The policy is dead, long live the policy—Revealing science, technology, and innovation policy priorities and government transitions via network analysis

Abstract Each government has priorities for science, technology, and innovation policies (STIP). How can we identify the changing or reinforced STIP research priorities induced by government transitions? This study aims to unveil the STIP changing structure in the public funding research call contents in Colombia from 2007 to 2022, applying a co-word and network analysis approach to 389 funding research calls. We showed each government’s changing distribution of the betweenness centrality of 334 fields from Health, Life, Physical, and Social Sciences. We found that STIP priorities are getting denser and more diverse in terms of research fields. Despite this complexity, just 14 fields of Life, Physical, and Social Sciences, such as drug discovery, general business, management & accounting, and nature and landscape conservation, maintained their higher strategic relevance despite the government in office. This study maps the short and long-term focus of STIP regardless of the changing political tide inherent in democratic countries.


INTRODUCTION AND BACKGROUND
The policy is dead.Long live the policy!This alteration of the traditional proclamation unveils the political essence of STIP (science, technology, and innovation policies).Because STIP priorities change with every new government in office, what are the changing or reinforced STIP research fields induced by government transitions?We aim to unveil the STIP changing structure in the public funding research calls (RCs) content in a middle-income country employing a co-word and network analysis approach.Our proposal provides a methodological and empirical incursion to study the content's changing structure of STIP and identifies strategic research areas or fields adopted or discarded according to priorities established by the government while in office.
Our insights rely on the intersection between STIP studies and SciSci (science of science).STIP are rules, guidelines, and practices produced by governments to develop basic and applied research and their implementation in the economy within national borders (Edler, Berger et al., 2012;Meyer-Krahmer, 1984;Neal, Smith, & McCormick, 2008).SciSci, on the other hand, is a set of theoretical and quantitative techniques to unveil the determinants of scientific discovery based on the structure composed of researchers, institutions, and ideas (Bernal, 1939;Fortunato, Bergstrom et al., 2018;Price, 1961Price, , 1963)).
Since Vannevar Bush's report to the President of the United States, "Science, The endless frontier" (1945), STIP studies have produced well-established theoretical, conceptual, and empirical frameworks in higher-income regions, predominantly Europe.After the turn of the century, some argued that STIP should adopt a systematic perspective, assessing a cluster of actors or program strategies instead of a single organization or STIP instrument.Then, essential questions were formulated, such as, Are we doing the right thing?What results from our actions?Could we do it better?, followed by a portfolio of interventions (Arnold, 2004).
An extensive review of STIP evaluations in Europe found that the most used methods were descriptive statistics, context, documents, and case studies (Edler et al., 2012).On a minor note of particular interest, a minuscule fraction of all the studies reviewed used SciSci or network analysis approaches.Findings from examining the effect induced by STIP priority fluctuations stated that narrowing the research fields supported in RCs fostered a clustering of small and medium enterprises (SMEs) to exploit innovation.In contrast, widening research fields encouraged the clustering of larger and more diversified firms to support innovation (Ahrweiler, Schilperoord et al., 2015).As most STIP evaluations are limited to a project or program level, one study proposed a "system-oriented innovation policy evaluation," which comprises four attributes: coverage, perspective, temporality, and expertise (Borrás & Laatsit, 2019).
The qualitative shift from STIP government to STIP governance advocated for formulating its objectives via negotiation with a broader sphere of actors within and outside the research ecosystem (Pohoryles, 2006).It also acknowledged the crucial importance of STIP instruments, which change according to the entry of new governments or the formation of political coalitions (Pohoryles, 2006).Failed or unintended consequences of STIP are explained partly by institutional bottlenecks such as administrative/government deficits (Svarc, Laznjak, & Perkovic, 2011).Therefore, STIP fails to achieve all its desired outcomes not only due to market failures, but also institutional and government bottlenecks (Arnold, 2004;Gök & Edler, 2012).Hence the pertinence of proposing new angles, such as modeling, developing, and describing the rationale of STIP programs (Jordan, 2010).
The literature on STIP and SciSci proved fruitful, particularly on the effects of funding on scientific productivity and impact.This agenda seems to begin with findings of scientific performance in ocean currents and protein crystallography in the United Kingdom (Crouch, Irvine, & Martin, 1986).It explored the value of low-cost scientometrics reports on research evaluation for the Advisory Board for the Research Councils (ABRC).Along the same lines, further research delved into the Wissenschaftsrat (the German Science and Humanities Council) and its role in advising the German government on STIP (Block & Krull, 1990).A study highlighted the invariable relevance of SciSci in STIP national assessments to measure scientific production, recognition, and collaboration (Katz, 2000).Under a similar empirical framework, the influence of political forces on the scale-invariant pattern of the global innovation system (Katz, 2016) was established.Finally, a series of studies gave mixed results on the impact of funding on research productivity/impact and collaboration, such as weak positive relationships (Rigby, 2011), differences between countries (Gök, Rigby, & Shapira, 2016), and funding allocation biased towards already well-financed researchers, which as a result produced low-impact publications despite abundant resources (Shibayama & Baba, 2015).
In sum, our study contributes to this research agenda threefold.First, it expands the implementation of SciSci and network analysis techniques to STIP in middle and lowincome countries.Concretely, the relevance of individual case understanding of STIP and

Quantitative Science Studies 318
The policy is dead, long live the policy SciSci settings in middle and lower income countries has been highlighted explicitly but seldom addressed (Avellar & Botelho, 2018;Cortés & Ramírez-Cajiao, 2023;Ito, Li, & Wang, 2017;Li, Mao et al., 2022;Rodríguez-Navarro & Brito, 2022).Second, it delves into a "common ground" in recent STIP studies governance: the changing dynamics of government/coalition priorities-expressed and implemented via STIP.And third, it provides a mapping overview for public-private sectors to identify the short and long-term focus of STIP regardless of the changing political tide produced by the natural exercise in democratic countries.
We focused our analysis on Colombia.Colombia ranked 47th among the countries in the SCImago Journal and Country Ranking, above Argentina and below Romania (SCImago, 2020).In Latin America and the Caribbean, it ranked in fourth place, below Brazil in first place, Mexico (second) and Chile (third).A comparative study between Nobel Prize laureates and scientists awarded the most reputable science award in Colombia showed scientists of the latter elite with a higher composite citation indicator than the former (Cortés & Andrade, 2022).Colombia achieved such results despite negligible investment in research and development (R&D) and an infinitesimal scientific workforce.From 1996 to 2020, the R&D expenditure as a percentage of gross domestic product (GDP) maintained below 0.4% (the world average in 2020 was 2.63%) (World Bank, 2022).The number of researchers in R&D per million people between 2013 to 2017 was maintained below 89 (the world average in 2018 was 1,597) (World Bank, 2022).Therefore, we consider Colombia a country with exceptional STI output performance despite noticeable STIP instrument hardships.A closer look at the changing STIP in terms of strategic research areas or fields adopted and discarded according to priorities established by recent governments can provide insights into the STIP framework that certainly has influenced such processes.

Materials
We reviewed the Colombian STIP since the 1990s.It was a crucial decade because the government opened the economy and organized its STI national system (Fundación Alejandro Ángel Escobar, 2007;UNESCO, 2010).We searched the National Planning Department (DNP) public library for national STIP (DNP, 2021).We found just five STIP from 1991 to 2021.There is no clear temporal pattern for their official publication.Two policies were published in the 1990s with a 2-year interval, two were published during 2000-2009 and one in 2021.The term innovation was not added until the last two (2009,2021).Analyzing five STIP will not provide granular evidence of the sort required.
We found the national government and Ministry of STI (formerly Colciencias) open data portal and RC digital archive.The Ministry of STI is the national institution with responsibility to "formulate policies pertaining to their office, direct administrative activity and execute the law" (Colciencias, 2005;Constitución Política de Colombia, 1991;MinCiencias, 2021a).The open data portal has information on 3,907 RCs from 2009 to 2021 (MinCiencias, 2021b).However, information is "not available" (i.e., there are missing values) for 2,926 "science area" and 1,191 "'thematic area" RCs.Also, as we further discussed, RCs were classified in just one area/thematic area, which is misleading.The second source, the digital archive, keeps RC information from 2005 on (Colciencias, 2005;MinCiencias, 2022).The public RC archive has no standardized metadata storage (e.g., some RCs have specific objectives and disciplines/research areas supported, but others do not).Therefore, we hand-curated

Quantitative Science Studies 319
The policy is dead, long live the policy essential information by RC by looking at the complete terms of reference in digital scanned format: • The search was focused on research-oriented RCs, such as basic/applied research project funding, research evaluation funding, private R&D activities funding, international academic mobility, junior researcher, Masters and PhD grants, and research institutes/centers strengthening.We excluded RCs concerned with tax incentives, national research group assessments, national scientific journal assessments, technical services outsourcing for the Ministry of STI, and similar.• Terms of reference documents were manually examined.The scanning focused on identifying the explicit mention of areas of research fields to be supported or fostered by the RC (e.g., biotechnology, agricultural production, museology, and so forth).Figure 1 displays an example of key research areas or fields sourced from the STIP aim section (in English: "energy and natural resources, biotechnology, health, material science and electronics, information and telecommunication technologies, logistics and design, citizenship building and social inclusion").• Different keywords were used for the same research field in the RC (e.g., sustainability, sustainability science).The standardization method we used was to match the hand-curated key terms with the All Science Journal Classification (ASJC) (Scopus, 2020).The ASJC is a standard system designed by Scopus to assign a serial title to single or multiple fields.There are 334 fields in five areas: physical sciences, life sciences, health sciences, social sciences and humanities, and multidisciplinary.In doing so, key terms mentioned in the RCs, such as biotech and bio-inspired tech, were unified under the ASJC field of Biotechnology.• For the unmatched key terms with no correspondent ASJC, we searched for the most cited article in the bibliographic database Scopus (Scopus, 2022).We searched using the RC key terms on the article title, ensuring the key terms' centrality to the article's research topic (Nakamura, Pendlebury et al., 2019).Then we used the ASJC of the journal in which such an article was published.All of them were considered if a journal was assigned to multiple ASJC fields.For instance, for key terms with no ASJC match, such as "Cultural processes," the most cited article with a key term in its title was "Demography and cultural evolution: How adaptive cultural processes can produce maladaptive losses-the Tasmanian case," published in the American Antiquity, which is indexed with the ASJC: Museology; History; Archeology.• For an RC and its ASJC to be processed, it should have more than two research fields to form at least a triad of concepts.

Quantitative Science Studies 320
The policy is dead, long live the policy The elected government term lasts 4 years.To identify the STIP priority by government, we matched the RC year with government periods.The government of 2011-2014 was reelected for the period 2015-2018.We processed 389 RCs. Figure 2 presents the number of RCs and unique terms extracted from them.The average number of RCs per year was ~26 and the number of unique terms by year was ~288.Note that the latter average is before the ASJC standardization.The number of RCs has three peaks, in 2007, 2017, and 2020.The number of unique terms identified in each RC covaries with the number of RCs.In contrast, the lowest RC number coincides with the following year after the 2008 global economic crisis.

Co-word analysis
The seminal work of Callon, Courtial et al. (1983) introduced co-word analysis to visualize the structure of problems and concepts in relation to others.In this study, co-word analysis can build a structure of problems or concepts by identifying collocated ASJCs.For instance, if a given RC has three ASJCs, those ASJCs (nodes) are collocated (linked) given that all of them are contained in the same RC.Table 1 shows an example.

Network and node indices
We implemented two widely used network and node indices to study the changing STIP structural properties between governments (Scott, 2009).The first index is density.Density estimates how connected the nodes (research fields) are compared to the potential connections.

Quantitative Science Studies 321
The policy is dead, long live the policy For example, a density of 1 indicates that each node is connected to all possible nodes in the network.The equation for network density is D = 2L/n(n − 1), where L is the number of links and n is the number of nodes (Scott, 2009).
The second index is betweenness centrality.Betweenness is one of modern network science's most influential node indices (Freeman, 1977;Shugars & Scarpino, 2021).It is a measure of the centrality of a node (research field), which measures the number of the shortest path that passes through a node.This property unveils the node's ability to enable-or constrain-the information flow between clusters.The equation for the betweenness calcula- , where g ij is the shortest path that links nodes p i and g ij (p k ) is the shortest path that links nodes p i and p j p k (Opsahl, Agneessens, & Skvoretz, 2010).We normalized the index from 0 to 1; the closer to 1 the higher its betweenness.
Consider the following analogy to get an intuitive interpretation of the betweenness centrality score.In a firm, an employee with a high betweenness centrality score can be a key connecter between different departments and teams.Regardless of whether that employee is of a higher rank, that person can play an important role in facilitating the flow of information and knowledge between different teams.In contrast, an employee with lower or no betweenness centrality at all is still part of the firm's network of employees, teams, and departments.However, they might not contribute significantly to connecting or transmitting information or knowledge from one team or department to another.They might be an isolated employee with a marginal or peripherical role in the firm's strategic tasks.In this research, a field with higher centrality is a field that might be constantly present in the RC, not only in terms of frequency but also in diverse and different RCs.Imagine a research field that mentioned RCs in both Health Sciences and Physical Sciences and Social Sciences and Humanities.In contrast, a research field with a lower betweenness centrality score could be a field that appears in just a few RCs or even in multiple but only marginal and rather specific RCs.This would limit its role to a peripheral and rather marginal area in the RC network.

Network visualization: Hive plots
The state of the art of visualizing networks provides multiple and rich task taxonomies and visual encoding (McGee, Ghoniem et al., 2019).Our approach was framed as a year-by-year comparison based on numerical attributes.We consider that it provides the highest value added for the research aim because, besides visualizing the overall network structure, it also allows us to compare networks to one another based on network indices year-by-year (McGee et al., 2019).Quantitative Science Studies 322 The policy is dead, long live the policy We used an axis-based node-link representation called a hive plot.A hive plot places nodes uniformly across spaced axes arranged radially.Hive plots have been used in multiple fields, such as distributed computed and bioinformatics (Engle & Whalen, 2012;Krzywinski, Birol et al., 2012).Compared to the network in Table 1, hive plots increase the interpretation of network patterns because they use network properties as foundations instead of aesthetic layout and make comparisons between networks easier because they are perceptually uniform (Krzywinski et al., 2012).Figure 3 presents an example of a hive plot with a few nodes.Node sizes are proportional to their betweenness centrality.

RESULTS
Figure 4 presents the number of fields by area with betweenness score (i.e., more than zero) period-by-period.This figure also displays on the right-hand axis the network density score.There is an overall increasing trend in the number of fields included in RCs with, at least, a minor strategic role.The only exception was for Health Sciences between 2011 and 2014, where the number of fields decreased compared with the previous period, 2007-2010.The

Quantitative Science Studies 323
The policy is dead, long live the policy areas with the most consistent involvement were Physical Sciences and Social Sciences and Humanities, followed by Life Science and Health Sciences.However, the differences between the two latter were marginal period by period.Concerning the network density, between 2007 and 2018, the score showed that the fields are more interrelated with each other.In 2019-2022, the increase is even more marked: The density score went from 0.5 to 0.97, revealing an almost wholly connected network for the last period.
Figures 5-8 present the co-word networks and rain plots with the betweenness centrality distribution of research fields by the ASJC area (logarithmically transformed).Among the fields with the highest betweenness centrality during 2007-2010 are Health Sciences led (e.g., General Medicine), followed by Physical Sciences (e.g., General Energy), Life Sciences (e.g., Biotechnology), and Social Sciences and Humanities (e.g., Tourism, Leisure and Hospitality Management) (Figure 5).
Comparing the betweenness distribution by fields between 2007-2010 and 2011-2014, we can emphasize that Health Sciences went from a positive-skewed distribution to near-normal; Life Sciences went from positive-skewed to negative-skewed; and Social Sciences and Humanities went from negative-skewed to positive skewed.Physical Sciences maintained a negative-skewed distribution.We also can identify betweenness outliers in Physical Sciences (i.e., Renewable Energy, Sustainability, and the Environment), Social Sciences and Humanities (i.e., Sociology and Political Science) and Health Sciences (i.e., General Medicine) (Figure 6).Between 2007Between -2010Between and 2011Between -2014, multiple fields were included for the first time as a new government priority, particularly in the Social Sciences and Humanities (e.g., Marketing) and Life Sciences (e.g., Microbiology).
There was a significant addition of Physical Sciences (e.g., Biomaterials) and Health Sciences (e.g., Psychiatry and Mental Health) fields, between the 2011-2014 and 2015-2018 periods.The betweenness distribution for 2015-2018 resembles more that of 2007-2010 than the previous period, namely Health Sciences and Life Sciences had a positive-skewed distribution, and Physical Sciences and Social Sciences and Humanities had a negative-skewed distribution (Figure 7).

Quantitative Science Studies 324
The policy is dead, long live the policy For the period 2019-2022, nearly all fields were connected.Hence, we can observe the substantial evolution to include research fields as time passes.There was a more balanced inclusion of new fields compared to 2015-2018 in all areas, with influential fields such as Immunology, Dermatology, Environmental Chemistry, and Library and Information Sciences (Figure 8).The fields' betweenness by area distributions of this period resemble those of 2011-2014, except for Social Sciences and Humanities, which is like the previous period.
The constant and substantial network density throughout governments reveals the STIP vision of directly or indirectly supporting an increasing number of fields, particularly for the 2019-2022 government.Health Sciences showed a highly influential set of strategic fields with high betweenness (i.e., most periods with a positive-skewed distribution) for all periods.Physical Sciences, the area with the highest overall number of fields, had sustained and

Quantitative Science Studies 325
The policy is dead, long live the policy increased participation throughout periods, with negative-skewed betweenness distributions, a trend that resembled that of Social Sciences and Humanities.The Life Sciences fields' betweenness distribution fluctuated period by period.
Figure 9 displays the betweenness score changes for those fields identified throughout 2007-2022.Out of 334 fields, 248 were mentioned at least once in all four periods.However, when we selected those with above median betweenness score, the subsample was reduced to 14.In doing so, we frame our analysis to individual fields with sustained relevance for all governments and highly strategic value through their betweenness score.The first outstanding feature is the absence of Health Sciences.Visibly, the top five fields of highly strategic relevance for all governments, despite transitions, were Drug Discovery, Nature and Landscape Conservation, Management Science and Operations Research, General Economics, Econometrics and Finance, and Safety, Risk, Reliability and Quality.Fields starting in 2007-2010 with highly strategic importance, such as Oceanography or Soil Science, were banished by the succeeding governments.Recent and higher importance fields, such as General Business, Management and Accounting, Visual Arts and Performing Arts, or Theoretical Computer Science, were not as significant back in 2007-2010.The policy is dead, long live the policy

DISCUSSION AND CONCLUSION
We aimed to reveal the changing structure of the STIP content in a middle-income country employing a co-word analysis.Our findings expanded the implementation of SciSci and social network analysis to STIP in middle and low-income countries, formalized the changing dynamics of government/coalition priorities expressed and implemented via STIP, and provided a detectable overview for public-private sectors to identify the short or long-term focus of STIP regardless of the changing political tide.In addition, we examined the evolution of the embedded network in STIP and the growing complexity in designing STIP by computing the research fields of importance and their plausible support over 16 years.
Our approach to identifying the changing priorities in science policy across governments relies on the computation and interpretation of the betweenness centrality for each field and how this score might change, or disappear, in the analysis presented.Connecting with the analogy presented in Section 2.2.2 the identification of research field and their betweenness centrality across periods enables both academics and practitioners in the public-private sector to detect those strategic fields, not only in the frequency of their mentions in RCs but also in how those fields help to connect seemingly different and unrelated research field clusters (e.g., Genetics and Marketing) through their explicit mentions in diverse and strategic RCs.In other words, our approach enables us to grasp the underlying role of a few strategic key research fields despite the natural changing priorities in democratic government transitions.
Colombian STIP is becoming complex.This is because the STIP research field content is constantly growing and intertwining, building an intricate structure of STIP research field priority for the country.Because almost all fields were connected in the last period, it seemed unfeasible to identify a clear focus or strategic fields.In addition, the changing strategic position in the STIP-field networks modeled (i.e., betweenness centrality distribution) also added to such complexity the discontinuity and main concerns of the government transitions and the STIP they were designing and deploying.This complexity, however, unveils specific and strategic fields for the local STIP.
Out of the few studies on STIP and priority fluctuations, it was clear that narrowing the research fields supported in RCs might produce a clustering of SMEs to exploit innovation, while widening the scope encouraged the clustering of larger and more diversified firms to support innovation (Ahrweiler et al., 2015).However, our findings do not integrate the effects of STIP priority fluctuations and research/innovation outputs, nor the amount of funding by fields in the same framework, so as a consequence this explanatory factor should be treated with caution.
Our results show a persistent widening in research fields, although with a changing distribution in their importance in government transitions, and few fields with consistent involvement for the local STIP.We outline two conceivable factors for this expanding STIP diversification.First, comparing governments 2007First, comparing governments -2010First, comparing governments and 2011First, comparing governments -2014, there was a marked rise in the number of net and strategic fields during the 2015-2018 government.During the 2007-2010 and 2011-2014 governments research and development expenditure was less than 0.31%, but in 2015 it peaked at 0.37%, the highest since 1996 (World Bank-DataBank, 2022).It is plausible that this unprecedented budget for science gave the policymaking actors a boost in their sphere of action (i.e., countries with a higher number of resources for investing in science can adopt a strategy of diversification based on comparative advantage (Janavi, Mansourzadeh, & Samandar Ali Eshtehardi, 2020).Second, the national STIP adheres to a global STIP valuation model in which the quantity of academic accomplishments is the driving force for a successful career and the primary benchmark for international comparison: the number of publications, patents, scientists, citations, and (as discussed here) the number of research fields included in

Quantitative Science Studies 327
The policy is dead, long live the policy STIP and financing initiatives (Chu & Evans, 2021).This trend is also visible in the structure of scientific production and citations.Increasing coauthorship and multiuniversity research collaboration could explain the cross-subdiscipline interrelation growth (Jones, Wuchty, & Uzzi, 2008;Varga, 2019;Wuchty, Jones, & Uzzi, 2007).
We noticed that Health Sciences was an area with no consistent presence among the most strategic fields, despite its importance for the production of final goods and economic performance (Pinto & Teixeira, 2020).A study of the science policy in Health Sciences in Colombia from 1990 until 2016 outlined a set of remarks on the subject (Escobar-Díaz & Agudelo-Calderón, 2016).Science policies for Health Sciences from 1970 to 2007 increased the amount of funding in the 1990s fields such as basic and clinical research and health care systems.Despite advancements, it was not until 2007 that the Ministry of Social Protection-also in charge of national health issues-and Universidad del Valle, produced specific inputs for a national health science policy which, it turns out, were not implemented.
A new breath for a national health science policy emerged in 2015 with Law 1751.This law established the responsibility of the state to coordinate the creation of a STIP for health sciences with the aim of producing "new knowledge, the acquisition and production of technologies, equipment and tools necessary to provide a high-quality health service".The continuity of this line of policymaking action is reflected in the STIP policy for the upcoming decade (Consejo Nacional de Política Económica y Social (CONPES) 4069 Política Nacional de Ciencia, Tecnología e Innovación 2022-2031), a product of the second version of the Mision Internacional de Sabios (i.e., The International Mission of the Wise) (MinCiencias, 2019).In this STIP, there is a thematic focus on Life and Health Sciences, led by accomplished scientists such as Nubia Muñoz.In sum, the impact of these recent efforts in Health Sciences STIP to maintain the importance and inclusion of Health Sciences remains to be assessed in the upcoming years, considering the lack of consistent influence.
We also contrasted the fluctuation of relevance of Health Sciences in Colombian STIP with its research output.We noticed that Health Sciences have experienced a substantial and sustained research growth rate since 1996, superior even to Physical, Life and Social Sciences (SCImago, 2020).It leads us to contemplate additional explanatory factors besides fields prioritized in STIP, such as an area or field path dependency capacity.Institutions (i.e., a set of formal/informal norms or rules) carry history.In public/private (Health Sciences) organizations, what they do today depends on two things: what they could do yesterday and what was learned in the process (North, 1990).Adapting that perspective to our interest in STIP, new elements (e.g., government vision of science) should adapt and interlock with the existing conditions of the system, including its productivity dynamic (David, 1994).Elements explaining such a self-reinforcing dynamic are the accumulation of experience, crystallization of expectations, widening circle of their diffusion, diffusion of the knowledge thereof, and actions predicated upon that knowledge.
Path dependency is also related to positive returns for (Health Sciences) organizations.For instance, doing things in a particular already-known fashion yields the effects of doing similar things the next time (Coombs & Hull, 1998).STIP failures or unintended consequences-such as the gap between STIP design and implementation and actual results-are explained partly by institutional bottlenecks, such as administrative/government deficits (Svarc et al., 2011).
Our results are in line with the qualitative shift from STIP government to STIP governance.The latter advocates drafting its objectives via negotiation with a broader sphere of actors within and outside the research ecosystem, as STIP priorities can change according to the entry of new governments or the formation of political coalitions (Pohoryles, 2006).A clear example

Quantitative Science Studies 328
The policy is dead, long live the policy in that line was the 2019-2022 government.They campaigned for the concept of "Economía Naranja" (i.e., orange economy).Economía Naranja is an economic development model based on producing goods and services related to culture and entertainment (Ministerio de Cultura, 2018), a synonym for cultural industries.Before this administration, the concept was absent from STIP.However, in the most significant STIP policy for the upcoming decade, CONPES 4069 Política Nacional de Ciencia, Tecnología e Innovación 2022-2031 (i.e., the STIP outlines for 2022-2031), there was a whole focus group/STIP segment dedicated to creative and cultural industries.Between 2019 and 2022, there were 15 RCs in which the creative and cultural industries were among the prioritized areas/fields.Now that a new government is in office for 2022-2026, policies related to Economía Naranja will be ended or given another title (e.g., instead of "orange" it will be "multicolor") (Cambio, 2022;Portafolio, 2022).
Finally, this study's limitations lie in a single country and the absence of the directionalityshaping capacity of STIP and its actual effect on the structure of the national research and innovation system, including output, impact, structure (i.e., collaboration), and scientists and researchers' careers.Further studies could extend the scope to multiple countries and integrate the longitudinal effect of the changing STIP in national systems structure in a single framework.

Figure 1 .
Figure 1.Example of policy text section with key areas and research fields.Source: DNP (2021).

Figure 2 .
Figure 2. Number of RCs by year (left y-axis) and number of unique hand-curated terms extracted from RCs (right y-axis).

Figure 4 .
Figure 4. Number of fields identified in RCs with betweenness centrality score by area (left y-axis) and network density score (right y-axis) by period.

Figure 5 .
Figure 5. Co-word network and rain plots with the betweenness centrality distribution of research fields by ASJC area 2007-2010.Note: Visible nodes scored a betweenness centrality different from zero.Only the top five betweenness nodes by area are labeled.

Figure 7 .
Figure 7. Co-word network and rain plots with the betweenness centrality distribution of research fields by ASJC area 2015-2018.Note: Visible nodes scored a betweenness centrality different from zero.Only the top five betweenness nodes by area are labeled.

Figure 6 .
Figure 6.Co-word network and rain plots with the betweenness centrality distribution of research fields by ASJC area 2011-2014.Note: Visible nodes scored a betweenness centrality different from zero.Only the top five betweenness nodes by area are labeled.

Figure 8 .
Figure 8. Co-word network and rain plots with the betweenness centrality distribution of research fields by ASJC area 2019-2022.Note: Visible nodes scored a betweenness centrality different from zero.Only the top five betweenness nodes by area are labeled.