Update log
- 7 September 2026 - First edition. Drawn from the architecture programme's source-classification chapter and concept contract, the era and power map, and the portal's own site plan. Axis names and the shape of their ranges are given; the full enumerations are not reproduced.
Short answer
Most roadmaps use one word per item: planned, in progress, done. The architecture programme identifies that habit as an error it had itself committed in an earlier revision - it "incorrectly mixed implementation, deployment, readiness, governance, truth ownership, and knowledge state" - and replaces the single word with six axes, each answering a different question about the same statement:
- Implementation stage - how far it exists, from absent through doctrine, specification and prototype to implemented.
- Deployment scope - where it runs, from nowhere through local, single durable node, distributed simulation, private network, hostile devnet and public testnet to mainnet.
- Governance state - what has been decided about it, from unratified through candidate, ratified, implementation-authorised and activation-eligible to activated - or suspended, revoked, deprecated, superseded, rejected, reserved, appealed.
- Truth role - what kind of thing it is: canonical protocol, canonical state, governance, policy, evidence, external claim, projection, operational, ephemeral, product view, or research.
- Epistemic status - what is known about it: declared, observed, derived, established, unestablished, or conflicted.
- Assurance level - how strongly it has been checked, from unassessed through self-tested, internally verified and independently reproduced to adversarially validated and operationally observed.
The rule that binds them is one sentence: "Do not compress the axes into one status." Earlier one-dimensional maturity labels "may exist only as documented compatibility projections with explicit mapping and information-loss warnings." The portal's vocabulary - research, implemented locally, in development, planned, reserved, and the claim scope proven today, in development, later gates, not claimed - is exactly such a projection, and this dossier says what it loses.
What is established, what is in development, what is not claimed
| Claim | Status | Basis | Limit |
|---|---|---|---|
| The line practises the discipline the grammar formalises: every crossing states what it admits, what it denies and what it does not claim | Proven today | Crossing acceptance records; era and power map | Practised in records, not yet emitted as six-axis fields |
| The portal carries an explicit status on every product, module and claim, drawn from a fixed vocabulary | Proven today | Site plan, truth discipline | A five-word projection of the six axes, with information loss |
| Six source axes with a legal combination matrix and machine-detectable illegal combinations | Research | Architecture programme, source classification | Design of the programme's own record-keeping |
| A concept contract for every named primitive: power delta, reachable consequences, properties, explicit non-claims, six-axis status, acceptance gates, falsifiers, supersession rule | Research | Architecture programme, concept registry | Design; "names without this contract are rejected from the active architecture" |
| Earlier maturity labels kept only as documented projections with information-loss warnings | Research | Architecture programme | Design |
| A public, machine-readable status feed for outsiders | Later gate | Adoption ladder | Not offered today |
| A statement anywhere on this portal that reaches mainnet on the deployment axis, or operationally observed on the assurance axis | Not claimed | Every page | None exists |
Why six questions instead of one word
A single status answers "how done is it?" - and hides five other questions inside the answer. Consider a sentence the portal could have written: "reconstruction is done." On the six axes, the honest classification of the reconstruction family is: implemented; deployed locally; ratified and activated on the line; evidence, not canonical state; established relative to the cases exercised; internally verified, with independent reproduction as a later gate. Six answers, none of them "done".
one word six axes
+--------------+ +-----------------------------------------+
| done | loses --> | implementation ... implemented |
+--------------+ | deployment ....... local |
| governance ....... activated (the line) |
| truth role ....... evidence |
| epistemic ........ established (scoped) |
| assurance ........ internally verified |
+-----------------------------------------+
Figure 1 - The same fact, once compressed and once classified. The right-hand side is what an evaluator can check.
The programme's own reason is stated in its history: an earlier revision had mixed the questions, and the mixture made illegal states expressible. Restoring the meaning of the truth axis and "separating the other questions" is presented not as bureaucracy but as a repair.
The legal combination matrix
The point of separate axes is that some combinations are impossible, and the grammar wants those impossibilities detected by a machine rather than argued about by people. The programme gives examples, and each one is a sentence a careless project would happily write:
- An absent implementation requires a deployment scope of none. Nothing can be deployed that does not exist.
- Mainnet does not imply operationally observed. Being deployed is not the same as having been watched behaving under real conditions.
- Canonical state requires activated governance. A fact cannot be canonical because someone implemented it; it is canonical because the governance axis says so.
- An external claim never becomes canonical state by stronger assurance. A bank's receipt, however well attested, stays an external claim; assurance moves along its own axis and cannot change what kind of truth a thing is.
- "Established" is always relative to an explicit subject and method. There is no free-standing "it works".
illegal by construction (examples)
+----------------------------------------------------------------+
| implementation = absent AND deployment != none |
| truth role = canonical state AND governance != activated |
| truth role = external claim --> canonical state (by any path) |
| epistemic = established without subject and method |
+----------------------------------------------------------------+
Figure 2 - Combinations the grammar refuses. An evaluator does not have to catch them; the record cannot express them.
The fourth rule is the one institutions should notice. It is the six-axis form of the settlement-truth law described in the evidence-objects dossier: a rail's delivery message is evidence about the world, and no amount of signing turns it into protocol state.
The concept contract
The grammar reaches below statements to names. Every primitive the programme admits must carry a contract: what power it adds, what consequences become reachable, which safety, liveness and privacy properties it claims, what it explicitly does not claim, its six-axis status, the trigger that would start implementation, the gates that would accept it, the falsifiers that would reject it, and the rule by which it may be superseded. "Names without this contract are rejected from the active architecture rather than left for Builder invention."
This is why the programme carries a table of near-duplicate names it has adjudicated - a portable denial protocol against refusal portability, a contradiction escrow against a conflict isolation cell, a proof economy against a proof budget - and records for each which is the canonical concept and which is a historical label. A name that could mean two things on the six axes is a name that will eventually be used to claim the stronger one.
The portal's projection
This portal uses five status words and a four-part claim scope. Under the grammar those are a compatibility projection, and honesty requires saying what the projection loses.
portal word six-axis reading (typical) information lost
-------------------- --------------------------------------- --------------------------
implemented locally implemented . local . activated (line) assurance level, epistemic
. evidence or canonical protocol scope of "established"
in development prototyped or specified . none governance state
planned doctrine or specified . none trigger and gates
research doctrine . none . candidate falsifiers
reserved absent . none . reserved nothing: this one is exact
Figure 3 - The portal's vocabulary as a projection. Each row is honest; each row is lossy, and the dossiers exist to restore what the word drops.
Two consequences are already visible on the portal. The claim tables in every dossier are a partial restoration of the axes: the Basis column carries the assurance axis, the Limit column carries the epistemic scope. And the evidence snapshot on the home page states its subject and method - which run, which merge, which census - because "established" without them is not a permitted state.
What the discipline costs
The programme does not pretend the grammar is free. It requires the classification for every source and every primitive, a schema for every machine-readable artifact, and a closed-world observation for every absence claim; its evidence-complexity budget counts, per wave, the number of owners, cross-references, alias ambiguities, state machines, unexecutable invariants, parsers and reviewer effort, and publishes the delta. Six axes are more expensive than one word. The programme's position is that the expense is where the honesty lives, and that a scalar composite "may be a display aid only; raw dimensions remain governing."
claim --> six-axis classification --> legal-matrix check --> record
| |
| +--> illegal: rejected, not argued
+--> concept contract required for every new name
Figure 4 - Where the cost is paid. Every claim passes the same gate; none is exempt because it sounds finished.
What the six axes do not do
- They do not make anything more implemented, deployed or assured; they describe.
- They are the programme's design for its own records; the line's current records practise the discipline in prose.
- The portal's five words remain a projection; where precision matters, read the dossier's claim table, not the badge.
- Nothing on this portal reaches mainnet on the deployment axis or operationally observed on the assurance axis, and the grammar would flag any sentence that implied it.
Verdict
The six axes are the architecture's refusal to say "done". They separate how far a thing exists from where it runs, from what has been decided about it, from what kind of truth it is, from what is known, from how hard it has been checked - and they make the tempting shortcuts, an external claim promoted to state, a deployment mistaken for observation, unexpressible rather than merely discouraged. The portal speaks a five-word dialect of that grammar, and says so.
Frequently asked questions
Is "implemented locally" a good status or a bad one?
It is an accurate one. On the six axes it reads: implemented, local, activated on the line, internally verified. What it withholds is as informative as what it grants: no network, no external reproduction yet.
Why can't a bank's confirmation become protocol state?
Because the truth-role axis is not the assurance axis. A confirmation is an external claim; more signatures make it better evidence, never a different kind of thing. The settlement-truth laws follow from this.
Does the line already use six-axis fields?
No. The crossing records practise the separation in prose - what is admitted, what is denied, what is not claimed - and the programme's design is to make it a machine-checked grammar.
What does "established" mean here?
Known, relative to a stated subject and a stated method. The evidence snapshot on the home page is the portal's example: a number, a run, a merge, a date.
Why does the portal not use all six axes?
Because a public page needs a short vocabulary. The five words are a documented projection; the dossiers restore what they drop.
Sources and methodology
Drawn from the architecture programme's source-classification chapter, its legal combination matrix, its concept contract and alias adjudications, and the era and power map, together with the portal's site plan. Design claims carry the status research; the portal's own practice is described as practised.
The programme is not published on this portal and is not reproduced here; axis names and the shape of their ranges are reported, the full enumerations are not. Figures shown anywhere on this portal come from a single dated snapshot; none are introduced by this dossier.
Related reading: The acceptance discipline; Receipts, denials and capsules for the settlement-truth law the fourth rule protects; Sealed history.