Epilogue

Eight years after my first epistemological crisis, during which I spent five years attempting to understand and find ways to improve the sustainability of science, I see no way around it: We need new systems to intersectionally address the systemic issues of science. Now, most reform fiddles with minor superficial knobs when and where we are allowed to. Piecemeal reform, such as giving virtual stickers for sharing data, is just not going to cut it when we want to do something about the sustainability of science. In this Epilogue, I reflect on the dire need for radical and uncompromising reform in science and I propose a set of demands.

The fundamental question when choosing between reformism and radicalism is whether one thinks the current system mostly works. Reformist thinking is conditional on the current system, whereas radical thinking is unconditioned. In that sense, reformist thinking accepts the very system it is trying to change — which makes perfect sense if one explicitly or implicitly believes the current system mostly works or if it serves them. Hence, there is a clear interest for established actors to argue for (conservative) reform. Conversely, radicalism does not condition itself on the current system and reimagines what is necessary where it is necessary. I consider that the current system, taken to its logical conclusion, does not work, hence, is unsustainable. For that reason, I see radical thinking as the only viable option.

That the current, article based, scientific system is too broken to fix has become painfully clear. Putting aside how invested the established institutions are in the current system, why should it objectively persist in its current form? Not only must all changes to the publication system, even effective ones, go through the gatekeepers of that system, those gatekeepers do not serve the scientific system. The publishers logically serve their shareholders in the end. Even if we could alter the rules such that ‘mischief’ no longer pays in the publishing game (Bakker, Dijk, and Wicherts 2012), framing it in terms of bettering science need not convince publishers. Moreover, my research suggests that it is unfeasible to detect mischief at large, indicating it may be more fruitful to aim for a system that prevents it in the first place. Only under the sunk cost fallacy does it make sense to me to keep the current system.

In order to understand the need for radical reform from another angle, we need to consider how the complex systemic issues of science are intertwined with the complex systemic issues of society. One societal issue that affects issues in science, for example, is austerity policy in government spending. Spending cuts affect the available national (research) budgets, which in turn intensifies competition among scientists, negatively affecting their mental health6 and feeding a race to the bottom of validity of findings (Smaldino and McElreath 2016). Vice versa, we cannot neglect the impact the scientific system has on societal issues. For example, we increasingly add to the ecological crises through conference travel as careers advance (Wynes et al. 2019); labs also produce large amounts of waste (one lab estimates approximately 1 ton of plastics per researcher per year; Urbina, Watts, and Reardon 2015).

Many complex interactions between issues of science and society exist. These include restrictive access to publications, causing many people to resort to illegitimate means to gain access for legitimate purposes (Bohannon 2016b)7. Additionally, in an attempt to manage the risks Open Access poses to their subscription models (and shareholders), the highly centralized oligarchy of academic publishers are shifting to business models resembling the problematic and data-exploitative models of Facebook. But even for researchers, the cost to play in publishing Open Access is substantial and rising faster than inflation (Grossmann and Brembs 2019), reaffirming structural and global inequalities. This extends existing inequalities in participation, such as participation in scientific education that is increasingly difficult as collective financing programs are replaced with individualized debt8. These complex issues do not even begin to grapple with the distortion caused by biased publication of results. If the studied effects are not large (the low hanging fruit) or widely reproduced using open data, the rules of the game promote contradicting results, serving confirmation bias. Moreover, intellectual property laws restrict our capabilities to deal with information overload in science. On top of all of this and more, systemic factors resulting in conscious and unconscious individual or institutional ableism, ageism, classism, homophobia, racism, sexism, and transphobia multiply these struggles for already underrepresented scientists, making it probable they become even more underrepresented as time passes. How can we create or expect valid, reproducible, and applicable knowledge if these issues are its context?

We need to start considering these interconnected issues together, not separately, and consider that current powerful stakeholders are invested in maintaining the status quo. Parallel scientific systems provide a space to envision something totally different. In order to build parallel systems that may provide hope in this web of complex issues, we need to integrate the critical study of scientific practices with critical studies of society, economics, organizations, information technology, and other relevant fields. Only that way will we be able to determine and interconnect the issues that need to be addressed in their broadest sense. Parallel systems also allow us to do more than simply shift power to a new generation; we can use it to restructure and distribute power. To restructure power in such a way that when we learn about the immanent problems of the new systems we build, power no longer serves to maintain the status quo but to change it. To distribute power such that those traditionally underrepresented may finally become adequately represented.

What should we demand of a parallel scientific system to make it worthwhile? The following demands can serve as utopian expectations to strive towards and evaluate proposals.

A worthwhile scientific system must be widely applicable. It must be usable on its own, but also in parallel with current structures people might already be embedded in (e.g., universities). This would allow simultaneous participation in both old and new systems for researchers in institutions. This would also not force (early career) academics into zero-sum publishing decisions, which would effectively exacerbate the institutionalized issues they are already facing. Nonetheless, the parallel system should be usable on its own for researchers outside of academic institutions, who should not be forced to conform to the norms of the institutions that they are not a part of (or may have actively rejected if they left academia).

Beyond minimizing barriers to participation in the scientific system, everyday research practices must be facilitated in the most concrete way. Too many proposals for change in the scientific system simply add to the already overburdening research process, making it more, not less, difficult to do and participate in research. Especially (quick) technocratic fixes serve such alienation from research: technocrats formulate these rules to create a veil of progress, telling people what to do but (often) not how they might go about (effectively) doing so. Additionally, technocratic measures are inherently institution-specific and affirm top-down control mechanisms, further entrenching power inequities. A worthwhile alternative must boost bottom-up change by making it easier to organize information, retrieve it, discuss it, and use it in the most concrete, everyday sense.

Alternatives must provide a coherent and extensive answer to the functions of a scholarly communication system. These functions encompass access, registration, certification, archival, and incentives. By analyzing the limitations of the current system on these functions, we can design an alternative that differentiates itself substantially and completely. By providing a more coherent and parsimonious answer to these functions, a parallel system may also prove more convincing for bottom-up change.

Those designing parallel scientific systems must actively renew and embed the values in science. As a result, incrementalism, validity, reliability, and replications must become preferred and rewarded over innovativeness, unreliability, and spontaneous isolated discovery. There is nothing objective about what we subjectively decide to value. Only deep and active value changes allow for incentive superstructures that substantially differ from the current and problematic system.

Access to and reuse of information must be complete and unlimited, further removing barriers to participate in research. Unlimited access and reuse means that intellectual control and censorship of information by state, commercial, or other actors needs to be mitigated by design. These actors benefit from bottlenecks in information flow that can be tightened, controlled, and surveilled. This is exactly how the services we now use are (being) designed. Hence, current centralized and restrictive models need to be replaced for decentralized and permissive models.

We must demand the potential of unlimited ways to deal with information overload. Full access and unlimited reuse permits us to deal with information overload in better ways by making it easier to commodify ways to consume and discover information. Creating such a market was part of the European Digital Single Market strategy, but it failed to do so. Oligarchic control over access and reuse prevents its existence today. By demanding decentralization and distribution of information, we may break up both the oligarchic publishing structures in science and break up the oligarchic control over the data on which services can be built.

To better deal with information overload we must also demand new information structures. Transparency is often proposed as the key to reforming science, but without somewhat standardized structures to make sense of what is available, there is a risk of obfuscation. I have spent many hours trying to understand someone else’s project structure with hundreds of files. This does not scale for sustainable information consumption and production. Questions about the unit of communication and what is communicable seem timely when the lack of chronology causes some of the issues in science (e.g., HARKing; Kerr 1998). As we create new information structures, we must also demand schemas that are flexible enough to allow for heterogeneous forms of research and prevent excessive rigidity that homogenizes and may even straitjacket research.

Due to the delicate balances and interests that need to be sought, we must demand that these systems are built in an open, dialectical, and actively inclusive way. It is unacceptable to create new (temporary) exclusionary privileges under the header of progress (e.g., the temporary Open Access privilege the Bill and Melinda Gates Foundation created; Van Noorden 2017). This means that any form of oligarchic control over how these demands are implemented needs to be actively prevented and bottom-up procedures need to actively be promoted.

Finally, we must demand systems that are sustainable for both knowledge production and the planetary environment. This is the acute societal context that we must not neglect (IPCC 2018). In order to reduce the need for consumptive resources (often fossil fuel based; Pirani 2018), we must reimagine how we can interact with other researchers. Doing so may drastically reduce the need for flights for conferences (and benefit our schedules). The digital services that are provided also consume substantial amounts of energy, especially given the rise of machine learning algorithms whose training is substantial in its energy use (Strubell, Ganesh, and McCallum 2019). Any radical proposals must be evaluated on their sustainability given the acute urgency of the ecological crises.

Only together do these demands produce a framework to build parallel systems that may start to substantially change the sustainability of science. Many scientific issues intersect with the societal issues that we are embedded in; ignoring this intersection harms effective and sustainable change. Reformist changes often lack this intersectionalism by being highly specific, which ultimately results in more effort spent for (maybe) the same goals. Regardless, the described framework served me in creating a concrete radical proposal (cf. chapters 8 and 9) and provides the basis as I implement this in my work moving forward. Nonetheless, there is a multiplicity of ways to think radically and achieve these demands. With pluralism of radical ideas, substantial change may happen. In the end, it does not matter who makes the change, but that the change happens.

References

Bakker, Marjan, Annette van Dijk, and Jelte M Wicherts. 2012. “The rules of the game called psychological science.” Perspectives on Psychological Science 7 (6): 543–54. doi:10.1177/1745691612459060.

Bohannon, John. 2016b. “Who’s Downloading Pirated Papers? Everyone.” Science 352 (6285). American Association for the Advancement of Science (AAAS): 508–12. doi:10.1126/science.352.6285.508.

Grossmann, Alexander, and Björn Brembs. 2019. “Assessing the Size of the Affordability Problem in Scholarly Publishing,” June. PeerJ. doi:10.7287/peerj.preprints.27809v1.

IPCC. 2018. “Global Warming of 1.5c. an Ipcc Special Report on the Impacts of Global Warming of 1.5c Above Pre-Industrial Levels and Related Global Greenhouse Gas Emission Pathways, in the Context of Strengthening the Global Response to the Threat of Climate Change, Sustainable Development, and Efforts to Eradicate Poverty.” https://www.ipcc.ch/report/sr15/.

Kerr, Norbert L. 1998. “HARKing: Hypothesizing After the Results Are Known.” Personality and Social Psychology Review 2 (3). SAGE Publications: 196–217. doi:10.1207/s15327957pspr0203_4.

Pirani, Simon. 2018. Burning up.

Smaldino, Paul E., and Richard McElreath. 2016. “The Natural Selection of Bad Science.” Royal Society Open Science 3 (9). The Royal Society: 160384. doi:10.1098/rsos.160384.

Strubell, Emma, Ananya Ganesh, and Andrew McCallum. 2019. “Energy and Policy Considerations for Deep Learning in NLP.” CoRR abs/1906.02243. http://arxiv.org/abs/1906.02243.

Urbina, Mauricio A., Andrew J. R. Watts, and Erin E. Reardon. 2015. “Labs Should Cut Plastic Waste Too.” Nature 528 (7583). Springer Science; Business Media LLC: 479–79. doi:10.1038/528479c.

Van Noorden, Richard. 2017. “Science Journals Permit Open-Access Publishing for Gates Foundation Scholars.” Nature, February. Springer Science; Business Media LLC. doi:10.1038/nature.2017.21486.

Wynes, Seth, Simon D. Donner, Steuart Tannason, and Noni Nabors. 2019. “Academic Air Travel Has a Limited Influence on Professional Success.” Journal of Cleaner Production 226 (July). Elsevier BV: 959–67. doi:10.1016/j.jclepro.2019.04.109.

  1. I am afraid this example is all too familiar to many.↩

  2. It even happens that governments seek university interns to get access↩

  3. Again, due to austerity measures of governments.↩