The road to success: drawing parallels between 'road' and 'research data' infrastructures to foster understanding between service providers, funders and policymakers

Background: The work of data research infrastructure operators is poorly understood, yet the services they provide are used by millions of scientists across the planet. Policy and implications: As the data services and the underlying infrastructure are typically funded through the public purse, it is essential that policymakers, research funders, experts reviewing funding proposals, and possibly even end-users are equipped with a good understanding of the daily tasks of service providers. Recommendations: We suggest drawing parallels between research data infrastructure and road infrastructure. To trigger the imagination and foster understanding, this policy brief contains a table of corresponding aspects of the two classes of infrastructure, and a table of policy implications. Conclusions: Just as economists and specialist evaluators are typically brought in to inform policies and funding decisions for road infrastructure, we encourage this to also be done for research infrastructures


Introduction
Data-intensive research depends on data and data services, such as databases, software and tools, and standards. These are often made available to end-users through research infrastructure. Such a research infrastructure for biological data and bioinformatics service in Europe is ELIXIR. 1 As is common for many types of infrastructure (especially those that are free at the point of use), the existence of research infrastructure, like the services provided by ELIXIR, is often taken for granted by their users. Their importance is only noticed when they are (temporarily) unavailable, or worse, when they disappear due to discontinued funding. 2,3 In the Kano model, infrastructure services are must-be qualities: their proper functioning does not make the users happy, but service disruptions are strong dissatisfiers.
As research infrastructures and their services are typically funded through the public purse, it is essential that policy makers, research funders, experts reviewing funding proposals, and possibly even end-users are equipped with a good understanding of the daily tasks of service providers. We have noticed that this is not often the case, and this becomes an issue when this lack of understanding affects the funding of research infrastructures (funding decisions are typically made based on scientific advice to funding bodies). To foster better understanding, this policy brief provides a comparison table of features of data infrastructure and their relatable counterparts in road infrastructure (Table 1). Any further responses from the reviewers can be found at the end of the article We believe that this approach can help, firstly because the change in mindset makes it possible to see consequences of certain choices more clearly, especially those linked to funding of research infrastructures. Secondly, many people (decision makers and those influencing the decisions), even those working in research, are much more accustomed to road infrastructure setup and road infrastructure disruptions than to research infrastructure setup and disruptions; this increased familiarity increases the chances that consequences of policy decisions are foreseen.

Policy outcomes and implications
The comparison between research data infrastructure and road infrastructure has many hooks to support productive discussions, and decisions, on research infrastructure funding and sustainability governance. Some examples of policy implications following from the comparison are given in Table 2.
Given that we, the co-authors, all work in research infrastructures, there is an inherent bias in the approach presented, just like when Sutherland et al. 4 published their "20 things politicians need to know about science". However, we hope our thoughts have practical use: policymaking is complex and multifaceted, as astutely explained in "20 things scientists need to know about policy". The comparison tables are simply our contribution to fostering longer-term sustainability of Knowledge in national governments about the companies building and maintaining roads Table 2. Policy implications of the comparison between data infrastructure road infrastructure.

Data infrastructure Road infrastructure
Thinking data infrastructure is too expensive (or lacking sufficient accommodation of one's specific needs) and preferring to build everything yourself to exact specifications; underestimating the value of services (e.g. 24/7 support) and underestimating the real cost of building everything yourself (excluding e.g. energy bills, and the fact that the postdoc employed for the task is not productive for research work while making weekly backups and troubleshooting issues) Not believing in road taxes nor other centralised tax systems (or complaining that you have to walk the last bit from the parking lot), underestimating the value provided (to its users) by the road infrastructure and underestimating the cost incurred to get exactly from A->B without roads (i.e. using an all-terrain car), e.g. forgetting that there won't be any gas stations or other support services along the way either Not knowing whether the data infrastructure will be sustained for the life of your project (and its successor), and therefore building a makeshift infrastructure yourself Not knowing whether the roads will be maintained for 10 years, and therefore planning your vehicles around off-road travel Infrastructure investing in services based on community needs Government prioritising road investments based on transport needs Asking a research infrastructure provider what scientific breakthroughs the infrastructure will be making. A RI facilitates/enables breakthroughs (as well as enabling more routine research to be carried out) but does not make them nor predict what they will be Asking a road construction/maintenance company where it will be driving its own cars on the new road, rather than asking what new economic activity the new road will facilitate Requiring that a funding proposal for "infrastructure" the enables scientific breakthroughs during the lifetime of the grant Giving a road construction company money to build a road only if they guarantee it will meet a threshold of user journeys during the 3 years it takes to build Asking top scientists to review proposals for research infrastructure funding. The services that would be offered by the infrastructure would be competing with what they (and only similar top scientists) could achieve in their own labs Asking those who own expensive Jeeps whether they approve the construction of a public road. They and others with off-road cars do not see the need for roads, they can get along just fine without them infrastructures that already exist, that are widely used across the world, and that have typically received significant public financing over the years.
Furthermore, the comparison tables are likely to support efforts of both research infrastructure operators and policymakers in more accurately conveying to taxpayers the public value of research infrastructures, in addition to their role as enablers of scientific discovery and applications of societal benefit. The word 'enablers' is perhaps the most important message to convey: just like a road enables travel (and a research infrastructure enables research), it is questionable whether it is right to ask the road construction/maintenance company (and the research infrastructure operator) whether the road (and the research infrastructure) brings value to society. Economists and evaluation specialists are very well placed and qualified (and likely unbiased) to answer complex questions around the public value of financing roads and research infrastructures. 5,6 Actionable recommendations Recommendation 1: When formulating opinions and decisions on research data infrastructure funding and sustainability governance, compare them with that of road infrastructure. The change of frame may bring new insights.
Recommendation 2: Consider informing policies and funding decisions relating to existing and future research infrastructure with support from economists and specialist evaluators.

Conclusion
We welcome any additional ideas for the comparison as well as discussion on improving the existing tables as comments to this paper. For instance, the parallels could be improved by considering other infrastructures delivering public services, such as water supply and sewage systems. We tried, but found it difficult, to broaden the set of comparisons to also include a sustainable travel angle (e.g. examples covering public transport versus private car travel). Considering the climate emergency, this would be a useful and still relatable expansion of the approach.

Data availability
No data are associated with this article.

Open Peer Review
understood and justified.
The article is well-written and to some extent illuminating. The parallel with road infrastructures is fitting quite well with the purpose of the authors. The parallel may go even further in depicting the parallels with other infrastructures delivering public services such as water supply, sewage system and so on. The article however does not fully exploit the potential of this parallel and remains a bit open and generic in the explanations.
First, the table, which is pretty much the core of the article, presents some elements that should be better qualified, explained and perhaps even reconsidered in a few cases (e.g. "data service" corresponds perhaps more to 'transport service' than to "road"; it is not clear why "research project " corresponds to "car" and "project" to "trip, journey"; "oversubscribed service" may be 'crowded road' more than "traffic jam"), so it would be better to add a short explanation or a narrative section where the content of the table is described, justified and put in context.
Second, the policy options and implications do not come immediately from the description of the main common features, especially for those who are not familiar with both research and more traditional infrastructures. Perhaps the table could be split in two to distinguish the key features (first half) from the policy implications on their funding and public justification (second half more or less).
Overall, the article is very interesting and well-conceived, but it would benefit from a small additional effort to better explain the main claim and its implications.

Does the paper provide a comprehensive overview of the policy and the context of its implementation in a way which is accessible to a general reader? Yes
Is the discussion on the implications clearly and accurately presented and does it cite the current literature? Partly

Are the recommendations made clear, balanced, and justified on the basis of the presented arguments? Yes
Competing Interests: I hereby declare that, in reviewing the article 'The road to success: drawing parallels between 'road' and 'research data' infrastructures to foster understanding between service providers, funders and policymakers' by Hooft RWW and Martin CST, no conflict of interest applied. In the interest of full transparency, I report herewith the non-financial relationships I have with one of the authors of the paper, Martin C.S.: -During the period 01/01/2018-01/06/2020, on behalf of our own institutes (CSIL and ELIXIR) we were both members of the research team of the project 'Charting Impact Pathways of Investment in Research Infrastructure' a research project granted by H2020 programme aimed at developing a model describing the socio-economic impact of research infrastructures and of their related financial investments. In the context of this project, we co-authored the report Deliverable 5.1 of the RI-PATHS Project. -Since the 1st of September 2022, we are both part of the PathOS project, a Horizon Europe programme aiming to collect concrete evidence of Open Science effects and study the pathways of Open Science practices. The project will last four years and we will collaborate to develop a joint case study. We work on PathOS for two different institutions, being partners in the project. They are supposed to be one of the case studies and we are supposed to perform an analysis of impacts generated by one of the open resources they have. Two people are working from their side and three experts are working from our side, in addition to all the other project partners (8 institutions for a total of more than 30 people involved). However, both Martin and I have a more supervising role, while the other colleagues are more hands-on.

Frédéric Sgard OECD, Paris, France
This policy brief addresses a very important element of research policy, which is the understanding by decision-makers of the role, relevance, and impact of research infrastructures. In this particular case, of research data infrastructure, by using a comparison with a much well known type of infrastructures: roads. This comparison with a much more mundane example (roads) is aptly supported by a strong argument regarding the usefulness of the infrastructure mostly appears when its activity is disrupted, as normal operation is often taken for granted and thus invisible. The comparison between data research infrastructures and roads is clearly and convincingly argumented. This brief is also timely as funders are increasingly stretched to provide support to existing and new research infrastructures and have to make selective choices in their investments.

Does the paper provide a comprehensive overview of the policy and the context of its implementation in a way which is accessible to a general reader? Yes
Is the discussion on the implications clearly and accurately presented and does it cite the current literature?