11-discussion.Rmd

# Epilogue {-}

Eight years after my first epistemological crisis, which led me to
spend 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 and fundamentally 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 unconditional. 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 for deep-seated change
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, but its 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 [@doi:10.1177/1745691612459060],
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.

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 health [@doi:10.1126/science.caredit.a1700028] and feeding a race to the bottom
of validity of findings [@doi:10.1098/rsos.160384]. 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
[@doi:10.1016/j.jclepro.2019.04.109]; labs also produce large amounts
of waste [one lab estimates approximately 1 ton of plastics per
researcher per year; @doi:10.1038/528479c].

Many complex interactions between issues of science and society
exist. These include restrictive access to publications, causing many
people to resort to illegal means to gain access for legitimate
purposes [@doi:10.1126/science.352.6285.508]^[Governments have been
documented to seek university interns to get access (see also
<https://therostrumblog.wordpress.com/2015/01/12/why-all-phd-students-should-do-a-policy-placement/>).].
Additionally, in an attempt to manage the risks Open Access poses to
their subscription models (and shareholders), the highly centralized
oligarchy of academic publishers [@doi:10.1371/journal.pone.0127502]
are shifting to business models resembling the problematic and
data-exploitative models of Facebook. But even for researchers, the
costs to play in publishing Open Access is substantial and rising
faster than inflation [@doi:10.7287/peerj.preprints.27809v1],
reaffirming structural and global inequalities, creating a paywall at the beginning of the publication process. This affirms existing
inequalities in participation, such as participation in scientific
education that is increasingly difficult as collective financing
programs are replaced with individualized debt [@isbn:9781781686195]. 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 chronically 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 that
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 (e.g.,
not being able to publish Open Access in the most career rewarding
journals). 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. Examples include stricter data regulations but with no support
structures to achieve them. 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 reconsider
what ought to be valued in science. As a result, cooperation, incrementalism,
validity, reliability, and replications must become preferred and
rewarded over competition, 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 published information must be complete and
unlimited, further removing barriers to participate in research. Not
all information should be published to begin with (e.g., sensitive
data), but published information should not be additionally
fortified. 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 new 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^[https://ec.europa.eu/digital-single-market/en/policies/digitising-european-industry],
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;
@doi:10.1207/s15327957pspr0203_4]. 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; @doi:10.1038/nature.2017.21486]. 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
[@ipccGlobalWarmingIPCC2018]. In order to reduce the need for
consumptive resources [often fossil fuel based; @isbn:9780745335612],
we must reimagine how we can interact with other researchers. Doing so
may drastically reduce the need for flights for conferences. 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
[@DBLP:journals/corr/abs-1906-02243]. 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.