift reflects a systemic prioritization of operational efficiency over resilience, consistent with the centralized performance-driven mandates governing the CAS framework. Among the selected regions, only the early-phase LZ organization approaches this newly defined optimum, particularly before structural redundancy adjustments were implemented. In contrast, the other three organizations exhibit slightly diminished robustness, with the XJ regional organization still progressing towards the optimal state. The BJ regional organization demonstrates relatively stable operation, with limited deviation from the modeled robustness curve. Hence, the alignment between the empirical α values and the theoretical robustness envelope lends support to H2, validating the application of flow-network theory to instrument-sharing systems and confirming the existence of definable efficiency-resilience trade-off zones. Simultaneously, the deviation of high-throughput years (characterized by elevated TST) from this optimal window further provides evidence in support of H3, demonstrating that throughput expansion without structural coherence leads to compromised robustness and systemic fragility.

Relationships between use-time and ecological efficiency

To examine the relationship between scientific instrument use and ecological efficiency, a Chi-square test was conducted on the relationship between annual average use-time and α values (N = 28). The test met statistical validity assumptions, with all expected frequencies in the contingency matrix ≥5. The result revealed a significant association between average use-time and degree of order (α) (Chi-square = 5.143, p = 0.023), with a Cohen’s W = 0.43, indicating a moderately strong correlation. This result challenges prevailing assumptions that increasing usage inherently enhances operational efficiency, as suggested in studies of scale economies in shared infrastructure (e.g., ESFRI, 2020). It implies that Chinese instrument-sharing systems may operate under distinct structural constraints, such as regional redundancies or coordination inefficiencies (Xu et al. 2022).

To further analyze relationships among key variables, Spearman’s rank correlation test was applied to the full set of surveyed regional organizations. Figure 5 presents a correlation heatmap, revealing several strongly positive relationships: between system scale and total use-time; between total use-time and sharing time; and between average use-time and average sharing time (sharing rate). These results confirm that resource expansion contributes directly to service output, and that active sharing is positively associated with increased technical usage across sites. However, a notable negative correlation was identified between platform scale and sharing rate. This means that as scientific instrument resources expand, their marginal sharing ratio declines, suggesting a potential scale-sharing paradox. This paradox echoes findings in public goods governance, where resource abundance may suppress collective coordination and mutual reliance (Ostrom, 2010). From a management perspective, this dynamic may reflect crowding effects, a lack of cross-institutional incentives, or mismatches in demand scheduling.

Fig. 5

Correlation heatmap of use-time flow networks based on data from fifteen regional organizations (N = 28).

Within the ecological network analysis (ENA) framework, system development capacity (C) is defined as the sum of ascendency (A) and redundancy (φ). As expected from their mathematical relationships, these three indicators are highly collinear and jointly represent the expansion potential of the scientific instrument-sharing network. This result indicates that the overall instrument management system is still in a formative stage. A strong correlation is also observed between ascendency and the Average Mutual Information (AMI), consistent with the fact that AMI, together with total use-time (TST), constitutes the two core components of ascendency. Furthermore, the degree of order (α) shows a significant positive correlation with theoretical robustness (R), reaffirming that most regional systems remain positioned on the left side of the unimodal robustness curve introduced in Fig. 4. In general, total use-time and sharing time are positively associated with all three ENA indicators (A, φ, C), further confirming that the platform remains in a development-driven phase characterized by increasing scale and connectivity.

Despite the general trends of growth and positive correlation, three variable pairs display unexpected negative relationships. First, average use-time negatively correlates with the degree of order (α), meaning that more intensive utilization does not necessarily translate into higher ecological efficiency. This finding directly supports H3, highlighting a trade-off between output-driven sharing and systemic coordination. Second, sharing time also shows a weak or even inverse relationship with α, challenging to the common assumption that increased sharing inherently enhances system performance. Together, these results call into question simplistic notions of output maximization and point to potential structural trade-offs embedded within current sharing practices. Several mechanisms may help explain this paradox:

• 1. Monopoly effect: instrument access may be disproportionately controlled by dominant institutions, limiting equitable usage opportunities for smaller or emerging participants.

• 2. Redundancy saturation: repeated or routine usage may indicate service duplication with diminishing scientific returns.

• 3. Inefficient scheduling: instruments may be heavily booked but remain underutilized in practice, especially for low-priority or non-critical tasks.

These distortions can undermine the intended benefits of shared infrastructure by reducing system flexibility and thereby lowering the degree of order (α). Such patterns highlight the need for more dynamic and responsive coordination models (such as quota-based allocation, cost-informed reservation systems, or adaptive sharing thresholds) that can better balance usage intensity with systemic efficiency.

Collectively, the results presented above provide empirical support for the three hypotheses proposed in the conceptual framework. First, H1 is validated by the observation that platform expansion increases overall throughput but does not necessarily translate to improved network efficiency. Second, H2 is confirmed by the identification of use-time as a measurable and responsive systemic flow, with corresponding variations in the degree of order (α) and robustness curves effectively capturing dynamic shifts in system stability. Third, H3 is substantiated by multiple indicators, including declining AMI contributions, α trends, and the observed paradoxes between sharing and utilization, which demonstrate that output-oriented growth may compromise structural resilience in the absence of coherent network organization.

Discussion

Conceptual implications

At the core of this study are several key contributions that shed light on the mechanisms and measurement frameworks governing scientific instrument Open Sharing in China. A persistent challenge in the management of scientific instrument platforms lies in balancing efficiency gains with broader goals of systemic resilience and equitable access (Wang et al. 2021b). While ecological efficiency emphasizes a system’s ability to self-organize and absorb perturbations, current evaluation mechanisms in China continue to rely predominantly on linear metrics such as total or average use-time (Wang et al. 2024). This mismatch may inadvertently create distorted incentives, prompting platform managers prioritize throughput over coordinated development or service equity. Such tendencies risk generating localized overloads, monopolization of resources, or even systemic stagnation, especially in regions with limited institutional capacity.

These systemic tensions are illustrated in Table 2, which summarizes the trade-offs across different efficiency regimes. Efficiency-driven systems, often found in eastern regions such as BJ and SH, tend to maximize output through centralized evaluation and use-time quotas. In contrast, resource-driven organizations, more prevalent in western regions, prioritize basic availability and equitable access, but often lack strong performance accountability. Both types are constrained by rigid evaluation framework that primarily evaluate performance based on aggregate throughput, rather than meaningful coordination or equitable sharing outcomes. For instance, centralized metrics may reward high total use-time while ignoring distortions like internal monopolization or ineffective cross-unit flows. Absent adaptive indicators and flexible policy levers, such as flow-based benchmarking, tiered quotas, or dynamic reservation systems, these regimes risk becoming locked into unsustainable incentive patterns. Conceptually, this reflects a broader governance paradox: linear metrics fail to capture the complex, networked, and dynamic nature of shared infrastructure performance.

Table 2 Key trade-offs in scientific instrument-sharing management systems.

Beyond use-time metrics and regional disparities, staffing limitations and evaluation standards constitute critical institutional bottlenecks in instrument-sharing systems (Xiong et al. 2023). As shown in Table 2, current regional organizations with different operating orientations all face human resource issues. Nevertheless, prevailing policy levers continue to emphasize performance outputs, while neglecting structural indicators such as the technician-to-instrument ratio. Despite longstanding recognition of the essential role played by technical support teams in enabling effective open-sharing platforms (Yu, 2022), substantive implementation efforts remain lacking. Consequently, this oversight reveals a form of governance myopia: a systematic omission of human capital investments from institutional performance frameworks, which undermines the development of robust scientific and technological capacity. When technical labor, such as maintenance, calibration, cross-department scheduling, and user assistance, is treated as marginal, a parasitic infrastructure dynamic can emerge (Corvellec et al. 2022), wherein physical expansion of resources occurs alongside a steady erosion of supporting technological resilience.

In regional organizations like LZ, this imbalance is particularly pronounced. Over the past decade, both instrument scale and throughput have more than doubled, yet the growth of the technical workforce has lagged significantly behind, as shown in Fig. 2. This disparity has led to a persistent misalignment between user demand and service capacity. As a result, the shortage of personnel to manage reservations or coordinate cross-institutional access has compelled platforms to internalize usage and engage in output inflation practices (e.g., prolonged booking, underutilized slots, and even phantom operation) (Liu et al. 2019; Wang et al. 2021a). Conceptually, this highlights a novel form of resource mismatch between expanding physical assets and stagnant human capital. Because current performance indicators fail to capture such mismatch, there is an urgent need for ecologically grounded, labor-inclusive efficiency metrics to more accurately evaluate the real effectiveness of open-sharing scientific instrument systems.

Fortunately, ecological efficiency metrics such as the degree of order (α) offer a system-level alternative to traditional output-based indicators, enabling platforms to monitor both internal organizational structure and external service performance within a unified analytical framework. In this context, ecological network analysis (ENA) allows managers to distinguish between platform growth and systemic development, offering not only a diagnosis of potential fragility but also a basis for performance benchmarking and strategic adjustments (Zisopoulos et al. 2023). When interpreted dynamically, α and AMI reveal whether increases in output stem from enhanced coordination or are merely the result of scale expansion (Fath et al. 2019b). The metaphor of eutrophication in an ecosystem remains instructive. Abundant but unstructured flows, such as increased resource allocation or diversified quota chasing, can signal organizational regression rather than progress. By integrating flow-based ecological indicators, this study complements existing evaluation systems and offers new tools to detect latent inefficiencies that traditional throughput metrics may fail to capture.

Institutional dynamics and evolution of scientific instrument-sharing systems

The evolution of scientific instrument-sharing systems can be analyzed through the framework of ascendency theory, which distinguishes four evolutionary phases: growth, development, maturity, and overload (Ulanowicz, 1997). During the initial growth phase, abundant resource input and minimal coordination demands drive rapid increases in total system throughput (TST). China’s current scientific instrument-sharing landscape, characterized by continued infrastructure expansion and investment, remains largely situated in this first phase. However, evidence of high instrument utilization efficiency in regional alliances such as BJ and SH suggests that certain subsystems may be transitioning toward the development phase, marked by intensified internal connectivity and increasing system tension. Conversely, alliances exemplified by the LZ regional organization exhibit signs of regression through growing redundancy. Rather than facilitating a progression toward Phase Two, ongoing resource inflows in these cases appear to exacerbate organizational inefficiency, reflecting a stagnation or even reversal in evolutionary trajectory under existing institutional management frameworks.

A fundamental mechanism underlying this systemic tension is the persistent mismatch between resource allocation and actual research needs. When instrument-supporting units prioritize hitting quantitative sharing targets, such as the MOST’s 230-h benchmark, they may allocate instruments to low-impact or non-specialist users, thereby disrupting domain-specific workflows (e.g., allocating geoscience instruments to non-specialist projects). This phenomenon resembles the “tragedy of the commons,” wherein output-maximizing mandates degrade both the quality and intended function of resource use. Although ongoing investment contributes to system-level resilience through increased capacity and redundancy, this same growth can undermine organizational coordination and ecological efficiency (Fath et al. 2019a). As illustrated in Figs. 3 and 5, the observed sharing paradox partly arises from this top-down expansion logic, whereby additional capacity introduces structural fragility and reduces systemic order (α). Nonetheless, such risks may be considered acceptable in the Chinese context, given the state’s long-term strategy to renew depreciating instrument stocks across research institutions. This is further supported by the availability of a robust technical talent and recent policy shifts, including revised evaluation measures that emphasize the role of technical personnel and require the documentation of exemplary service cases.

Conversely, the use of instrument use-time as a dominant performance indicator represents a pragmatic administrative compromise. It offers a quantifiable and comparable output metric that enables cross-institutional comparisons among heterogeneous research entities (Alkaabneh et al. 2021). However, the findings of this study show that maximizing use-time alone does not ensure sustainable ecological efficiency. Quota-oriented practices may incentivize the monopolization of high-demand instruments, known as policy-induced hoarding, which restricts access for smaller or peripheral research groups. An illustrative case occurred within a member institution of the LZ regional organization in 2016, when a recently merged unit allocated over half of its annual use-time to a few state-owned enterprises, nominally undermining equity and access across the whole platform. More broadly, linearly accumulated use-time is often recognized only at the end of the innovation cycle, whereas the managerial burden to achieve those machine-hours is immediate and inflexible. This misalignment generates skewed incentives, encouraging service providers to prioritize time accumulation over scientific impact or collaborative inclusion.

To address these systemic distortions, it is essential to develop meso-level governance structures, such as CAS’s regional alliances. Within such frameworks, localized improvements at the member level can be strategically aligned to raise system-level performance, particularly through resource mechanisms of complementation and user recombination strategies (Figge et al. 2021a). Drawing on growth pole theory, centralized instrument-sharing centers may serve as attractors that integrate smaller units, thereby improving collective efficiency and curbing unnecessary infrastructure expansion. In parallel, the adoption of shared rule-making processes rooted in social choice theory offers a pathway to performance enhancement without exacerbating distributional inequities (Goetz et al. 2017). Empirical findings from this study further suggest that geographical proximity and regional concentration enable policy spillover, which in turn stabilize both sharing intensity and performance variance (Yang et al. 2024). Thus, ecological efficiency assessments should expand to include indicators of network centrality, inter-unit coordination density, and alignment across institutional actors, thereby improving both diagnostic accuracy and policy usability.

From an ecological perspective, scale constitutes a critical factor in interpreting infrastructure performance and resilience. The physical scale of economic systems, rather than their embedding ecosystem, becomes integral to understanding and unpacking ecological and economic dilemmas (Malghan, 2010). As shown in Table 1 and Fig. 5, TST is primarily driven by network size, whereas increases in ascendency (A) are more closely tied to the quality of internal coordination than to scale alone. The observed negative contributions from AMI indicate internal coordination bottlenecks, particularly in regional organizations like LZ. This finding mirrors a well-documented dilemma: large-scale nodes are often more productive (Rapposelli et al. 2023), but simultaneously diminish system resilience more easily if structural diversity and functional redundancy are not maintained. Contrary to prevailing assumptions that smaller or underperforming units should be eliminated, this study supports their deliberate retention, as such nodes contribute to systemic efficiency through diversity-enhanced stability and load-balancing capacity (Figge et al. 2021a). Hence, a sustainable scientific instrument-sharing network must strategically balance centralization and dispersion, maintaining both high-throughput nodes and peripheral adaptive links.

From Open Sharing to sharing prosperity

Natural ecosystems achieve long-term sustainability by maintaining a dynamic balance between efficiency and resilience, a condition described by Ulanowicz and colleagues as the “window of vitality” (Ulanowicz et al. 2009; Zisopoulos et al. 2022). In this study, the reconstructed robustness curve of China’s CAS regional use-time networks reveals a rightward shift from the natural window, prompting the conceptual introduction of an “efficiency window” to represent a new equilibrium more applicable to human-designed systems. Notably, the observed negative correlation between degree of order (α) and average sharing time indicates that interactions among internal member units alone may dilute systemic coherence, particularly when redundant or diffuse flows are prevalent. This is exemplified by the 2020 case in the LZ regional organization, where an emergency surge in internal sharing under pandemic-related disruptions caused α to fall below the ecological threshold. This event underscores a critical insight: not all sharing volume contribute positively to system robustness. Instead, ecological efficiency is contingent upon the presence of structured, goal-oriented interactions rather than mere activity volume.

While Open Sharing initiatives are designed to broaden access, the configuration of incentive structures and performance metrics may inadvertently reinforce systemic disparities. For example, larger instrument-sharing centers or their member units may engage in quota exchanges to meet mandated policy targets, inflating sharing rates without genuinely expanding accessibility, and leading to an outcome akin to soft monopolization (Philippon, 2019; Webster, 2021). This phenomenon resonates with Metcalfe’s Law, wherein the value of a network increases quadratically with its size, structurally disadvantaging low-capacity nodes and smaller institutions. Such dynamics may prompt a strategic shift from collective optimization toward Nash-equilibria-like behaviors, where institutions prioritize self-interest over collaborative efficiency (Golany et al. 2015; Figge et al. 2021b)). Concurrently, policy indicators appear to be losing traction: while national evaluations prominently emphasized sharing time during 2018–2019, references to this metric have declined so far, with the sharing volume at that time plateauing at approximately 230 h (Su et al. 2023; Xiang et al. 2024). This stagnation suggests diminishing relevance or legitimacy of the indicator, reflecting a possible adjustment in institutional consensus or policy alignment. In this context, persistent access barriers underscore the limitations of symbolic or encouragement-based governance. Achieving inclusive and equitable utilization of shared scientific infrastructure requires a transition toward enforceable fairness mechanisms and adaptive, real-time monitoring systems.

Sustainable governance of shared scientific resources requires resilient institutional architectures capable of managing access rights, ensuring equitable benefit distribution, and responding adaptively to systemic changes (Andrews et al. 2024). As shown in Fig. 6, a comprehensive policy should may include: (1) mandatory thresholds for openness and usage equity; (2) enforceable penalties for access restriction or quota hoarding; (3) dynamic evaluation frameworks that replace static performance benchmarks; and (4) the classification of instrument platforms by disciplinary specialization rather than generic multi-purpose mandates. These mechanisms can help maintain context-sensitive, equitable, and scientifically relevant sharing practices. Furthermore, the reallocation of idle resources and the re-integration of underutilized instruments must be institutionalized through standardized procedures rather than case-by-case decisions. Recurrent underutilization continues to stem from a structural mismatch between the physical availability of instruments and the capacity of technical staff, limiting both service coverage and the completion of innovation cycles (Xiong et al. 2023). Addressing this imbalance remains critical for ensuring long-term operational sustainability and scientific productivity.

Fig. 6

A policy-oriented toolkit for achieving shared prosperity in scientific instrument utilization.

Unlike natural resources, where conservation prolongs utility, scientific instruments are intrinsically scarce, depreciable, and often institutionally inaccessible. To promote their rational use, Chinese public agencies have adopted reward-penalty mechanisms, awarding approximately 675 million yuan to high-performing institutions and issuing dozens of warnings to underperformers between 2018 and 2022 (Su et al. 2023). While this framework has gained policy traction, its long-term legitimacy depends on the development of reciprocity-based compromise mechanisms, whereby stakeholders forgo maximal individual benefit in exchange for system-wide stability (Picavet, 2009). Scientific instrument use-time may function as one such compromise metric, not because it fully captures performance, but because it is observable, standardized, and institutionally entrenched. In contexts of intensified resource competition, the establishment of enforceable access rights becomes a prerequisite for sustainable cooperation (Andrews et al. 2024). Drawing from ecological management, the concept of maximum sustainable yield offers a useful analog: beyond a certain threshold, additional utilization depletes systemic resilience. Accordingly, while the open-sharing initiative is consistent with broader principles of the sharing economy, the pursuit of sharing prosperity remains an evolving endeavor, demanding continuous innovation in governance mechanisms, equity protections, and performance recalibration.

Limitations and future research directions

The first limitation of this study concerns the interpretation of theoretical robustness. While the robustness curve provides a valuable tool for assessing relative system resilience, its maximum point does not necessarily represent an optimal state, nor does it fully reflect the real-world complexity of organizational performance. This highlights a fundamental constraint of simulation-based robustness analysis: namely, that it often simplifies system dynamics and assumes behavioral uniformity, which may reduce representational fidelity (Ludovisi and Scharler, 2017). Industrial systems as one kind of man-made system may be inferior robustness than ecosystems in general, not only because their processes are overly constrained, but also because the network connections are too redundant (Morris et al. 2021). More broadly, the challenge of threshold determination, specifically, identifying what values of α or AMI correspond to “optimal” ecological efficiency, remains an unresolved theoretical issue in ecological network analysis (ENA) applications (Fath et al. 2019a). Despite these, trend-based monitoring of α values and their deviation from the identified efficiency window remains useful for monitoring systemic transitions in open-sharing governance (Zisopoulos et al. 2022).

Second, limitations in data granularity and availability present significant challenges for spatial-temporal comparisons. While statistical methods such as Chi-square tests can confirm associations, they are insufficient for uncovering mechanisms or establishing causal relationships at the organizational level. In many instances, provincial-level indicators are substituted for institutional-level data due to the absence of localized metrics (e.g., supporting project type, scientific instrument configuration). Furthermore, the organizational heterogeneity among regional instrument alliances complicates standardized comparative analysis. Variations in member unit count, disciplinary composition, and network centrality introduce structural differences that, if unadjusted, may bias interpretations. As a result, disparities in average use-time between organizations may reflect differences in policy scope, personnel capacity, or instrument specialization rather than purely operational performance. Moreover, scientific instruments with zero-reported use-time may fall outside formal evaluation mechanisms, distorting assessments of sharing effectiveness by altering the denominator in performance calculations. This issue reflects a broader limitation of the current use-time-based evaluation system at the institutional level, which requires refinement to more accurately capture ground-level operational realities.

Future studies on scientific instrument sharing systems should extend the current ascendancy framework by integrating more granular, facility-level data, including funding streams, user characteristics, and scientific outputs (Hallonsten, 2014; Qiao et al. 2016). Such refinements would enable a more precise understanding of the performance variations observed across regions and disciplines. Particular attention should also be paid to discipline-specific dynamics, as instrument use-time logic and sharing modalities can differ substantially between fields such as geosciences and life sciences. At present, limited access to data from non-CAS institutions or universities, cross-departmental usage flows, and regional policy heterogeneity pose significant barriers to generalizing results beyond the studied sample. Addressing these limitations calls for more inclusive, adaptive, and technologically supported research strategies. Specifically, future research should prioritize the following directions:

• 1. Cross-institutional and cross-disciplinary mapping of use-time flows, to uncover interaction patterns among departments and institutions that go beyond organizational silos.

• 2. Stakeholder-inclusive performance design, particularly involving emerging users and peripheral institutions, to ensure that evaluation frameworks accommodate diverse operational contexts.

• 3. Experimental testing of dynamic evaluation mechanisms, such as adaptive efficiency windows, tiered quotas, and rolling benchmarks, through longitudinal and comparative case studies.

• 4. Collaborative data infrastructures, including open-access, real-time data platforms, to enable interoperability, enhance system-wide monitoring, and reduce information asymmetries among users, funders, and policymakers.

These efforts should be complemented by multi-stakeholder collaboration in both data reporting and policy implementation. Importantly, no single indicator, such as use-time, can adequately capture sustainability or equity in instrument-sharing systems. Rather, evaluation frameworks must adopt context-sensitive toolkits that reflect ecological efficiency, administrative feasibility, and disciplinary diversity. Realizing such balance requires transparent, modular data infrastructures capable of supporting robust assessments without distorting institutional behavior or long-term systemic resilience.

Conclusion

This study applies ecological network analysis (ENA) to examine the sustainability and efficiency of scientific instrument-sharing systems in China, focusing on the evolution of the Chinese Academy of Sciences (CAS) regional alliances from 2013 to 2022. Empirical findings confirm all three hypotheses proposed in the conceptual framework. Regarding H1, results reveal a nonlinear and occasionally paradoxical relationship between increased use-time and ecological efficiency: higher sharing volume does not always correspond to a higher degree of order (α), suggesting existing structural inefficiencies in resource coordination. For H2, instrument use-time is effectively modeled as a systemic flow network, wherein network expansion enhances TST, but improvements in resilience and internal coherence (as captured by AMI) often lag behind. For H3, the analysis identifies an efficiency window, based on ENA-derived robustness curves, within which platform operations strike an optimal balance between efficiency and flexibility. Over the study period, CAS regional organizations showed increasing system ascendency but diverging performance in terms of α and AMI. These patterns indicate that observed gains in use-time have been driven more by instrument-scale expansion than improvement in coordination or management capacity. These findings underscore the importance of differentiating between output maximization and sustainable system optimization.

Based on the foregoing analysis, this study proposes three strategic priorities for enhancing the governance of scientific instrument-sharing systems. First, a shift from static to dynamic evaluation metrics is essential. The current reliance on aggregate or average use-time fail to capture systemic trade-offs and latent inefficiencies. Incorporating ecological indicators such as ascendency, redundancy, and degree of order (α) can provide a more accurate and real-time assessment. Second, demand-side coordination must be significantly strengthened. Effective governance should move beyond quota-driven mandates toward responsive sharing networks that integrate user needs, technical capacity, and human capital allocation. This calls for cross-unit collaboration mechanisms, adequate staffing strategies, and the integration of service workflows with broader research agendas. Third, enabling adaptive and inclusive sharing frameworks is critical to avoiding monopolization and resource misallocation. Institutional designs must proactively anticipate risks related to access inequality and instrumental overconcentration. Policy interventions, such as differentiated incentives, anti-monopoly regulations, and dynamic scheduling, can help ensure equitable distribution and system-wide efficiency. These priorities are especially pertinent in the Chinese context, where centralized resource management intersects with pronounced regional disparities and ambitious innovation objectives. Nonetheless, the methodological framework proposed in this study, particularly the use of ENA in performance evaluation, offers transferable insights for advancing infrastructure sustainability in other emerging economies.

This study contributes to the literature by extending ecological efficiency principles, originally formulated for natural ecosystems, into the realm of scientific resource governance. By introducing the degree of order (α) as a dynamic performance indicator, it offers a novel bridge between network structure, resource flows, and sustainability assessment. Ultimately, this study also points out such a fact that scientific instruments should be regarded not only as high-value technical assets but also as platforms for collaborative knowledge production and public innovation. The case of the LZ Heavy Ion Irradiation Laboratory illustrates how strategic specialization can catalyze regional scientific ecosystems. To replicate and scale such successful models, the development of interdisciplinary sharing platforms, transparent data governance structures, and sustained public-private collaboration will be essential. Embedding ecological thinking into performance evaluation represents more than a methodological advancement; it signals a paradigm shift in how stakeholders conceptualize, allocate, and maintain scientific infrastructure. By moving from static metrics toward ecology-informed, flow-based, and fairness-oriented systems, instrument-sharing networks can evolve from resource distribution mechanisms into engines of inclusive, efficient, and sustainable development.