Every diplomatic agreement begins to decay the moment it is signed. The rate of decay is not random. It follows a pattern: commitments with specific language, public attribution, and enforcement mechanisms tied to existing institutions survive into implementation. Commitments with aspirational language, bilateral goodwill, and no operational anchor do not. They are not abandoned. They are simply not acted upon, which is a quieter version of the same thing.
The September 24 joint statement was unusually light on specificity. Trade language was principles. AI language was framework. The South China Sea was silence. The G20 imbalance question was deferred to Miami in December. A communiqué built on principles and frameworks decays faster than one built on mechanisms.
This post locks the indicators before observing the results. Four tracks. Specific signals. Dated deadlines. In December, the audit runs against this list.
Track One — Trade Enforcement
Window: 90 days | Deadline: December 23, 2026. Joint statement language: aspirational — no enforcement provisions. Baseline (Sep 28): no indicators present.
The joint statement contained no origin verification language, no bilateral customs data sharing provisions, and no enforcement mechanisms targeting transshipping routes. The 90-day window tests whether the absence reflected a deferred commitment or no commitment.
→ CBP guidance: US Customs and Border Protection issues updated certificate-of-origin guidance specifically referencing Vietnam, Malaysia, and Mexico as transshipping intermediaries. Publication in the Federal Register counts. An internal memo does not.
→ Entity list additions: Commerce BIS adds entities specifically identified as origin-laundering nodes — not end users of controlled technology, but transshipping intermediaries. The distinction is the signal.
→ Joint customs statement: Any joint statement from US Customs and China's General Administration of Customs on data sharing or coordinated enforcement. Working group announcement counts as partial; operational agreement counts as full.
Track Two — AI Governance
Window: 60 days | Deadline: November 23, 2026. Joint statement language: framework — working groups, renewed dialogue. Baseline (Sep 28): no indicators present.
The shorter window is deliberate. Convening a working group is logistically fast when political will exists. If the joint working group has not met by November 23, the AI governance commitments are in the same category as the 2015 Obama-Xi cyber agreement: announced publicly, never operationalized.
→ Working group convening: The joint AI safety working group holds its first formal meeting, both delegations present, public agenda on record. Announcement of intent does not count.
→ Incident reporting mechanism: Either government publishes a mechanism for notifying the other of AI incidents, with a named notification timeline. "We agree to communicate promptly" is aspirational. "72 hours after confirmed classification" is a mechanism.
→ Regulatory citation: Any US federal agency — NIST, NSF, AISI, OSTP — issues guidance or rulemaking that specifically cites the September 24 summit agreement as a basis for the action. Federal Register publication counts. Press briefing mention does not.
Track Three — South China Sea
Window: 90 days | Deadline: December 23, 2026. Joint statement language: silent. Baseline (Sep 28): AMTI data as of September 24.
Silence in a communiqué on an active bilateral issue can mean three things: a diplomatic trade-off, a back-channel de-escalation agreement, or no agreement. Behavior in the 90-day window distinguishes the last two.
→ AMTI operational tempo: Chinese Coast Guard vessel presence at Second Thomas Shoal, Scarborough Shoal, and the Paracels for the 90-day post-summit window vs. the 90-day pre-summit baseline. Increase: no behavioral change. Decrease of 15% or more: consistent with back-channel agreement. Flat: inconclusive.
→ PLA theater appointments: PLA Navy or Air Force command appointments in the Southern Theater Command within 90 days. Appointment of known hardline commanders is a behavioral signal inconsistent with diplomatic constraint. No major appointments, or appointment of commanders with dialogue records, is weakly positive.
→ US-Philippines and US-Vietnam incident reporting: Continued reporting of aggressive Chinese vessel encounters during the implementation window is behavioral disconfirmation of any informal de-escalation agreement.
Analytical note: This track is subject to confounds — domestic Chinese political cycles, US positioning, third-party incidents. The December scoring will flag these explicitly rather than forcing a clean read the data does not support.
Track Four — G20 Imbalance Language
Event: G20 Leaders' Summit, Miami | Date: December 15, 2026. Test: binary. Prior signal: China declined Asheville language (August 2026).
At Asheville, 19 of 20 G20 finance ministers signed language acknowledging large current account surpluses as a structural risk to global stability. China did not sign. The Miami Leaders' Summit produces its own communiqué. The question is whether China signs equivalent language at the Leaders' level that it refused at the Finance Ministers' level.
This is the most informative single signal available in the December window. Unlike trade enforcement, which can hide in agency guidance, and unlike AI governance, which can stall invisibly, the G20 communiqué is public, binary, and attributed.
→ China signs: The Miami communiqué contains language acknowledging surplus risks and China is among the signatories. This is the clearest available signal that September 24 moved something structural on the core trade imbalance question.
→ China does not sign: The communiqué omits the language entirely (to preserve consensus) or China again declines. The structural position is unchanged. The summit produced no movement on the issue underlying every other trade mechanism.
The decay rate of a communiqué is not a metaphor. It is observable. In November the Track Two window closes. In December the Trade, South China Sea, and G20 tracks close together. Each indicator above will have a reading by then — present, absent, or inconclusive — and the audit runs against this post, not against whatever seems most significant at the time.
The record is maintained at rhodesresearch.org/forecasting-record. A communiqué has a half-life. The clock started September 24.
J. Ken Rhodes writes through Rhodes Research (rhodesresearch.org) on Chinese economic statecraft, dollar architecture, and institutional dysfunction.