Open First, Ask Later: Rethinking Data Ethics Among International Agencies, Donors, and NGOs

5 minute read

Aggregating and opening data presents opportunities for deepening civic engagement, addressing decades-long environmental coordination challenges, or highlighting inefficient institutional processes and procedures, among many other potential uses, benefits and information gains. The same data, branded as a public good, is also the subject of many debates regarding risks, agency, and privacy among others. There is a rising need to sort out the ethics of use and power dynamics over decision-making when it comes to open/ing data for ‘development’.

Launched in 2013, the Data for Development Challenge (D4D), a challenge run by mobile network operator Orange, opens mobile phone usage data up to researchers selected to analyze it, in order to facilitate the use of big data in key development sectors, defined in conjunction with government ministries and organizations (for example, the D4D Senegal Challenge focuses on health, agriculture, transport/urban planning, energy, and national statistics). There is an additional category, defined by Orange themselves for those “advancing anonymisation techniques to allow sharing of data that is relevant for society while respecting privacy. A constructive dialogue with the Personal Data Commission was opened up at the start of the project, focused, in part, to work on the creation of robust sharing formats and exploration of the uses and limitations of the synthetic data sets.” In 2013, the European Union committed to implementing the G8 Open Data Charter, which outlines five principles for government data: open government data by default; release as much data as possible (quantity) and then improve as needed (quality); usable by all- free, easy to access, open formats; release data for improved governance- transparency about processes and sharing of lessons learned with other governments; and release data for innovation- encourage use of open data in civil society. The EU’s open data commitments cover all grants provided by EU donor and development agencies. USAID announced its Open Data Policy in 2014, which commits all partners and subcontractors to submit any and all USAID-funded data, again open by default “with limited exceptions,” in a standardized format to their Development Data Library, tagging with relevant labels to facilitate metadata and consumer use.

In each of these cases, “opening” data represents a continuum of principles, commitments, processes, and actions related to generating, organizing and sharing [and making decisions based on] information. The rapid adoption of open standards would signal their promise, and has in many ways enabled and accelerated its momentum.

International organizations (non-governmental, multilateral, and bilateral) aren’t unique because of the data they work with, but because they play a unique, and often unclear role in civic, social, political, and economic life around the world - with a particularly heavy presence in conflict-affected countries and countries in transition. The number of NGOs alone operating in active conflict areas has grown exponentially in the past 30 years. According to Weizman, “while in 1980 there were about 40 NGOs dealing with the Ethiopian famine, a decade later 250 were operating during the Yugoslavian war; by 2004, 2,500 were involved in Afghanistan.”

The growing imperative among international agencies, donors, and organizations to open first and ask later needs rethinking. Here’s why:

• IOs are represented and play a role in highly sensitive, transitional or conflict-affected environments, where risks and considerations around the collection of information that is linked to an individual, even if not about that individual have different implications.
• The legitimacy of international organizations’ presence, projects, and services is often derived from other international bodies, and accountability requirements are most focused on populations in countries where organizations are headquartered. How people in different countries, individually or collectively, interact with, consent to, and/or share opinions related to international organizations and institutions is by no means uniform, where present at all, unless there is a particular evaluation underway and a dedicated staff person is deployed to collect it.
• There is a higher concentration of uncoordinated, and frequently evolving efforts, resulting in duplicated efforts and survey and research fatigue. All the fragmented data associated with different initiatives, that may be de-identified by an individual organization and seem innocuous as reported, has the potential for meta-analysis which doesn’t rely on individual details.
• The regulations and policies regarding data vary widely between countries, and it is unclear which country’s standards take precedent or apply.
• Where standards and protocols are unclear, program staff or field office directors often make the call. When there are standards imposed on situations for which they simply do not fit, meeting reporting obligations may take priority without clear knowledge of logic or consequences associated with decisions.
• IOs are accountable to a wide number of stakeholders, with the population they seek to serve, represent, or research coming in at the bottom of the list. They must often report to their headquarters, most likely located in another country, host governments, and donor(s).

With the volume of organizations growing alongside the volume of data being produced or opened through their actions or advocacy, questions about ethical data use, protocols, and standards quickly become daunting in their sheer volume; making them all the more necessary to ask at all levels.

SIMLab has benefited greatly from the work that others are doing on this topic and our own work in the past, including a data integrity guide, focused on data ethics when using low-end mobile technologies, published in 2011. In the coming months, SIMLab will continue taking stock of data ethics questions and approaches in international organizations, as open data and data-driven decision-making, accelerate into accepted, even expected, standards.