From Data Poverty and Fragility to the Long Road to Data Equity

*In part one I argued that the people we pledge hardest to reach are the ones our data systems are built to miss, and that some of that invisibility is the unintended cost of our own protective caution. This is the answer I promised: data equity, why it is not the opposite of data protection, and what pushing for it actually requires.*

In part one, I described a drought severity map my team built in Afghanistan that could show, district by district, where famine-like conditions and the need for agricultural, nutrition, food, WASH, and health assistance were most acute, but could not name a single one of the people inside those locations who were meant to receive it. I used that gap to trace a wider trap. A compounding cycle in which the under-counted become the under-served become the more-invisible, accelerated by fragility, and deepened, uncomfortably, by the very data-protection instincts we adopted to keep people safe. I closed by proposing data equity as the way to address those unintended consequences. This is what I meant.

So is there a balance? I think there is, but only if we stop framing this as a straight trade-off between visibility and protection, because that framing is the trap. The choice is not "expose people" versus "protect people into oblivion." The choice is whether we are willing to do the harder, more relational work that lets people be both seen and safe. That work has a name now, and it is data equity.

Data equity starts from a different premise than data minimisation. It treats representation as a right, not a risk to be managed downward. The Inclusive Data Charter frames it well. The goal is data that is genuinely representative of those usually marginalised, collected for all people regardless of location, ethnicity, gender, age, disability, or status. The point is not to collect more for its own sake, and certainly not to surveil. It is to deliberately close the representation gap that decades of passive and active exclusion have produced, and to do so on terms the represented communities actually control.

That last clause is where the balance lives. The reason granularity feels dangerous is that, historically, it has been extracted from communities and used on them, without their consent or benefit. The answer to that is not less data about the marginalised. It is a different relationship to it. This is precisely what the CARE Principles for Indigenous Data Governance, Collective benefit, Authority to control, Responsibility, and Ethics, were built to articulate. That the people in the data should hold meaningful authority over how it is collected, governed, and used. Pair CARE-style governance with the technical discipline of data responsibility and you get something better than minimisation. You get data that is granular enough to make people visible to the systems that serve them, and governed tightly enough that the visibility does not become exposure. Seen and safe, on their own terms, rather than absent and abandoned in the name of protection.

I am wary of essays that diagnose a deep structural problem and then resolve it with three bullet points, so let me be honest about scale. Nothing here is quick, and some of it runs against the grain of how the sector is funded. But the direction is clear, and it is actionable.

It means treating the foundational systems, civil registration, legal identity, inclusive national statistics, as core development infrastructure deserving sustained investment, not as technical afterthoughts. A child registered at birth is a child the system can never again pretend not to see. It means building data systems designed for the conditions that actually prevail in fragile and marginalised contexts, contested authority, insecure collection, low trust, rather than systems that assume the stable conditions of the places that designed them. It means funding disaggregation deliberately, because the groups who are left behind are invisible in any aggregate, and an average will never reveal them. It means shifting governance toward the communities in the data, so that being counted becomes something done with people rather than to them, which is also, not coincidentally, what makes granularity safe enough to be worth having, and which echoes the case I made in The Voices Our Data Systems Were Built to Silence and in building systems governments can own. And it means the institutions that allocate global resources learning to read missing data as a warning sign rather than an all-clear, because in the places that matter most, the silence in the dataset is the loudest signal there is.

None of this dissolves the tension I sat inside throughout part one. Protection and visibility will always pull against each other, and the balance is genuinely hard to strike, especially when there are real people in front of you. But the current equilibrium is not neutral. It tilts, decade after decade, toward the invisibility of the already-excluded, and it does so partly through choices we made for good reasons and never revisited. Data equity is the attempt to retune that balance deliberately, with the people most affected holding the dial, rather than letting it settle by default in the place that leaves them out.

I still think about the families behind that severity map, the ones the data could place inside a shaded district but never name. We have spent years promising people like them that no one will be left behind. The quiet, difficult truth is that the promise cannot be kept by goodwill or funding alone. It can only be kept if we are willing to see people, carefully, accountably, and with their consent, because a person no system can see is a person every system will eventually fail.

Data equity is not a software feature or a single reform. It is a decision to finish the ethic we started. To keep everything we learned about protecting people from the harm of being counted badly, and add to it an equal commitment to protecting them from the harm of not being counted at all. Leaving no one behind was always, underneath the slogan, a data problem. It is long past time we treated it like one.

And here is the revelation in that, the part I wish we said aloud more often. The barrier was never our compassion. We have never lacked the will to include people. We have lacked the means to see them. "Leave no one behind" asks us to carry everyone forward while staying quiet on the humbler act everything else depends on, which is seeing them in the first place. To be counted, carefully and on your own terms, is the first form of being valued by a system. It is the moment a person stops being an abstraction and becomes someone a clinic, a cash transfer, or a convoy can actually be sent toward. So the promise is not waiting on a greater conscience or a larger cheque. It is waiting on a choice we are fully able to make. To build the systems that let the unseen become visible, safely, and to treat that visibility as the foundation of the dignity we keep pledging. The people we have promised not to leave behind are not lost. They are uncounted. And what is uncounted can still be counted safely, and with dignity, the moment we decide it must be.