n.V7-8.01 | WHY WE PUBLISH THE REGISTER
01| Everything at IkoCiti becomes public within 48 hours. That rule was not written for good news. Its only real test is what we do with the list of things that could destroy the platform — and this book part is that list. We publish it for a plain reason first: any serious funder will assemble it anyway, out of our own documents, in a room we are not sitting in. We would rather write it ourselves and be judged on the quality of our thinking about it than be handed it by someone else's analyst.
02| The second reason matters more. Big claims carry a duty. We claim that impact can be made self-perpetuating. We claim a barrio can carry its own rebuilding. We claim IkoCiti can become financially permanent. Claims that size are only honest if we state in advance what result would prove them false. A model that no evidence can kill is not a model. It is a belief, and beliefs do not belong in a business plan.
03| Risk and uncertainty are not the same word. A risk has a known shape: we can bound it, price it, buy against it. An uncertainty has no distribution we can see yet — we know the thing could happen and we cannot say how often, how hard, or under what conditions. Almost everything in this part is uncertainty. We use the word "register" because it is the standard term, not because we are pretending to actuarial precision we do not have.
04| What we refuse is mitigation theatre. In most risk registers, a serious threat is named in one sentence and then defused in the next by phrases like careful monitoring, adaptive management, or ongoing stakeholder engagement. That is vocabulary, not mitigation. A mitigation is a decision taken in advance and written down before it costs anything to take: a threshold, a named owner, an action that follows automatically, and a point at which we stop. Where we have that, we state it. Where we do not have it, we say we do not have it. There are places in this part where the honest answer is that the only mitigation available is the willingness to stop.
05| The ordering is deliberate. The conventional register sorts risks into model, market, operational, technology and financial — filing-cabinet logic, useful for auditors and useless for judgement, because it puts a threat to the core claim in the same weight class as a procurement delay. We sort by what the risk kills. Some risks kill the model: if they fire, IkoCiti is wrong and no amount of execution saves it. Some kill a CITI: painful, local, survivable. Some kill the clock: the model may be right and still die of slowness, because momentum runs out before proof arrives.
06| Four chapters follow. The loop may not replicate — the risk to the core claim. The wins may stay invisible — the risk to the clock. The barrio may reject or capture the platform — the risk to individual CITIs. And the money may not arrive, in any of the four ways it can fail to arrive. Technology risk is not given its own chapter; it is not a separate danger but a cost assumption, and it is priced where it belongs, inside the economics.
07| One finding sets the tone for everything after it. Eva Vivalt assembled 15,024 effect estimates from 635 studies across twenty intervention types and asked how much the same intervention varies from place to place. The answer is that development results scatter far more than we are used to in other fields:
"We use a new data set of 15,024 estimates from 635 papers on 20 types of interventions in international development, gathered in the course of meta-analysis, and find that the results reported are more heterogeneous than in other fields, such as medicine."— Eva Vivalt (How Much Can We Generalize from Impact Evaluations?, working paper version, p. 1; published in the Journal of the European Economic Association, 2020, Vol. 18, No. 6)
08| Read that against our own scaling case. IkoCiti's economics assume that what works in the first CITI works in the twentieth, at lower cost. Vivalt's data says that assumption is the exception rather than the rule, and that we should expect our own results to travel worse than we hope. The same paper carries a second finding that lands directly on our core market, since our long-run revenue runs through public budgets:
"Government-implemented programs also had smaller effect sizes than academic/NGO-implemented programs, even after controlling for sample size."— Eva Vivalt (How Much Can We Generalize from Impact Evaluations?, working paper version, p. 25)
09| We do not have a clean answer to that. Our design response — the platform carries the method, so quality does not depend on who employs the staff — is an argument, not evidence. It is exactly the kind of claim this register exists to keep visible until data settles it.
10| Finally, the register is a live document, not a chapter. Every risk in it gets an owner inside the organization, a metric that would show it firing, and a review date. When a risk retires because evidence killed it, we publish that too. The point of writing this down is not to look candid. It is so that three years from now, nobody at IkoCiti — least of all us — can quietly redefine what counts as failure.
