n.V6-2.04 | The Core and the Satellites
01| There is one global platform. There are many local CITIs. Almost every argument in this volume comes back to what belongs to which.
02| Social progress is local and personal. It happens on a specific street, between people who know each other's mothers. Nothing about that is standardisable, and we should not try. But the machinery underneath it — the training, the exchange, the project pipeline, the quality bar, the accounting — has no reason to be different in Bogotá than in Lagos. Building it twice would be stupid. Building it fifty times would be fatal.
03| So: one core, built once, replicated. Many satellites, each embedded in a place we do not control and do not fully understand.
04| What is identical everywhere. The way a project is planned, listed, funded, implemented and followed up. The quality standards a project has to clear. The training path from rookie to Fellow. The way contributions are recorded and rewarded. The rules of the exchange. These are the platform. A CITI that changes them is not a CITI.
05| What is local. Language and currency, obviously. Which components a CITI switches on, because not every barrio needs everything on day one. And everything that is genuinely about the place: which neighbourhoods, which problems, which projects, which partners, which fights are worth having. That is not a small allowance. That is the entire content of the work. We standardise the how and leave the what alone.
06| Why this is the whole economic case. Once the core exists, adding a CITI costs a fraction of building one. The heavy investment amortises across a growing network instead of being repeated. Learning flows in one direction and accumulates — what CITI number four discovers about recruitment improves CITIs one through three and every CITI after. Fifty separate systems would produce fifty separate lessons that never meet. This is not a cost-saving preference. It is the difference between a model that can reach scale and a model that cannot.
07| What we want. A core so standard that a new CITI is a configuration rather than a construction project.
08| What that takes. Saying no, constantly, to reasonable people. Every local team will have a genuinely good reason why their situation requires an exception. Most of those reasons will be correct on their own terms. Consensus on this is unanimous and unsentimental: platforms that grant local exceptions early end up maintaining a dozen incompatible versions, and the cost of every future change multiplies by the number of variants.
09| What we sacrifice. Fit. Something in the standard core will be wrong for every single CITI — irritating, badly matched to how things work there, obviously improvable by anyone standing on that street. We are going to tell them no anyway. That is a real loss of quality at the local level, traded for the ability to exist at all at the network level, and local teams will resent it. They will be right to. I would rather have that argument every year than run fifty codebases.
10| White-label is a different animal. Licensing the technology to outside organisations who want their own version — customised, rebranded, under their own philosophy — breaks every property above. Their changes cannot amortise. Their learning does not come back. Their quality is not ours to control but is attached to our work. It is a legitimate revenue line and V7 argues it seriously. It is not CITI expansion, and it does not start until the network has proven the loop works. Reversing that order would dilute the proof at exactly the moment the proof is everything.
11| The technical question — how you build a core that localises cleanly without forking — is not mine to answer. The user-side call is: identical core, local content, and the CITI does not get to change the machinery because it does not like the shape of it.
