Engendering public trust and epistemic standards
A significant assumption in many documents relating to open science (OS) is that openness about the scientific process and projects will engender public trust in science and in research results. This leads to transparency being identified as one of the enabling values underpinning OS, although transparency is also a prerequisite for other OS values such as scrutiny, critique, and reproducibility; accountability; quality and integrity, among others.
It is, however questionable whether full transparency, as an expression of the virtue of honesty, will always and automatically engender public trust – actual scientific processes are often messy and do not conform to the idealised processes that are written about in textbooks on methodology, or presented in the final scientific outputs in journals or more popular scientific writings. There is also considerable evidence that the public’s understanding of how science ‘works’, their ‘folk philosophy of science’ is very idealised, and that actual scientific practices are likely to fall significantly short of this idealised picture. Since scientific results are often complex and nuanced, there is a potential danger of “epistemic relativism” of many truths.
Epistemic injustice occurs when knowledge claims are unfairly rejected, or when the knowledge possessed by certain types of knowers are excluded or not taken seriously in violation of the principles responsibility, respect, and accountability. Issues of epistemic injustice are common in science where research results published by well-known groups at prestigious universities have often been evaluated as ‘better’ than publications from less well known groups, and where evidence presented in publications in high prestige journals have been often been evaluated as more reliable even though these journals have high retraction rates. Such bias goes against the principle of equality of opportunities.
In the OS context these epistemic reception biases leading to epistemic injustice are likely to persist, and they are likely to be extended to open data. Thus, the envisioned advantage of OS in terms of quality and integrity and consequently more efficient knowledge production might not materialise but rather the already existing advantages of the privileged will become even more entrenched.
Trust is essential for successful and accountable collaborations thus training in research ethics and methodologies is crucial for ensuring that researchers from diverse fields, as well as citizen scientists, share an understanding of the research standards and ethical norms.
Data collection is an important part of research itself and the effort and resources involved in the procurement and standardisation of data need to be sufficiently acknowledged. Data is valuable and researchers are under pressure to publish which might hinder their willingness to share data or share it early enough (as it would work against their self-interest or the interests of their institution). This may be a violation of the principle of openness.
Certain research methodologies and scientific fields, especially those that rely on large datasets and quantitative methods, are an especially good fit with OS. Qualitative data, often impossible to anonymize completely, cannot be shared as easily without identifying (and therefore potentially stigmatizing or harming) research participants, thus potentially violating the principles respect for persons and care.
Protecting research participants
The rights and interests of research participants lie at the core of RE and one of the major ethical challenges of practicing OS lies in how the goals of openness and data sharing can be fulfilled while also protecting the rights, dignity, and welfare of research participants.
The privacy of individual research participants can be fully protected in an OS data set if the data set can be completely anonymised. For many types of data this is possible (although often difficult). For other types of data, complete anonymisation is impossible, but it is nevertheless important to allow other researchers to use these datasets that might be unique.
Research participants may in some contexts have enduring control interests in relation to the data they have provided to researchers, e.g. in relation to what the data is used for and who uses the data; and they may perceive some uses of their data as misuses.
Autonomy of research participants is a crucial value and ensuring this in OS potentially requires alternative modes of engagement and consent. Given that research participants cannot withdraw their data once they have been deposited as an open data set in repository other mechanisms have to be developed that will allow research participants to protect their legitimate control interests.
Distributive justice in international knowledge production
OS is committed to the principle of openness expressed through the conceptualization of its products as public goods. OS practices are in principle reciprocal and symmetrical. Everyone contributes and can use the knowledge and data by making them openly accessible. This egalitarian picture of OS is, however, highly idealised. Many OS practices, e.g. preparing and annotating a dataset to fully comply with FAIR standards require resource, as does utilising a data set made openly available by other researchers. The ability of a researcher or group of researchers to fully comply with OS ideals and mandates, and their ability to fruitfully exploit what others make openly available thus depends on their access to financial and other resources. This means that researchers who are resource poor, e.g. researchers in LMICs and in the scientific periphery in more affluent countries, are systematically disadvantaged in relation to realising the benefits of OS.
This situation raises issues of justice and fairness that cannot be fully solved on a project by project basis, but need a systemic solution. Researchers have obligations to act fairly in the project collaborations, but other agents in the research system have obligations in relation to ensuring that resources are made available to researchers in LMICs that enable them to participate fully in OS.
Citizen scientists are valuable partners in many OS projects and their collaboration, participation, and inclusion are crucial for achieving a number of overall OS goals. However, participation of citizen scientists also has its challenges, (for example in terms blurring of the research object/subject roles in research or in terms of accommodating the activism of some citizen science a the more discovery-oriented stance). Transparency about the goals of research, openness regarding the various roles and interests of (citizen) scientists, and open data publication may help to alleviate these concerns.
Participatory research, while often offering valuable opportunities for all involved, has in some cases been associated with exploitation when citizen scientists are instrumentalised as a form of free labour and their contributions are not duly recognized (for example through authorship or ownership rights, if appropriate).
Scientific practice, while very diverse, still adheres to a set of basic research ethical norms intended to protect the participants and support the reliability and accountability of knowledge production. Data quality and integrity issues have been raised in this context as citizen scientists have often not been trained in research ethics and methodologies. Citizen scientists need to be included in having access to this knowledge and training.
Proper recognition of research contributions – alternative metrics
Another aspect of distributive justice in OS knowledge production is related to the proper recognition of all contributions to the research processes, and a proper alignment of the scientific reward system with the overarching goals of OS. This has been recognised as a major challenge for some time, but although many of the relevant organisations, RPOs, RFOs etc have officially signed up to take action in this area practical progress has been slow.
The lack of progress has both practical and more theoretical reasons. The main theoretical problem is that it is difficult to provide a principled account of how OS contributions, e.g. preparing a data set to FAIR standards and making it available equates to more traditional quantifiable contributions like authorship, citations, or grant success.
Openness beyond publications, data and code
In relation to a consensus commitment to openness and an implementation of OS practices most progress has been made in relation to open publishing, data, and code. There are, however many other elements of the research process that are not routinely shared openly and where there is no current consensus that they ought to be shared. This includes elements of the research process that are strictly necessary to reproduce particular research result (for example, highly specialised equipment and reagents, unique research sites, modified model organisms, etc.). These elements are currently often ‘traded’ for collaborative opportunities, or authorship or kept as proprietary ‘property’ in order to exclude competitors from utilising them in their own research. The situation is thus very similar to the traditional way in which research data was conceived of and handled before OS became generally accepted in relation to data. However, there seems to be no good reason to exclude many of these research elements from the obligation to openness and sharing. Many could be made public goods with a resource investment that is comparable to the investment necessary to make data fully FAIR compliant.
 Hofmann, B. (2022). Open Science Knowledge Production: Addressing Epistemological Challenges and Ethical Implications. Publications, 10(24). 10.3390/publications10030024
|This passage is part of D1.2: Suggested framework for addressing the (epistemic-ethical) challenges with knowledge production, that feeds into D1.3https://rosie-project.eu/wp-content/uploads/2023/03/D1.2-Suggested-normative-framework.pdf written by Søren Holm, Kadri Simm, Rosemarie Bernabe.|