In contemporary science and technology policy, fostering strong linkages between scientific research, technological development, and societal systems has become essential to driving meaningful social transformation. This essay proposes a typology of such linkages and examines the types of human resources required to enable them—resources that differ from conventional scientists and governmental personnel. These individuals must combine deep scientific expertise with multidisciplinary analytical capability at the meta-level and advanced skills for political coordination among stakeholders. Furthermore, to cultivate future talent of this kind, these competencies must be systematically but incrementally articulated and translated from tacit to explicit knowledge.

Designing a science, technology, and innovation (STI) policy often faces challenges in developing a vision and identifying policy issues due to its highly ambiguous and indeterminate nature. Accordingly, efforts have emerged across various academic fields to utilize scientific knowledge in order to promote more scientifically grounded STI policy design. This special issue surveys approaches to "better policy formation" from multiple academic perspectives and explores the possibilities for future developments in policy science.

Strategic intelligence (SI) has not been fully utilized in public policymaking since its introduction into the field of science, technology and innovation policy, partly because of a disconnect between the perspectives of practitioners and policymakers, as well as a divergence in the concerns of policy process research and policy analysis. This article proposes critical realism as the third philosophical paradigm to replace these ideological foundations. Critical realism—grounded in ontological realism, epistemological relativism, and judgmental rationality—opens up ways to a future-oriented SI that transcends mere accurate prediction or individual interpretation to engage with social reality. It also enables us to envision a liberating future that promotes social change through the critical exploration, reflexivity and open imagination of individual actors. We conclude by identifying interdisciplinarity and methodological plurality, ontological problem definition and commitment, and agency for future thinking as three critical realist perspectives essential for the future application of SI.

Science, technology, and innovation (STI) policy has grown more complex as societal goals gain prominence, obscuring what and how to evaluate. This paper analyzes STI policy through three frames— (1) the linear model, (2) the national innovation system, and (3) transformative change—and two system paradigms: cybernetics and autopoiesis. We examine how these perspectives shape system boundaries and policy intervention, using Japan's six Science and Technology Basic Plans as a case. We find that Phases 1–3 were rooted in Frame 1 but gradually incorporated Frame 2 elements. Phase 4 broadened the target system and showed early signs of Frame 3 while retaining PDCA-based control. Phase 5, with the Society 5.0 vision, strengthened autopoietic, learning-oriented approaches, marking a shift toward Frame 3. Phase 6 emphasized societal transformation but relied on politically led evaluation, creating inconsistencies. The study highlights the need to align evaluation methods with the system assumptions underpinning each policy frame.

The creation of unicorns has become a policy goal in innovation policy as a successful example of startups. However, when comparing the ratio of fundraising to the valuation of unicorn companies in Japan, the U.S., and China, the average ratio is 22% in the U.S., 26% in China, but only 8% in Japan. This paper analyzes the details of high-valuation startups in Japan. Specifically, it examines how the composition of various investor attributes in each funding round affects valuation estimates. The analysis reveals that when financial institutions with limited investment experience are more prominent, valuations tend to inflate relative to the amount raised. Additionally, foreign venture capital firms tend to make investments that suppress valuation inflation. Based on these results, there is a concern that startup valuations may be overestimated. While the presence of unicorns is undoubtedly important, using valuation alone as a policy goal is risky.

Innovation requires an ecosystem comprising talent, funding, support, infrastructure, and community. Academically, many studies quantify innovation efficiency, with Data Envelopment Analysis (DEA) being a key approach due to its ability to handle multiple inputs and outputs while leveraging each Decision Making Unit's (DMU) strengths. However, few studies systematically analyze efficiency across different startup stages while considering ecosystem significance.

This study applies Network DEA to evaluate startup creation and growth across three processes: knowledge production, knowledge commercialization, and business growth. It assesses "knowledge production efficiency," "knowledge commercialization efficiency," "business growth efficiency," and "overall efficiency." Findings reveal that Japan excels in knowledge production and commercialization efficiency but lags slightly (in comparison with the highest group) in business growth efficiency compared to top-performing countries. These insights help identify focus areas for Japan's future startup policy.

This study explores the potential role of statistical causal discovery (SCD) in enhancing evidence-based policy making (EBPM), alongside traditional approaches of statistical causal inference that estimate causal effects under assumed structural models. Focusing on the domain of science, technology, and innovation (STI) policy, the paper provides a technical overview of major SCD algorithms— including PC, FCI, and LiNGAM—and discusses extensions that account for latent confounders. A case study is also presented in which these methods are applied to examine the causal impact of policy-related variables on the number of entrants to doctoral programs in Japan. The analysis reveals that economic support has a positive effect on enrollment behavior, particularly in research-intensive universities, and that the determinants of DC1 fellowship awards provided by Japan Society for the Promotion of Science (JSPS), may depend on the institutional type. Finally, the paper highlights several future directions, including the need for well-designed datasets, consistency with domain knowledge, the integration of AI-assisted analysis, and collaborative engagement between analysts and policymakers.

This paper elucidates the structural challenges scientists face in reflecting scientific knowledge in public policy and the institutional conditions required to overcome them. I define scientists who proactively lead policy change based on their research as "Science–Policy Entrepreneurs (SPE)." SPEs operate in a domain where the different norms of science and policy intersect, forcing them to rely on unevaluated strategic practical knowledge, or "shadow work," leading to an "unsustainable dependency on specific individuals" premised on excessive personal burdens. Through a comparative case study of 48 projects from the JST-RISTEX program, my analysis reveals that the success or failure of policy implementation depends less on the qualities of individual SPEs and more on the dual dysfunction of "Quasi-Boundary Organizations (QBOs)": limited support functions and short-sighted evaluation metrics. Japanese QBOs, exemplified by RISTEX, fall into an "extractive" model due to their time-limited design, unilaterally exploiting the self-sacrificing efforts of SPEs rather than supporting their diverse activities long-term. To overcome this institutional pathology, it is crucial to establish a "reciprocal" ecosystem that organizationally supports and legitimately rewards individual vitality. The core solution lies in moving beyond temporary QBOs and establishing a sustained co‑creation platform equipped with stable support functions, appropriate evaluation criteria, and specialized personnel.

This study examines how visions for the relationship between society and technology are developed in technology-driven transformative innovation policies. Given the complexity and sophistication required, governments often face challenges in leading such efforts. Previous research suggests that networks involving stakeholders or government-affiliated intermediary organizations can play key roles in vision development. However, empirical evidence remains limited. Focusing on the "Industrie 4.0" concept, which envisions transformations in the relationship between society and technology in manufacturing, this paper analyzes the German government's policy documents to clarify the actors involved in vision formation. The findings reveal that, first, the government fostered policy-planning networks by creating private forums and providing public funding to existing intermediary organizations. Second, these networks were responsible for deriving, specifying, and developing the roadmap of the concept, which the government integrated into its strategies. This highlights a division of labor between government and private actors in addressing complex socio-technical challenges.

This study explores structural challenges in industry–academia–government collaboration in Japan's pharmaceutical sector, focusing on intermediary practitioners in technology transfer. Although open innovation is institutionally promoted, practical challenges remain. Semi-structured interviews with ten professionals from academia, pharmaceutical companies, and research support organizations were analyzed using the Modified Grounded Theory Approach. Three core challenges were identified: (1) institutional and resource constraints hindering academic drug discovery, (2) a shortage of personnel bridging technology and value, and (3) ritualization of matching activities between academic seeds and industrial needs. These challenges interact, generating structural risks such as the diminished attractiveness of research seeds, misalignment with corporate needs, and declining motivation, which collectively hinder sustainable technology transfer. A structural dilemma was also identified: academia's difficulty in pursuing scientific deepening, a bottleneck in the transfer process. The study suggests support measures, including intermediary training, flexible funding systems, and enhanced matching quality, as priorities for future policy and system design.

This study aims to construct a method for concretely inferring the conceptual structure of manufacturing that enriches the spiritual affluence of craftsmen, by using the action decomposition tree method, which makes the conceptual structures of manufacturing explicit, and statistical methods based on an online questionnaire survey concerning the conceptual structure of craftsmen and their spiritual affluence. The statistical methods use structural equation modeling.

As findings, we conducted interviews with 13 manufacturing skilled workers and extracted the 40 common conceptual structures of manufacturing. By constructing a knowledge processing system, we could infer the conceptual structure of manufacturing that further enhances the spiritual affluence of craftsmen. Using the action decomposition tree method and the statistical methods, it was shown that a knowledge processing system can be constructed, and a method for concretely inferring the conceptual structure of manufacturing that further enriches the spiritual affluence of craftsmen was constructed.

3D CAD is considered one of the industrial software applications that has most profoundly impacted manufacturing, and it is widely used in corporate design environments. Whether 3D CAD can create complex shapes depends heavily on a core technology called the solid modeling kernel (hereafter kernel), which handles geometric shape generation. Developing this kernel requires enormous costs and advanced technical expertise, which long served as a barrier to entry in the 3D CAD market. However, the release of two commercial kernels, ACIS and Parasolid, removed this barrier and created a new market (the mid-range CAD market). This paper examines the competitive strategies of these two kernels. Both strategies showed early signs of The Two-Sided Markets Strategy, but they failed to achieve the expected results. Focusing on the characteristics of 3D CAD as industrial software, this analysis suggests that network effects have both positive and negative aspects and can influence strategy.