n.V1-9.07 | What Reaches the Settlement

01| Stand on a paved footpath in an upgraded section of Mathare, in Nairobi, and ask a set of ordinary questions. Who built this? What did it cost? Who decided it should be here rather than two streets over? Is there a plan for the parts of the settlement still under mud? How many people does the organisation that paid for it say it helped, and how would anyone know?

02| Every one of those questions has an answer somewhere. What is interesting is how far you have to travel to reach it, and what happens to the answer on the way.

03| Large development institutions produce five things: physical works, services, policy frameworks, publications, and numbers. The argument here is not that these outputs are fake. Most of them are real. It is that they are shaped for the people who pay for them, and that the further you get from the funder and the closer to the footpath, the thinner the accounting becomes.

The path is real

04| The Kenya Informal Settlements Improvement Project was approved in 2011, run by the Government of Kenya with financing from the World Bank's IDA, the Swedish development agency and the Agence Française de Développement. Its stated purpose was to improve living conditions in informal settlements by strengthening tenure security and building infrastructure based on community-developed plans. Its second phase has continued that work, including in Mathare: roads and footpaths, drainage, street lighting, water and sanitation.

05| None of this is invented. The works exist and can be walked. They are also partial in a way that is obvious on the ground — specific corridors were upgraded and the sections either side were not. That is not a scandal; money is finite and upgrading is sequenced. But it means the settlement contains, permanently, a visible line between the part that was reached and the part that was not, and no document a resident can obtain explains why the line falls where it does.

06| What is genuinely hard to find is the price. Aggregate figures are published. The cost of a specific stretch of paved path, the value of the contract, the firm that held it, and what comparable work cost in another settlement are not published in any form a resident could look up and compare. That is the plainest version of the problem: the thing is public, the accounting is not. A person can stand on an asset built with public money and have no route to finding out what it cost.

The numbers that describe the path

07| Now the outputs that travel furthest: the counts. The project reported its results against its targets, and reported them well. Some 262,780 people were recorded as benefiting from improved water sources, against a target of 162,000. About 125,525 were recorded as gaining improved tenure security, against a target of 100,000. Of twelve intermediate results indicators, all but one were fully achieved.

08| Look at what those sentences do. They are precise to the individual person. They exceed their targets. They are, as far as anything suggests, honestly compiled. And they cannot answer the question the resident on the footpath is asking, because "benefited from an improved water source" is a category, not an experience. It does not say whether the water arrives, at what pressure, at what hours, at what price, or whether it still arrives three years later.

09| A number can be accurate and still be about something other than the thing you want to know. That gap between precision and meaning is where the sector's numbers live. And there is a case that shows what happens to such numbers once they leave the building.

Kibera, and the number nobody could source

10| For roughly two decades Kibera — the settlement on the other side of Nairobi — was described in the international press, in NGO material and in United Nations documents as home to about one million people. Some accounts said two million. Mike Davis, in the best-known book on global slums, put it around 800,000. UN-Habitat's own published estimates ranged from 350,000 to a million, and in 2010 it was still offering 500,000 to 700,000.

11| The 2009 Kenyan census counted 170,070 people.

12| The gap is not a rounding error. It is a factor of five or six against the most repeated figure. Some of it is explained by genuine disagreement about where Kibera's boundaries lie, which is a real methodological problem in settlements without formal edges. The Map Kibera project, which mapped the settlement in detail with residents, arrived at 235,000 to 275,000 — higher than the census, far below the myth.

13| But the most damaging detail is not the discrepancy. It is what happened when researchers went looking for the origin of the million. They traced the figure through the media and institutional documents that carried it and could not find a source. It had been quoted everywhere, including by the United Nations, and attributed to nobody. It was passing from document to document on the strength of having appeared in the previous document.

14| That is the mechanism, caught in the open. Not fabrication — nobody invented the million to deceive. Repetition, in a system where a large number is more useful than a small one at every point in the chain. It justifies the programme, impresses the donor, warrants the press attention, and justifies the next programme. Nobody in that chain is rewarded for checking, and until 2009 nobody had to.

15| Set that beside the beneficiary counts and the question sharpens. If the population of a settlement could travel unsourced for twenty years, the number of people a project reached inside such settlements is not obviously more solid. There is no shared register across the organisations working in one settlement, no de-duplication between them, and no independent audit trail. A resident who receives a water connection from one project, a health outreach from a second and a training session from a third is counted three times, in three annual reports, by three organisations, each honestly.

Services, frameworks, publications

16| The other outputs fail the resident's questions in their own ways. Services are mostly not delivered by the institutions that fund them. Water comes through a utility, sanitation through utilities and small operators, health through a mix of public clinics, NGO facilities and private providers, schooling in large part through low-fee private schools that exist because public provision does not stretch. The financing and the reliability are connected loosely, through several other organisations. So when the tap fails, the chain back to the funder is long enough that nobody along it is accountable — which is worse than the earlier problem, because there is no single body to hold responsible.

17| Frameworks exist in abundance and at every altitude: the New Urban Agenda, the SDG 11 monitoring framework, national urban policy, the county development plan. Each is real and citable. None descends to the level of a specific settlement. A resident asking what the plan is for their settlement finds fragments scattered across documents written for different audiences on different timescales. The frameworks are said to cascade downward. What they demonstrably do is legitimise activity at the level where funding decisions are made.

18| Publications are the largest output by volume and the least reachable by the people they describe. The accumulated published material about Nairobi's settlements runs to many thousands of pages. The material written for the people living in them, in the languages they speak, through the channels they use, is a small fraction. This is not hypocrisy. It is a consequence of who commissioned the writing.

The reply, and why it fails

19| There is an obvious defence: settlements are genuinely hard to count. Boundaries are unclear, people move, structures change use, there is no register. If the numbers are bad, perhaps that is the nature of the problem.

20| The defence does not hold, and the reason is the most important thing in this chapter.

21| Muungano wa Wanavijiji, the Kenyan federation of residents of informal settlements, was formed in 1996 and now reports more than 64,000 members across some 300 settlements in fifteen counties. Working with Slum Dwellers International, it developed and has run at scale the practice of community-led enumeration and profiling: household surveys conducted by residents that identify and record every structure in a settlement and what it is used for. Not a sample. All of them.

22| The purpose was never only data. Enumeration was built to give a settlement a shared and defensible account of itself — one that could be put on a table in front of a county government, and that came from neither political interest nor the property market. It also changes what residents are in the transaction. A settlement that has counted itself is not a population waiting to be estimated by somebody else.

23| So the counting problem has a demonstrated solution, invented by the people being counted, running across hundreds of settlements for nearly thirty years. Which means the interesting question is not why settlements are hard to count. It is why, with that method available and proven, the numbers reported upward through the funding system are still produced the way they are.

24| Put the five outputs against the questions we opened with. The footpath is real and its price is not discoverable. The services are financed by one set of organisations and delivered by another, so failure has no address. The frameworks are real and operate above the level of the place they govern. The publications are extensive and addressed to other readers. The numbers are precise, exceed their targets, and cannot tell a resident whether the water runs.

25| None of this requires anyone to be dishonest. It requires only that the reporting line runs upward — that every output is finished, formatted and audited for the people who fund the work, and that no comparable obligation runs the other way. What reaches the settlement is the physical residue of a process whose actual product is a report delivered somewhere else. The Kibera number is the cleanest illustration available because it can be checked: wrong by a factor of five, circulating for twenty years through the most reputable institutions in the field, surviving because in the direction that mattered, it worked.