POLICIES TO PROMOTE OPEN SCIENCE: EVIDENCE FROM OECD COUNTRIES Giulia Ajmone Marsan Directorate for Science, Technology and Innovation OECD

OECD Principles and Guidelines for Access to Research Data from Public Funding (2006-07) Initial discussion of Open Science at CSTP in 2011 Many open science related activities on-going (PSI, open gov data, open educational resources, MOOCS) Once upon at time at the OECD 2

Decision to ask TIP WP to work on Open Science over 2013-14 first project on Open Science at the OECD (focusing on open access and open data mostly) Probably follow up activities in the near future today and tomorrow 3

ICTs offering new possibilities to share results Science is becoming increasingly data-driven Science: an evolving scenario 4 TDM-related scientific articles 1995-2014, per thousand article Source: OECD (2014), Measuring the Digital Economy: A New Perspective, OECD Publishing, Paris.

Main finding of forthcoming OECD report on OS 5

There is no formal definition of open science. In the OECD study, the term refers to efforts by governments, research funding agencies or the scientific community itself to make the primary outputs of publicly funded research results publications and the research data publicly accessible in digital format with no or minimal restriction Some definitions... 6

Open data are data that can be used by anyone without technical or legal restrictions. The use encompasses both access and reuse. Whether such openness exists from the legal perspective depends on the applicability of possible legal restrictions (or otherwise, whether the restrictions are removed by a free licence). Research data are factual records used a primary sources for scientific research, and that are commonly accepted in the scientific community as necessary to validate research findings. They are collected and produced in a wide range of formats: digital spreadsheets and databases, compilations from surveys, images, or objects. The consultation and usage of research data often involves use of specific computer programmes, software, etc. Some definitions 7

Open science is more than open access to publications or data; it includes many aspects and stages of research processes. Open science is a broader concept that also includes the interoperability of scientific infrastructure, open and shared research methodologies (such as open applications and informatics code), and machine-friendly tools allowing, for example, text and data mining. But open science is more than access to research publications and data 8

Improving efficiency in science Increasing transparency and quality Speeding the transfer of knowledge Increasing knowledge spillovers to the economy and society Addressing global challenges more effectively Promoting citizens engagement in science Why do we care about Open Science? 9 Open science is a means and not an ends. Policy makers need to promote openness in science while at the same time preserving competition. Competition is a key aspect of the scientific enterprise: pushing for open access and open data too early may be counterproductive in some cases.

PubMedCentral show that 25% of the daily unique users are from universities, 17% from companies, 40% are individual citizens and the rest are government or other categories (UNESCO 2012) A recent study on R&D-intensive SMEs in Denmark (Houghton, Swan and Brown 2011) found that 48% of those SMEs consider research outcomes very important for their business activities and more than 2/3 reported difficulties in accessing research material Ware (2009) conducted a survey on UK SMEs and found evidence that the equivalent of 10% to 20% of articles were not easily accessible for his survey respondents 10 OA can also contribute to innovationbut

Open science a relatively new phenomenon (at least for policy makers) Evidence on open access citation advantage (but quantification of this advantage is subjected to debate) Different behaviours in different fields (why? Over time?) Scientists tends to like open science in surveys (what about in reality?) Many estimates of the economic impact of data sharing (mostly on open gov data) Fewer estimates on research data sharing impact especially on innovation the impacts of open science need to be fully understood 11 Need to assess these impacts (to whom and on what?)

The open science ecosystem: a complex picture 12 Researchers Government ministries Research funding agencies Universities and PROs Libraries, repositories, data centers Private no profit organisations and foundations Private scientific publishers Businesses IGOs

Open science policies should be principle-based but adapted to local realities. Open science policies require a diversity of approaches, taking into account the needs of the different actors involved in research projects. For example, if a research project involves business sector partners and commercial interests are present, the requirements for sharing research results may be different from the case in which only public actors are involved. In other cases, privacy or confidentiality concerns may apply to the treatment of certain classes of individual data. Consultative approaches that involve all relevant actors for open science are a key component of successful open science strategies. Open science efforts involve different communities and different actors: researchers, governmental institutions, universities and research centres, libraries and data centres, private non-profit organisations, business sector organisations including private academic publishers, supra-national entities, citizens, etc. These actors do not necessarily have the same incentives, goals or expectations. A successful strategy needs to take into account this diversity, and react accordingly. As a consequence: 13

Open science: a hot issue for OECD and non- OECD countries 14 National STI strategy or plan

Open data policies and initiatives 15

The National Policy on Management of Research Data and Scientific Information: developed in order to implement the OECD Recommendations on Access to Research Data from Public Funding, designed in 2012 in order to optimize and rationalize the use of public resources involved in the generation and manage of knowledge; increase the access to research data and scientific information; reach international standards in the matter of access to research data, considering OECD recommendations Chile 16

China Scientific Data Sharing Programme since 2001 To break the data barrier before 1990s. To date, supported by the Programme, 24 platforms have been established to share scientific data from public funding, especially the observational data and basic data. 1500 S&T projects and engineering projects benefited from the data sharing. Data exchange with foreign organizations and institutions. 17

The Open Science and Research Initiative was established in 2014 by the Finnish Ministry of Education and Culture focuses also on research data Focus on data management plans, data storage and preservation in the long run Skills are a key component of the strategy: guidelines for data management have been developed for researchers and scientist Finland 18

In the Priority Initiative Digital Information of the Alliance of the German Science Organisations has dedicated working group focusing on research data. In June 2010, the Alliance of German Science Organisations adopted Principles for the Handling of Research Data, to establish structures to enable the collection, archiving and subsequent reuse of primary research data in all applicable disciplines. Focus on Skills The major research organisations are actively contributing to strengthen OA and OD. This includes rising awareness and providing information for scientists as well as offering support and advice: e.g. the Helmholtz Association has established regular training courses on managing research data with regard to Open Science. Germany 19

Sharing research data is encouraged but not mandatory: the Foundation for Science and Technology encourages (although it does not mandate) researchers to share the data resulting from research activities partially or totally funded by FCT, while assuring the protection of privacy and confidentiality Researchers are also encouraged to develop data management plans to ensure preservation Portugal 20

Open research data Research Councils (RCUKs) Guidance on Open Access states that all research papers, if applicable, should include a statement on how underlying research materials (such as data) can be accessed. However, the policy does not require that the data must be made open Focus on Skills Much skills policy in the UK surrounds providing skills training to students in numerical subjects. These skills policies are focused more around teaching students and academics how to use the big data sets that will emerge out of open data, rather than skills necessary for open science policies per se Large scale projects: The Open Data Institute: this 10m project will provide data from across the public sector on an open access basis to enable industrial and academic exploitation 160m investment in high-performance computing and networks (2011): the investments included high capacity networking, a national supercomputing facility and for the Hartree centre, offering leading edge supercomputing capabilities alongside software development expertise for industrial, academic, governmental and research organisations. 189m investment in Big Data: these investments included energy efficient computing, establishing a network of Administrative Data Research Centres (the ADRN), and the Square Kilometre Array (SKA) platforms. The United Kingdom 21

the White House Office of Science and Technology Policy (OSTP) memo explicitly refers to access to digital data Focus on Skills The OSTP Memo directs federal science agencies to coordination with other agencies and the private sector to support training, education, and workforce development related to scientific data management, analysis, storage, preservation, and stewardship The United States 22

Policy measures may include different efforts and initiatives, such as: enablers, incentive mechanisms or mandatory rules Open science policy trends: enablers, incentives and requirements 23 Enablers are, for example, the infrastructure developed to share articles or data, initiatives undertaken to develop an open science culture, amendments to the legal framework to make them increasingly open-science friendly or the development of the skills necessary for researchers to share and re-use the research outputs produced by others.

ICT Infrastructure: Argentina has developed the SICyTAR database with information on the CVs, publications and affiliations of researchers In Spain, RECOLECTA is the national repositories and main infrastructure that allows researchers and other stakeholders to freely archive and access research publications in Mexico, CONACYT supports the creation of institutional repositories Skills: the Finnish Ministry of Education and Culture created a Data Management Guide to teach researchers to develop data management plans in the UK, the Data Capability Strategy focuses on human capital and skill development for data analytics as well as data accessibility and data sharing skills in consumers, business and academia. The creation of centers for doctoral training on big data has been announced in several universities and higher education institutions in the country Legal framework: Australia, Finland are currently discussing modifications of the existing legal framework around the publication of publicly funded research results to make the copyright legislation increasingly open science friendly Germany has modified the national copyright law to grants scientists the right to upload publications on the internet even if they have transferred all exploitation rights to the publisher the UK has recently passed a series of amendments to its copyright legal framework (coming into force in 2014), which include a greater freedom of re-use of copied or recorded material for education and non- commercial research purposes For example 24

Data-related skill development is essential. Researchers skills needed for sharing articles or data sets openly on line are unevenly distributed. Some disciplines such as computer science or physics may have a longer tradition of uploading research material on repositories and curating and maintaining large data sets. Researchers in other disciplines, however, may need to be trained to develop the necessary skills to make open science happen. At the same time, students and citizens need to acquire the skills to take advantage of, use and reuse data set shared by the research community. Some countries are currently developing data science curricula to address this issue. Clear legal frameworks for the sharing of publications and reuse of data sets are needed at the national and international levels. A lack of clarity on the interpretation of national and international legal frameworks may prevent the sharing or reuse of research results. In addition, clear guidelines around text and data mining are needed, as this tool will become increasingly used by researchers in the future. Some OECD countries are currently discussing or have recently modified national legal frameworks to make them increasingly open science-friendly. Training of and awareness-raising among researchers is important for the development of an open science culture. Recent surveys on the behaviour of scientists reveal that not all researchers are necessarily aware of the possibilities offered by open science. In some countries, different institutions regularly organise workshops and training sessions to make researchers aware of these possibilities. Repositories and online platforms will not have impact if the information they contain is not of good quality. If repositories are not user-friendly and the data sets they contain have not been properly cleaned and curated, or the metadata have not been sufficiently developed, it may be difficult to maximise their usage. Policy messages in the OECD report 25

Policy measures may include different efforts and initiatives, such as: enablers, incentive mechanisms or mandatory rules Open science policy trends: enablers, incentives and requirements 26 Carrots (incentive mechanisms) may be in the form of financial incentives to cover open access publishing or the release of datasets. They may also be in the form of proper acknowledgment of open science efforts of researchers and academics, for instance in the form of data set citations or career advancement mechanisms partly based on metrics that take into account open science or data sharing efforts.

Better incentive mechanisms to promote data-sharing practices among researchers are needed. While all public sector researchers have an interest in maximising the sharing of published research articles, the same is not true for research data sets, especially at the pre-publication stage. In addition, data cleaning and curation (for example, by developing metadata) is a time-consuming activity that is rarely acknowledged in evaluation mechanisms or grant allocation procedures. Most evaluations of universities and researchers are almost entirely based on teaching and bibliometric indicators, attributing little value to the sharing of pre-publication inputs and post-publication outcomes, such as data and other relevant information. Extending citation mechanisms to data sets can partly address this issue. Policy messages in the OECD report 27

Policy measures may include different efforts and initiatives, such as: enablers, incentive mechanisms or mandatory rules Open science policy trends: enablers, incentives and requirements 28 Sticks (mandatory rules) are often implemented in the form of requirements in research grant agreements or in some cases are defined in national strategies or institutional policy frameworks.

major funding agencies in Australia, Costa Rica, Denmark, Estonia, Germany, Switzerland, the United Kingdom and the United States have mandated public access to the results of the research they fund More countries are also considering adopting rules for mandatory open access and data management plans For example 29

The long-term preservation costs of openly available research output need to be considered. Open access is not without costs. Many governments and research institutions are currently bearing the costs of offering open access to articles and to data, as well as the costs of storage and the preservation of data sets on line. Given the rapidly increasing amounts of data, public institutions will be challenged to find sustainable funding and business models. Public- private partnerships with private service providers may offer innovative solutions. Policy messages in the OECD report 30

Science and research do not stop at borders. Co-ordinated international efforts facilitate transfer of knowledge. Examples of international efforts to promote open science and overcome barriers include: and Open science: international efforts 31 OECD principles (2007) UNESCO: Open Access Policy Guidelines 2012; Charter for the Preservation of the Digital Heritage In Europe, Horizon2020 open science requirements, as promoted by the European Commission LAC countries: LA Referencia, a federated network of national repositories of scientific documentation for 9 LAC countries, financed by Inter American Bank

International collaboration in the area of open science is necessary to address global challenges. International collaboration is becoming more important than ever, as publications and data in electronic form travel across national frontiers. Shared and interoperable infrastructure is necessary to disseminate research results and promote scientific collaboration. Such efforts can help avoid the duplication of effort, as well as helping share the risks or the associated investments. In addition, BRIC countries Brazil, China and India are also adopting open science policies and data infrastructure roadmaps. International co-ordination and co- operation in this area will become even more important as the global production of knowledge and R&D increasingly shifts towards the emerging economies. Furthermore, tackling global challenges will require greater access to and sharing of national public research data sets and consequently, greater co-operation at a global level. Policy messages in the OECD report 32

Summing up Policies to promote open data are less mature than those to promote open access to scientific publications. While the principle of open access to scientific data is well established in OECD countries, the scope of access varies greatly across countries. This is due to the fact that data sets are not as easily identified and defined as scholarly research articles. Diversity of scientific data and differing traditions and standards in their treatment are also issues. Some of the additional challenges related to data sets include the definition of ownership of large-scale data sets, potentially collected by machines or software providers; privacy; confidentiality; and even national security issues. In addition, certain classes of data, such as medical and biological records, are particularly sensitive due to privacy issues. Many funding agencies are introducing open access requirements however, few policies targeting incentives other than funding are in place Many initiatives to invest in open access enablers (online platform, repositories, ) but less attention to skills development

Thank you 34 OECD report on Open Science Policy Trends will be released soon on: https://www.innovationpolicyplatform.org/ [email protected]

Documents

POLICIES TO PROMOTE OPEN SCIENCE: EVIDENCE FROM OECD COUNTRIES Giulia Ajmone Marsan Directorate for Science, Technology and Innovation OECD