Return to homepage

From the 2020 update report

Update to the Roadmap for Action

Acting in conditions of high uncertainty is particularly difficult. These are actions that libraries, in particular, and academic institutions, in general, could take regardless of the current situation.

10 mins read

New Responses in the Context of the COVID-19 Crisis

Acting in conditions of high uncertainty is particularly difficult. Academic institutions do not know what will happen to their revenues and expenses in the next six to nine months, let alone a couple of years out, and anecdotal evidence suggests that library budgets will have to be repeatedly recast as 2020 progresses. Offering near-term relief and maintaining some degree of operational continuity is paramount, and makes any longerterm planning difficult. However, uncertainty will decrease over time, allowing academic institutions to refine economic forecasts. It is important to take steps now to help maintain viability in the near term, while offering options for launching strategic initiatives at a later stage.

In the original Roadmap, we identified three classes of action. Below, we’ve provided additional actions that could help accomplish the twin goals of supporting near-term viability while laying the groundwork for future strategic initiatives:

Risk Mitigation Actions

These are actions that libraries, in particular, and academic institutions, in general, could take regardless of the current situation, although these actions are especially germane given the current crisis. For example:

  1. Limit cash outlays. If there are subscription collection expirations, this is the time to avoid renewing them, regardless of the intention to renew at a later stage or not. Negotiations could be put on hold until librarians have better visibility on 1) their budget for the next academic year and 2) the likely outcome of the possible new mandate for OA for federally funded research in the US. The value of subscriptions could even decline substantially in the years to come as a result of regulatory developments in various countries, and maintaining flexibility seems particularly valuable now. In this environment, publishers will likely be eager to be seen as collaborating with academic institutions and to avoid a constant stream of cancellations news in the press, so they will probably be more willing to offer discounts to minimize public controversies.

    We would also urge academic institutions to pursue alternatives before accepting inclusive access deals that raise total tuition costs for students and/or institutional spending (if their cost is wrapped into tuitions). OER is a practical alternative that limits total spending at a time when this issue is particularly urgent.

  2. Negotiate terms and conditions affecting sustainability. Historically, the academic library community has focused a great deal of its relationships (and tensions) with publishers on the rising costs associated with subscription contracts. There are several other terms and conditions that should also be put on the table now; the following are all essential elements of sustainability that should be pursued: billing in local currency, retention of perpetual rights to subscribed content, default author retention of copyright, financial hardships clauses, and price recalibration clauses as open content increases. Most important, the academic community should rally around open identifiers, and demand that all the relevant data that drives the data infrastructure is consistently open. This would create, over time, a “more level” playing field for new participants in the academic data and data analytics industry, as they would be able to build services that use the same data as the existing services. ORCID and DOI are good examples of these identifiers, but more are needed linking them also to data sets, grants, etc.

  3. Avoid “bigger deals” and distribution agreements that penalize alternative infrastructure. We outlined earlier the issues posed by linking data analytics and journal subscriptions, as well as by GetFTR. It is plausible that some vendors will try to bundle multiple products and services as a strategy to offset the likely pricing pressure they will encounter in the months to come. It would be highly problematic to accept even bigger bundles that would only limit the future flexibility of academic institutions and libraries. Similarly, adopting services that depress the roles played by repositories and other distribution channels should not be done lightly. Anindependent assessment of the real value of GetFTR to the academic communityshould be conducted before libraries and academic institutions sign on to GetFTR.

  4. Support the adoption of OER. Publishers will be eager to use this crisis to establish credentials for future sales (for example, by offering temporary free access to their digital courseware). Cengage, for example, has stated that it views Cengage Unlimited as a key element of its future strategy. Limiting the uptake of digital offerings from commercial vendors, in particular inclusive access and unlimited offerings, also seems particularly valuable because it limits the amount of student data that it will be possible to gather in the meanwhile. OER get very high marks for quality from both faculty members and students once they have tried them, but adoption is still limited. This is the time to launch a concerted effort to expand the support and resources available to faculty for the adoption, adaptation and creation of OER in lieu of digital courseware from commercial vendors.

  5. Adopt stringent data management and privacy policies and require commercial vendors to comply. Academic activities and campus life already generate vast amounts of data on both faculty and students, and the COVID-19 pandemic will only increase this trend. Moving online will add even more data and transfer a lot of it from the academic community to commercial vendors. In addition, a lot of this data can and will be used to “assess” faculty and students’ abilities and behavior, often with limited human supervision. For example, there have been reports charging that software used to detect cheating may disadvantage minorities, the poor and students with health issues.1 Just as it is necessary for academic institutions to introduce safeguards around the data they hold, there should be safeguards that protect faculty and students if they are forced to move to an online world.

In the original Roadmap, we advocated for academic institutions to identify a list of “principles” as a basis for adopting appropriate data policies (Exhibit 2). It is unrealistic to argue that all these principles are equally nonnegotiable in an emergency (for example, demanding an independent audit of algorithms may be unrealistic in the current crisis).

Exhibit 2: Principles of Data Analytics Usage 2
  • Transparency. Open source software, disclosure to enable testing for biases, auditing and evaluation requirements, etc.
  • Strong privacy protection. Consent, control over the use of data, right to erasure and correction, right to restrict processing, etc.
  • Accountability. Remedy for automated decisions, ability to appeal, etc.
  • Equity. Identification/correction of errors/biases, fairness, environmental impact, etc.
  • Human control. Opt out of automated decisions, human review of recommendations, etc.
  • Customization. Definition of non-standard reports, development of tools for a subset of users, etc.
  • Governance. Effective input from all stakeholders, independent review mechanisms, etc.

However, academic institutions should consider adopting four non negotiable principles, and demand – in parallel – that commercial vendors also support them if they want to operate with their community:

  1. Strong privacy protection. Faculty and students should give informed consent to the collection of data and be entitled to an explanation as to how it will be used. They should also have a right to restrict processing to a specific list of tasks (without being forced to provide a blanket acceptance) and they should have a right to demand erasure after courses are completed. Individuals should also have a right to demand rectification of data that is wrong or incomplete. Within applicable laws, requests from any government for data should be notified immediately to the individuals involved and the data handed over only in the presence of a legitimate court order.

  2. Accountability. There should be rights in place to demand remedy for any decisions that are made by algorithms, and there should be clearly identified and accessible processes in place to appeal decisions.

  3. Human control. There should be a right to opt out of AI-driven decisions and demand a human process.

  4. Accessibility and Equity. There should be explicit indications that data and data services are accessible to all relevant constituencies, with no barriers or impediments. There should be rights in place to demand that any biases that are identified, including when identified outside the institution (for example, by other academic institutions or independent auditors) are immediately notified and corrected.

Strategy Actions

As we pointed out in the Roadmap for Action, this second category of actions is more complex, since it relates to decisions that will need to be made specifically based on each individual institution’s mission, culture and values. It also involves the establishment of an explicit, structured process to determine the position that each institution wants to take in regards to specific issues posed by the collection of data and the deployment of data analytics tools. Establishing such processes in the midst of a crisis is certainly complex. Running such processes will be more difficult at a time of campus closure, financial stress and planning uncertainty.

However, some of these issues are so important that they will need to be resolved urgently, and resolution should involve all relevant parties. For example, the need to reconcile student and faculty privacy with health protection will require choosing proper monitoring tools that adequately balance very divergent goals. Similarly, academic institutions will need to decide whether (and to what extent) they want to substitute humans with algorithms in a number of activities, from screening student admissions to student tutoring to remote exam monitoring. In each of these instances, all parties involved should have a proper voice as these themes are debated and decided.

Community/Collective Actions

A number of actions require such large efforts that only concerted action, sustained by several institutions, can accomplish real impact.

  1. Support community initiatives on dissemination and data infrastructure. Financial resources will shrink for the foreseeable future, penalizing new initiatives aimed at building the next generation of community-owned tools and infrastructure. Scarce financial resources should be pooled to achieve the most impact, rather than dispersed through individual library spending.

    This is an ideal time to pool whatever resources are available, including, if possible, some of the savings from the cancellation of collections subscriptions, into open infrastructure initiatives (such as IOI) and library-supported platforms.

  2. Advocate for the immediate opening of all articles to text and data mining. One of the lessons learned from this crisis is that text and data mining are becoming important tools in accelerating science. There is no morally defensible reason why such acceleration would be possible only to help treat COVID-19 patients. This is the time to ask publicly all publishers to relinquish text and data mining rights for all articles and to identify publicly any publishers refusing to do so.

  3. Seek opportunities to acquire courseware publisher content in order to make it open. The continuing decline in the size of the courseware market, coupled with the failure of the McGraw-Hill/Cengage merger, may lower valuations for publishers’ assets. It is now possible to contemplate an investment in existing, high-quality titles and technology aimed at transforming them into OER. This action would likely still require the financial support of funding bodies interested in supporting the provision of OER, with the additional incentive of making OER available more rapidly and efficiently.

  1. For example, see here 

  2. Based on Principled Artificial Intelligence: Mapping Consensus in Ethical and Rights-based Approaches to Principles for AI published under the auspices of the Berkman Klein Center at Harvard University 

About the authors

Portrait of Claudio Aspesi

Claudio Aspesi

A respected market analyst with over a decade of experience covering the academic publishing market, and leadership roles at Sanford C. Bernstein, and McKinsey.

Scholarly Publishing and Academic Resources Coalition

SPARC is a non-profit advocacy organization that supports systems for research and education that are open by default and equitable by design.