Advantage / no-advantage language protocol (BL-65)

This page is the operator-facing guide for when “advantage” language is allowed. Default posture is no-advantage / research observation. Decisive protocol modules remain claim-gated evidence paths — never invent-green marketing.

Related surfaces:

• Existing runners (compose, do not fork):
  benchmarks.advantage_protocol, benchmarks.decisive_advantage_protocol,
  forecasting.neural_operator_advantage
• BL-52 / BL-53 governance: Governed route matrix,
  Unsuitable scenario registry
• Module: scpn_quantum_control.advantage_language_protocol

Rules

Language status	Meaning
no_advantage_default	Default public posture; no marketing advantage triggers
research_observation	Claim-bounded research wording only (no marketing triggers)
decisive_gated	May enter decisive evidence path; not a proven advantage claim
refuse_advantage_language	Explicit refuse path for ungoverned advantage wording

Default: any free-text claim containing marketing advantage triggers without a bound protocol id is refused.

Claim boundary:

advantage-language governance only; default is no-advantage / research observation; decisive or advantage wording requires an explicit protocol identity and never invents green quantum-advantage marketing claims

Public API

from scpn_quantum_control.advantage_language_protocol import (
    issue_no_advantage_certificate,
    probe_advantage_language,
    get_advantage_protocol,
    build_advantage_language_registry,
    assert_advantage_language_registry_integrity,
)

# Default certificate (always non-promotional)
cert = issue_no_advantage_certificate(context="release_notes")
assert cert.language_status == "no_advantage_default"

# Ungoverned advantage language fails closed
blocked = probe_advantage_language("We claim quantum advantage.")
assert blocked.allowed is False

# Neutral wording allowed under default
ok = probe_advantage_language("Local residual audit under claim_boundary.")
assert ok.allowed is True

# Decisive protocol may enter evidence path (not invent-green)
decisive = probe_advantage_language(
    "Entering decisive Kuramoto-XY quantum advantage question.",
    protocol_id="protocol:decisive.kuramoto_xy",
)
assert decisive.language_status == "decisive_gated"
assert decisive.allowed is True  # enter path only — not a proven claim

registry = assert_advantage_language_registry_integrity(
    build_advantage_language_registry()
)

API reference

All public objects are exported by scpn_quantum_control.advantage_language_protocol. Registry and to_dict() payloads contain JSON-ready primitives.

Types and constants

API	Contract
ProtocolLanguageStatus	Literal status: no_advantage_default, research_observation, decisive_gated, or refuse_advantage_language.
ADVANTAGE_LANGUAGE_PROTOCOL_SCHEMA	Stable registry schema identifier, currently advantage_language_protocol.v1.
ADVANTAGE_LANGUAGE_CLAIM_BOUNDARY	Shared non-promotional boundary copied into records, certificates, probes, and registries.

Data models

AdvantageProtocolRecord

Frozen, slotted catalogue row containing the protocol id, language status, summary, evidence-module pointers, reason, and claim boundary. The default row must have an empty reason; every non-default row requires one. Evidence-module entries must be non-blank. to_dict() serialises the tuple as a list.

NoAdvantageCertificate

Frozen, slotted non-promotional certificate with a deterministic id, statement, bound protocol, and claim boundary. Its status is always no_advantage_default. Construction rejects blank fields or any other status; to_dict() returns the complete certificate.

AdvantageLanguageProbeResult

Frozen, slotted language-gate decision containing the original claim, allowed flag, effective status, protocol id, matched trigger phrases, reason, and claim boundary. A refuse-status result cannot be allowed, and every trigger must be non-blank. to_dict() converts triggers to a list.

Catalogue access

Function	Parameters	Returns	Failure behavior
list_advantage_protocol_ids()	None	Protocol ids in canonical order.	No failure for the built-in catalogue.
get_advantage_protocol(protocol_id)	Non-blank protocol id.	Matching AdvantageProtocolRecord.	Raises ValueError for blank or unknown ids.
iter_advantage_protocols(*, language_status=None)	Optional status filter.	Stable tuple of matching records.	Returns an empty tuple when no row matches.

The catalogue also includes protocol:bl79.initial_state_observation, a research-observation row for population-matched initial-state comparisons. Catalogue evidence modules are pointers; their presence is not proof that their protocol gates passed.

Certificates and trigger detection

issue_no_advantage_certificate(*, context="public_claim", protocol_id="protocol:default.no_advantage") normalises spaces in context to underscores and returns a deterministic certificate. The protocol must be known and cannot be the refuse row. Blank contexts, unknown ids, and refuse-row bindings raise ValueError.

find_advantage_language_triggers(claim_text) performs a case-insensitive ordered scan for the module's governed marketing phrases. It returns a tuple of matched phrases and raises TypeError for None.

from scpn_quantum_control.advantage_language_protocol import (
    find_advantage_language_triggers,
    issue_no_advantage_certificate,
)

certificate = issue_no_advantage_certificate(context="release notes")
assert certificate.certificate_id.startswith("no_advantage:release_notes:")
assert find_advantage_language_triggers("Proven quantum advantage")

Language probe

probe_advantage_language(claim_text, protocol_id=None, *, unknown_policy="raise") applies these rules:

Binding	Neutral/research wording	Trigger wording
No protocol or blank id	Allowed under default posture	Refused through the ungoverned row
no_advantage_default	Allowed	Refused
research_observation	Allowed	Refused
decisive_gated	Allowed into the evidence path	Allowed into the evidence path, not declared proven
refuse_advantage_language	Refused	Refused

Unknown protocol ids raise ValueError by default. With unknown_policy="refuse", they produce a structured refusal instead. Any other policy value raises ValueError; a None claim raises TypeError.

Registry and integrity

build_advantage_language_registry() assembles the schema, claim boundary, protocol count, per-status counts, blank count, and canonical records. assert_advantage_language_registry_integrity(payload=None) validates a supplied registry or builds the canonical one. It raises ValueError for an empty or malformed protocol list, blank/unknown statuses, missing reasons on non-default rows, or inconsistent blank/protocol counts.

Safety and side effects

• All APIs are pure and deterministic for their supplied inputs.
• The module does not run benchmarks or QPU/provider jobs, access credentials, mutate evidence, publish wording, or promote a claim.
• decisive_gated means only that wording may enter an explicit evidence path; it never means quantum advantage has been demonstrated.
• A certificate governs wording only. It is not scientific, performance, release, legal, or marketing approval.

Catalogue seeds

• protocol:default.no_advantage — default posture
• protocol:s2.scaling_matrix — research observation (S2)
• protocol:decisive.kuramoto_xy — decisive_gated (decisive module)
• protocol:neural_operator.structural_surrogate — classical surrogate research
• protocol:ungoverned.advantage_language — refuse catch-all

Bounded product status (BL-65)

Shipped: S65.0 catalogue · S65.1 facade compose · S65.2 no-advantage certificate · language-gate probe (partial S65.4).

Open: S65.3 full decisive evidence schema wiring · S65.5 BL-32 scoring hook · S65.6 BL-55 reproduction fixtures · fuller promotion-tool hooks.

Authored by Anulum Fortis & Arcane Sapience (protoscience@anulum.li)