There is a particular kind of power that does not announce itself. It watches. It learns the terrain. It secures its footing. It develops capability before making claims. Then, when conditions require action, it moves from preparation rather than impulse.
This is the useful core I take from the strategy commonly associated with Deng Xiaoping and often called the Chinese “24-character strategy.” English translations vary, but its recurring ideas are disciplined observation, a secure position, calm response, unadvertised capability, patience, modesty, and restraint in claiming leadership.
I do not import that doctrine whole, nor do I treat statecraft as a template for democratic institutions. The American-modern version must be transparent where public trust requires transparency, accountable where power affects people, and ambitious in service of human freedom. But the deeper lesson survives translation: do not confuse performance with readiness, visibility with value, or speed with direction.
From strategic concealment to quiet capability
In an American enterprise, “hide capability” should never mean hiding risk, bypassing oversight, or deceiving stakeholders. It means resisting premature claims. Build the prototype. Test the controls. Strengthen the team. Measure the outcome. Let evidence mature before the press release.
That is quiet capability: competence without theater. It is especially important in artificial intelligence, where the distance between a demonstration and a dependable operating system remains enormous.
| Strategic idea | American-modern interpretation | Technology practice |
|---|---|---|
| Observe calmly | Begin with evidence, not fashion | Map users, data, workflows, dependencies, threats, and consequences |
| Secure the position | Establish legitimacy and resilience | Name owners, classify data, control identity, fund recovery, and define authority |
| Handle change with composure | Use disciplined incident and change management | Stage releases, monitor behavior, preserve rollback, and rehearse response |
| Build capability quietly | Prove before promising | Prototype in bounded environments and promote only with evidence |
| Practice patience and modesty | Prefer service over spectacle | Measure mission outcomes rather than model novelty |
| Do not rush to command | Earn leadership through trust | Use open standards, shared governance, and portable architecture |
The Lujan operating doctrine
Across my work and enterprises—from public-sector modernization and critical-infrastructure security to Avanyu AI Architects, ZiaWolf, HYDROS, the Water Scribe, and the emerging Mythos concept—the method is consistent.
Observe before automating
Walk the real process. Listen to the person doing the work. Find the handwritten exception, the spreadsheet nobody documents, the approval that exists because something once failed. AI introduced before this observation does not remove disorder; it accelerates it.
Secure the foundation before scaling
Identity, data stewardship, records, privacy, procurement, architecture, accessibility, continuity, and cybersecurity are not cleanup activities. They form the load-bearing structure. A powerful model resting on weak institutional controls is not transformation. It is leverage applied to fragility.
Develop depth without performing certainty
Leaders should be candid about uncertainty while relentlessly improving readiness. We can experiment boldly inside bounded environments, keep humans accountable for consequential decisions, and require evidence before autonomy expands.
Lead by making others more capable
The goal is not to become the loudest technologist in the room. It is to build a system in which staff understand the mission, decisions are visible, knowledge survives turnover, and the institution can operate safely without dependence on one personality or vendor.
Mythos: narrative above, controls below
Mythos is an appropriate name for a new class of AI development because every institution already runs on stories. Policies tell stories about authority. Data tells stories about the past. Models construct probable stories about what comes next. The danger begins when a compelling machine-generated narrative is mistaken for truth, mandate, or permission.
Mythos therefore should not be designed as an oracle. It should be designed as a governed intelligence layer: a system that can gather institutional memory, connect evidence, expose patterns, simulate options, and help people reason—while clearly showing sources, confidence, permissions, and the boundary between recommendation and decision.
The architecture follows a simple rule: as above, so below.
Above the system are mission, law, ethics, public purpose, executive accountability, and human judgment. Below are identity, data classification, retrieval, models, tools, logs, approvals, network controls, and recovery mechanisms. The lower technical order must reflect the higher human order. If accountability is unclear above, autonomy will become unsafe below. If the mission is confused above, the model will optimize confusion below.
A governed pattern for emerging AI
- Observe: document the mission, users, process, data, affected communities, failure modes, and current evidence.
- Position: assign accountable owners, risk tier, approved models, permissions, data boundaries, and stop conditions.
- Prototype quietly: test inside a sandbox with synthetic or appropriately protected data and no unnecessary production authority.
- Measure honestly: evaluate accuracy, security, privacy, bias, accessibility, cost, latency, human workload, and mission value.
- Promote gradually: expand access and autonomy only when evidence justifies it, with rollback and human approval proportional to consequence.
- Remain watchful: monitor model drift, tool behavior, data changes, exceptions, emerging threats, provider risk, and whether the system still serves its purpose.
This pattern aligns naturally with the NIST AI Risk Management Framework and NIST Cybersecurity Framework 2.0. Govern sets the higher order. Map and Identify establish awareness. Measure, Protect, and Detect establish evidence. Manage, Respond, and Recover preserve mission authority when reality diverges from the plan.
Modernism with a human center
American modernism at its best is not decoration. It is structural honesty: reveal the material, remove the unnecessary, make function legible, and build for the life that occurs inside. That is also how trustworthy AI should be built.
The system should reveal who owns it, what it knows, what it may do, how it reached a conclusion, where uncertainty remains, and how a person can challenge or stop it. Elegant architecture is not merely clean code. It is clear authority.
New Mexico adds another truth. The desert teaches economy. Water teaches consequence. Ancient places teach that endurance is measured across generations, not release cycles. Technology worthy of this landscape should be powerful but not wasteful, modern but not rootless, intelligent but never detached from the people and places it serves.
The final position
Watch without paralysis. Prepare without boasting. Experiment without surrendering control. Act without panic. Lead without needing every room to know that you are leading.
That is the American-modern adaptation of the 24-character strategy. It is not secrecy. It is disciplined readiness. It is not timidity. It is strength held in reserve until purpose, evidence, and timing align.
For Mythos and the AI systems that will follow, the formula is straightforward: mission above, mechanism below; human authority above, machine agency below; public trust above, verifiable controls below.
As above, so below. Build both levels with intention.