What accountability regimes cost the people doing the most important work — and what a different recursion might look like.
Preface: The In-Between Is Where the System Lives
Thirty years of immersion in a community produces change that no programme design can replicate. The question worth sitting with is: why? What is the mechanism?
For me, it’s because deep, sustained presence builds relational density — the trust, the reciprocity, the shared meaning that accumulates between people and local institutions over time. That’s what actually translates change into practice. Grassroots relational density is vertical, deep within a system, whether it’s the Anganwadi, the school, or the local health network, and is therefore irreplaceable.
But each of these systems intersects with the others. The Anganwadi system with the health system, the education system, the justice system. It’s precisely at those intersections — the in-between across systems — where citizens fall through the gaps, where responses break down, but also where the most generative possibilities lie.
Complexity science has been saying this for decades: emergence happens at the edges, not inside the nodes. Neil Theise’s work makes the case for this operating at every scale, from cell biology to consciousness. And Buddhism has been saying it longer — that nothing exists independently, everything arises in relation. The in-between isn’t a gap to be managed; it’s where the system actually lives.
So the problem with heroic systems change isn’t just that it’s overconfident. It focuses on the nodes while leaving the in-between invisible and untended.
What this means is that we’re missing a practice and a theory for tending the in-between — not a retreat from systems change but a more honest accounting of what systems actually are and what our roles in them could be.
I. Against the Word “Impact”
The problems that matter most don’t have answers that exist before the work begins. Answers in human systems are generated inside the work, by the people doing it, in contact with the situation. They are path-dependent, context-bound, and revisable.
The word “impact” implies that the answer existed before the work began, and that the work was the delivery of that answer against friction. But the work most of us are doing is generative. The answer forms inside the work, through collective sensemaking among people who do not initially agree, by holding uncertainty long enough for the situation to speak back and something generative and new to emerge.
To demand pre-specified indicators for that kind of problem is not just imprecise but incoherent, since it asks for the answer before the work that produces it has been done.
II. The Translation Tax
There is a cost paid by most grassroots organisations that is never accounted for. This is the translation tax: the cost of converting three-dimensional relational work into the two-dimensional governance artefacts that funders require.
It can look like the hours spent on logframes, or the senior leader who spent three days on a board deck instead of three days in the community. Why? Because what cannot be counted cannot be funded.
This tax is paid by the people closest to the work, who have the least slack to pay it. It is paid continuously, not once. And it is paid in the same currency the work depends on — in attention, presence, and cognitive bandwidth. The substrate is being burned to fuel the report on the substrate.
The reason this matters is that the tax exists because the funder cannot sit in the uncertainty the work demands. Generative work does not have answers before it begins. The answer forms inside the work, over time, in conditions of not-knowing that the grassroots leader has to hold long enough for something the situation calls for to emerge. This holding is not a side-effect; it is the work.
The funder is accountable upward to a board, trustees, audit committees, and peer foundations. Their institutional position does not permit them to remain in uncertainty; they must produce certainty for the people who fund and govern them. So they ask the grantee to produce a document that performs knowing on their behalf. The artefact is a regulation device that calms the funder’s nervous system by translating the unbearable openness of generative work into the bearable closure of measurable activity.
The grassroots leader, already holding the uncertainty of the work itself, is now also asked to absorb the funder’s inability to hold it. They become, structurally, the regulatory device for the entire chain above them. They hold the situation, they hold the funder’s discomfort with the situation, and they hold the board’s discomfort with the funder.
The cost compounds, and the person at the bottom of the chain pays it in their own body. This is what burns out the best people in the sector. Not the work itself, although the work is hard, but the work plus the regulatory absorption of an entire funding chain’s anxiety about it.
III. From Outcomes Produced to Conditions Tended
If “impact” is the wrong word because it presumes the answer existed before the work began, and if the translation tax is the device through which funders transfer their own inability to hold uncertainty onto grassroots leaders, the question that follows is what an accountability regime that did not require this transfer would look like.
I don’t have an answer yet, but I have a direction.
The unit of accountability has to shift from outcomes produced to conditions tended. Not “what impact did you have” but “what conditions did you tend, and what emerged that you could not have predicted.” This is closer to an ecology report than a logframe. It requires the funder to read for what is generative rather than what is attributable, and it requires the grantee to describe the field rather than the deliverable.
This is difficult because the funder’s institutional position usually does not permit it. A foundation programme officer who decides, individually, to drop the impact reporting requirement is taking on personal political risk within their own institution that they cannot easily offload. Most cannot sustain the position alone, and the practice quietly reverts within a year or two. The structural pressure sits on both the funder and the grantee, just distributed differently. The grantee pays the tax in cognitive labour, and the funder pays it in being unable to see what they are doing.
The reform required, then, is not at the level of methodology but at the level of what capital is permitted to hold. As long as philanthropic capital must produce upward legibility to boards, trustees, peer foundations, and the regulatory environment, it will demand downward legibility from grantees. The chain holds. Replacing “impact” with “effect” does not change who is regulating whom.
Until funders and the institutions behind them develop the capacity to sit with uncertainty themselves — until they treat their own discomfort with not-knowing as their problem rather than the grantee’s — the translation tax will continue to be paid.
And it will continue to be paid within the very structure that produces the work the funder claims to support.
This is uncomfortable for everyone, including funders who almost always want to do the right thing.
IV. The Russian Doll
Two pieces read together illuminate something that usually stays invisible.
One is on what makes a great philanthropy CEO: discernment, relational intelligence, storytelling. These CEOs spend 20% to 50% of their time on board and principal engagement, acting as translator and sometimes buffer between the principal and the staff, creating the organisational container in which the principal’s vision can grow and thrive.
The other is on trauma-informed leadership: how leaders absorb pressure that would otherwise destabilise the system, how the labour involved is real and metabolic, and how it cascades through three domains — the leader’s own regulation, the relational field around them, and the wider system they sit inside.
What both pieces are pointing at, in different vocabularies, is that senior leaders perform a kind of regulation labour for the systems they lead, which is largely invisible in conventional metrics, and which the systems above them rarely acknowledge as labour at all. Most senior leaders will feel this in their bodies without having the vocabulary for it.
What becomes visible when you hold them together is something close to a Russian doll. The CEO holds the board and the principal. The grassroots leader holds the CEO. The community holds the grassroots leader. At each level, someone is absorbing the dysregulation of the level above so that the level above can continue functioning without having to confront its own inability to hold uncertainty. The role celebrated as a virtue at one level produces the translation tax at the next. The shape is recursive, it cascades and amplifies as it goes down, and the people most depleted are the ones closest to the actual work.
No amount of better leadership at any single level fixes this. You cannot reform philanthropy by training better CEOs, because the CEO is not the person whose holding capacity is the binding constraint. The work that has to happen lies in the whole system’s capacity to sit with uncertainty and not knowing.
V. Changing the Shape of the Recursion
The fourth section articulated the recursive nature of the challenge. It also made me wonder whether the recursion itself could be part of the solution.
Some African societies have organised political and economic life on fractal principles — the same shape repeating at every scale, from household to compound to settlement to region. When a dispute arises, it can be settled within any of these overlapping units that the situation calls for, with the incentive at every level being cohesion rather than verdict.
Recursion is necessary to scale, however we define scale: at the nature of the problem, the nature of the container, the nature of the substrate. The question is what kind of recursion. There is the centralised version that pushes authority downward and extracts value upward. There is also a version where authority can be transmitted laterally and value can circulate back to the source that produces it.
A fractal structure will not guarantee fairness. The geometry is neutral. What matters is what is built into the recursion.
The argument I have been trying to make across this series is that the centralised extractive recursion is the one most of us are familiar with, that other designs do exist at scale, and that perhaps the work of reform is at the level of changing the shape of the recursion — not just optimising any single link inside it.
Originally written for LinkedIn as a six-part series, 13–29 April 2026. View originals: Preface · Part I · Part II · Part III · Part IV · Part V
Gautam John 
Leave a comment