-
Status: Active, v1.0, 2026-05-28
-
License: PMPL-1.0-or-later (MPL-2.0 automatic legal fallback)
-
Companion to: Toolchain Readiness Grades (TRG) v1.0, Component Readiness Grades (CRG) v2.2, Foundations Readiness Grades (FRG) v1.0
-
Part of: Rhodium Standard Repositories (RSR)
ARG grades the language-as-an-adoptable-artefact. It sits at the language-whole tier of the estate grading hierarchy:
-
CRG grades individual components.
-
TRG grades the toolchain.
-
ARG grades the language as a thing someone can adopt.
-
FRG grades the mathematical foundations of the language.
ARG uses the same letter grades as CRG and TRG (X | F | E | D | C | B | A)
and the same worst-of-any-required-element aggregation rule. The evidence
required at each grade asks a different question: "would anyone use this?"
Three reasons:
-
CRG grades components, TRG grades the toolchain — neither answers "is this language something a person can adopt?". That question belongs at the language-whole tier.
-
The hyperpolymath estate has 17+ languages in various states of maturity. ARG provides a uniform yardstick for comparing them on the adoption axis, independent of TRG or FRG progress.
-
The existing
READINESS.mdper-repo convention (KRL is the canonical example) can express component grades cleanly but has no rubric for language adoption as a whole. ARG provides that rubric.
| File | Purpose |
|---|---|
|
The normative spec. Read this first. |
|
This file. |
|
Machine-readable counterpart of the spec. |
|
ARG applied to itself (insofar as that’s meaningful — ARG grades languages, not specs; the doc set is self-assessed against internal-consistency). |
|
The per-language profile template referenced from §1.1 of the spec.
Each adopting language publishes |
| Grade | Meaning |
|---|---|
X |
Not yet a language |
F |
Released but actively user-harmful (rare) |
E |
Bare minimum: qualifies as a language, runs own stdlib. Nobody outside the maintainer would actually use it. |
D |
Toy-project usable: a weekend project is feasible |
C |
Alpha gate: self-validated, ≥10 dogfood users, ≥2 substantive domains |
B |
Beta gate: broadly adopted, ≥100 external users, ≥6 diverse projects, ecosystem-plugged |
A |
Stable: school-grade real (first-gen-Python equivalent), ≥1000 users, founder bus-factor passed |
Loss of any A-grade gate (founder steps away without replacement, educator withdraws, distinct-user count drops, FRG regresses, TRG drops, 6+6 signoff rescinded) immediately demotes the language. There is no grace period. ARG ≤ TRG always — if the toolchain regresses, adoption regresses with it. See spec §4.8.
See spec §11. Honest answer: ARG-A is calibrated against first-generation Python (c. 1995-2002) and modern Rust. The author expects approximately zero hyperpolymath languages to hit ARG-A in year one. That is the correct order of magnitude. If ARG-A becomes easy to hit, the bar is wrong and the standard MUST be raised.
ARG is the baseline. Each hyperpolymath language is expected to publish
a spec/ARG-PROFILE.adoc in its own repo that tightens ARG with
language-specific obligations. A profile may only tighten, never loosen.
See spec §1.1 and templates/ARG-PROFILE-TEMPLATE.adoc.
-
vs. CRG: CRG grades components inside the language project (a formatter, an LSP server, a stdlib module). ARG grades the language as a whole. Where CRG and ARG would both apply, both grades are published independently. ARG ≤ worst-of(CRG over all required adoption-surface components).
-
vs. TRG: TRG grades the toolchain. ARG asks whether anyone would adopt the language. ARG ≤ TRG always.
-
vs. FRG: FRG grades the mathematical foundations of the language (preservation theorems, soundness, the 10-level TypeLL progression). Independent of ARG, except that ARG-A requires FRG ≥ B.
-
vs. RSR: RSR grades repository hygiene. ARG-graded language repos MUST be RSR-compliant at the level required by TRG at the same grade.