Minimise Harm

Like all data infrastructure, a collaborative maintenance project involves several different communities. There is the community of contributors, a community of data users, a community of people impacted by how the data is collected, shared and used and, depending on the dataset, a community that is represented in the data. These communities might overlap with a single individual present in multiple communities. All of these communities might benefit from, or be naegatively impacted by the project or use of the data.

It is important to ensure that data is being ethically and legally collected, shared and used.

Use the data ethics canvas to help you think through the potential impact of collecting and using the data. Ensure you consult with the different communities to help understand potential harm. Repeat the exercise if the scope of the project changes.

Harm might also arise from bias in a dataset. For example, if you are collecting data that might be used to train an algorithm for machine learning system, then ensuring that the data is representative of the communities who may be impacted by use of that algorithm, will improve the quality of the dataset and reduce harms.

It is not just with personal data that harm may occur. For example, some projects collect data about animal sightings, and restricting access to information about sighting of endangered species could help limit poaching.

In addition to the risk of harm arising from the use of data, or bias in the data, contributors might incur risks during data collection, eg if they are carrying out surveys in busy cities or conflict regions. Ensure that you provide guidance to contributors about collecting data when it may impact their personal safety.