Before you read

Every foundational document has one sentence it usually leaves out: who wrote it.

Everything you read here was proposed by the founding group

The architecture, the compliance mechanism, the nine roles, every economic number, even the way nine missions are divided into three sets — all of it was built and proposed by the founding group. Not one line in this document has been through a community vote.

Saying that is necessary. Without it, a reader will assume this is the truth of the system. It is not. It is what one group of people chose, written down so that others can read it, argue with it, and change it.

A small group has to begin, because something must exist before there is anything to discuss. The difference between beginning and owning lies in exactly one place: whether the group that began will let others change it, how far they may change it, and until when.

Four labels, used consistently throughout

From here on, everything this document mentions belongs to exactly one of four states. Reading a claim without reading its label is the easiest way to mistake a proposal for a commitment.

The shortened reading rule: anything not marked ENGRAVED can be changed. And the ENGRAVED list is short on purpose — with a long engraved list, "the community decides" is only a figure of speech.

Figure 2 — The path of a decision. The gate in the middle is where the project blocks itself: see Nine roles.

Nine steps, and only one that engraves

9Chain's roadmap has nine steps — and it is the roadmap of the project itself, not an abridged version made for outsiders.

The backbone of those nine steps is still four periods: Testnet 1 → Testnet 2 → Testnet 3 → Mainnet (mainnet is also called the official network; the two words mean the same thing). The first three steps are the run-up to Testnet 1 — the period of trial builds and of making a place for the community to stand; the middle three are the three test periods; the last three are close the window, engrave, then let go.

Two places in this order deserve a closer reading than the rest.

Step 3 comes before step 4. The voting mechanism must exist before the three test periods begin, because otherwise the phrase "the community votes" across those periods is only a figure of speech: every decision would be settled before anyone could cast a ballot. This ordering is a commitment, not a scheduling detail.

Step 9 comes after step 8. This is where this roadmap differs from most others: the day the official network launches is not the finish line. At the moment of engraving, power is still more concentrated than it ought to be — saying otherwise would be a lie. Full decentralisation is therefore a step that comes afterwards, and it is the only step that does not end with a launch event: it is done when no task requires an administrative key any more, and that is a countable number (see The commitment to remove the scaffolding in Operations and governance).

Which step things stand at is a live fact, so it is not written here — how to read it for yourself is in "Check the network yourself".

This is the most important thing to carry into the rest of the document: not one line here is engraved before mainnet. The three testnet periods exist precisely so that what is written here can be tested, argued with, and corrected — by community vote, with almost no clause off the table. The single exception is the core, reduced to its minimum: the three memorial blocks that open the ledger, and the rule itself that everything else is decided by the community together (see LOVE9 economics).

Mainnet = Testnet 1 + Testnet 2 + Testnet 3 + Mainnet

The three test periods are not thrown away when the next one begins. Their history — who did what, who contributed what, who was verified how — is preserved and attached to the shared ledger of mainnet. Mainnet is therefore not a ledger opening on a blank page, but the sum of all four periods.

⚠️ HISTORY is preserved, not a balance promised. The inherited genesis principle stands intact: no period promises a one-for-one balance. What testnet contribution history converts into — or whether it converts at all — is marked LEFT BLANK, and mainnet is where the community votes on it. Anyone telling you that running a testnet will certainly convert into tokens is putting words in the project's mouth — nobody has decided that.

There is one more boundary to state fully, because it is easily read as something it is not. A test period can be wiped and rebuilt from scratch — that is exactly what the three testnet periods exist to do. When it happens, balances, addresses, contracts and block history of the abandoned build do not flow into the next one; do not build anything that depends on a test network living forever. The inheritance promise takes effect from the genesis of mainnet.

Preservation only means something if outsiders can check it, so the rule that comes with it is a countable action: before a network is rebuilt, a complete copy of the old ledger and its hash are published. Anyone can download it, hash it again, and reconcile what happened — even long after that network has stopped running. What carries over is a published contribution record with its hash, not a balance.

The window that is still open, and when it closes

Changing a proposal is cheap while it is still words. It gets more expensive once it becomes code. And it is all but unchangeable once it enters the genesis of mainnet — a wrong genesis cannot be patched, it can only be rebuilt as a whole network.

So the most valuable stretch in the whole life of this project is the one from now until mainnet: the stretch where a stranger's question can still change the design. The three testnet periods are three times that window is reopened.

The window closes at the genesis of mainnet, not at the genesis of a testnet — and a genesis should only happen after passing the gates in Check the network yourself. Within the nine steps, the closing point has a clear name: step 7, concluding the votes — after that comes engraving. Put another way, the window stays open exactly as wide as the remaining distance to level four — and you can measure that distance yourself, without asking anyone.

Three things the founding group binds itself to

Never ratify anything permanent with its own votes. Distributing voting power to the point where the project loses its veto must come before any ratification of anything permanent. The full reasoning is in Nine roles — the most candid page in this document.

State what does not work yet right beside what does, in the same table, so that nobody has to join two separate lists together.

Measure the promise to withdraw with a countable number: how many tasks still require an administrative key. If that number goes down, the promise is real; if it stands still, the promise is only words.

In short: a project that calls itself community-owned must answer two questions, not one: what can the community change, and what can no one change any more. The four labels above answer both — and you should hold that same table up against every new thing we say.