Data is power, which necessitates ethics, governance and legislation to keep it in check. Luckily there are established guidelines to draw on, the latest of which has been published for the geomatics industry – but it could be improved.
Data-driven workflows form part of most business and operational processes. Decisions made with data affect every aspect of citizens’ lives, something the US Federal Data Strategy states at the outset of its Data Ethics Framework.
According to consulting firm Accenture, “analysing and acting on insights from data can introduce entirely new classes of risk”. Given the size, intricacy and wide use of data, that risk and harm can be considerable. Data ethics can help limit those risks and harm, and serve to optimise societal benefit.
Data ethics also help build trust and avoid the unintended consequences of data use. It guides data users through difficult and in-the-moment decisions. The World Federation of Advertisers (WFA) even argues that the health of our digital societies depends on data ethics beyond what’s legally required.
In the age where data provides a competitive advantage, and data gathering is ubiquitous, data ethics benefit individuals and private businesses, governments, NGOs and all other data-driven organisations.
A geospatial data ethics charter
The geospatial industry is catching up with other industries, following the recent publication and launch of the Locus Charter, a proposed set of international principles and guidance for the ethical use of location data. The result of an 18-month collaboration between EthicalGEO and the Benchmark Initiative, the charter chiefly outlines ten principles (with two explanatory sentences beneath each point):
- Realise opportunities
- Understand impacts
- Do no harm
- Protect the vulnerable
- Address bias
- Minimise intrusion
- Minimise data
- Protect privacy
- Prevent identification of individuals
- Provide accountability
These principles will be familiar to anyone working in related fields such as artificial intelligence, as they echo the likes of Microsoft’s Responsible AI Principles and Google’s AI Principles. Articulating location data ethics can be daunting, and the overlaps with other industries obvious, since location data intersects with IT, data science, artificial intelligence and machine learning, big data and the internet of things.
Like these industries’ data ethics codes, location data ethics is rooted in privacy concerns, such as those outlined in the World Geospatial Industry Council’s Geospatial Information and Privacy 2020-2021 Report. Likewise the Locus Charter, takes a macro-ethics approach: an overall framework rather than contextualising ethics.
A macro-ethics approach is appealing in a globalised world. In their paper, What is Data Ethics? Luciano Floridi and Mariarosaria Taddeo from the University of Oxford even consider this the only viable approach, saying “only as a macro-ethics will data ethics provide solutions that can maximise the value of data science for our societies, for all of us and for our environments”. They argue that the complexity of ethical challenges posed by data science requires data ethics to be developed from the start as a macro-ethics.
A macro-ethics approach makes the charter amply broad, but its brevity is a boon and simultaneously a bane for its adoption and implementation. Unfortunately, the Locus Charter
- lacks intricacy and nuance.
- doesn’t position itself in relation to complementary data and applications.
- fails to address aspects unique to location data and its uses.
- seems abstract without illustrative use cases.
- ignores the complexity of data value chains and workflows.
- neglects the changing data uses over time.
- does not provide assisting measures or tools to guide implementation.
Learning from other industries
Location data users can learn data ethics from the advertising industry, itself in the midst of a crisis of trust, with the World Federation of Advertising’s Data Ethics – The Rise of Morality in Tech arguably being one of the most compelling reads on the topic. While its authors agree with a principle rather than rules-based approach, they believe “the application of data ethics is subjective, cultural and contextual… while ethical principles may be universal, what is morally acceptable will vary from market to market”. As a first principle, the federation therefore recommends data ethics respect cultural nuances. It also takes a more proactive approach to inclusion by emphasising diversity in the data itself.
With location data often used in crisis scenarios, the American Association for the Advancement of Science has created useful decision tree tools to complement its Location-Based Data in Crisis Situations: Principles and Guidelines Report. Specifically developed for individuals and organisations who are using geo-located data in humanitarian or crisis-based work, the decision trees focus on data collection and sharing. To make the influence of data ethics more tangible, they’ve also published case studies of data ethics failures and successes.
Another good implementation guide is the EU’s Data Ethics, which outlines data ethics principles alongside a questionnaire to guide users in understanding data ethics in their own context. The Open Data Institute has created its own Data Ethics Canvas, a framework to help identify and manage ethical issues throughout a project using data, and which suits most contexts.
Data’s value also changes over time and throughout a workflow, something the Locus Charter ignores, along with the need to regularly re-evaluate data ethics and practices. One of the best approaches to assess changing data is the UK Government’s Data Ethics Framework, which employs an easy-to-use scoring system for three different project stages (start, during and after completion).
Creating and enforcing data ethics
While universal ethics guides are useful starting points, organisations would do well to create their own. Four common principles pertain to respect, fairness, accountability and transparency. The WFA however cautions against designing data ethics principles before understanding their operational impact. They further recommend creating a code of data ethics that complements an organisation’s existing privacy and security policies and reflects its culture, values and purpose.
In similar advice, the Alan Turing Institute recommends “concentrate[ing] on the content and nature of computational operations, rather than on the variety of digital technologies that enable them”. In addition, Accenture’s recommendations that the highest priority should be to “respect the persons behind the data” and “account for the downstream uses of datasets”.
To aid adoption of a data ethics code in your organisation, the WFA suggests first bringing everyone in an organisation on board (general awareness), before focusing on groups with specific responsibilities (focus groups). Just as important is to provide employees and colleagues with opportunities to raise concerns without fear of retaliation.
For those interested, there is a plethora of short online courses on data ethics:
- The University of San Francisco’s Center for Applied Data Ethics offers an Intro to Data Ethics Certificate
- The University of Michigan offers a Data Science Ethics course
- LinkedIn Learning has a Data Ethics: Making Data-Driven Decisions course
- O’Reilly tech learning offers an Ethics and Data Science course
Setting the tone
The Locus Charter, despite its shortfalls, brings location data into the fold of data ethics, and empowers individuals and organisations with a starting point. It promotes existing data ethics and inspires further adoption. Already, The Real Estate Data Foundation has followed suit by establishing its own Data Ethics Steering Group in April 2021 to look at data ethics in property technology, building information modelling, and GIS.
Data ethics concern everyone in an organisation, and places an obligation on everyone. Codifying morality through data ethics might seem tedious or like yet another bureaucratic management headache. It does not need to be. On the contrary, it will bring consistency and psychological comfort to you and everyone in your organisation working with data.