The CARE principles

If you work with data you have probably heard of FAIR. Since 2016 these four principles (findable, accessible, interoperable, reusable) have helped scientists publish data that other people and machines can find and use. They describe the data itself.

The CARE principles start from people. They were drafted in November 2018 at a workshop in Gaborone, Botswana, led by Stephanie Russo Carroll and Maui Hudson, and published by the Global Indigenous Data Alliance in 2019. Their authors point out that open data principles focus on "characteristics of data that will facilitate increased data sharing" while "ignoring power differentials and historical contexts." For Indigenous Peoples who are working to control their own data, more sharing on its own creates a tension.

CARE stands for four things:

• Collective benefit. Data systems should be designed so that Indigenous Peoples benefit from the data, in ways grounded in their own worldviews.
• Authority to control. Indigenous Peoples' rights and interests in their data are recognised, including the right to decide how it is collected, represented, accessed and used.
• Responsibility. People who work with Indigenous data have to show how it supports self-determination and collective benefit, and report back.
• Ethics. The rights and wellbeing of Indigenous Peoples come first at every stage of the data's life, including future uses.

The authors do not ask anyone to drop FAIR. Their slogan is #BeFAIRandCARE.

A 2023 paper in Nature Ecology & Evolution applies CARE to ecology and biodiversity research, the field GainForest works in. It names a familiar problem: helicopter research, where outside scientists fly in, collect samples and data, and leave. We heard the same story in Manaus. Researchers had often come once, asked their questions and never shared the results.

Here is CARE as four questions you can ask of any project, each with an example from GainForest's own work:

• Who benefits? During the XPRIZE Rainforest finals, the Indigenous Data Council approved the price paid for every piece of data the communities collected. The team later gave its $250,000 prize to an endowment for Indigenous scientists.
• Who controls? In the current version of Taina, records stay on the community's own device until a member chooses to publish them.
• Who reports back and builds skills? After two workshops, Marina Mura and Gabriel Pirow ran their own Taina workshop for more than thirty students at the Federal University of Amazonas, with no European team members in the room.
• Whose idea of harm counts? At Parque das Tribos, people decided for themselves that rare and culturally important species stay off the public record.

CARE matters even more once AI is involved. A model turns data into answers, translations, risk scores and training material for the next model. Once knowledge is inside a model it is very hard to take back out. CARE asks you to settle the questions while you still can, before the data moves.