Agent Jungle Law 2: From "Technical Control" to "Order Building"

Over the past few days, the Ethereum ecosystem has been engaged in a profound debate over AI-accelerated mathematical research and its potential impact on traditional cryptographic assumptions.

Ethereum researcher Justin Drake has called for planning asset migration in advance and building quantum-resistant, stronger cryptographic protection, while Vitalik Buterin, even as he urged the community to take the risk seriously, cautioned against rushing a migration. What this debate reflects is not the established fact that existing cryptography has already been broken by AI, but rather frontier researchers’ vigilance about future technological risk — and their disagreement over the right response.

If AI can significantly accelerate mathematical research and cryptanalysis, and even push some long-standing security assumptions back onto the table for re-examination, then security windows once considered comfortably long may face new uncertainty. This does not mean existing cryptography has already failed; it reminds us that every concrete security scheme must face the continuing challenges brought by technological evolution.

The deeper question this debate reveals is far weightier than any dispute over a single technical route: when the very technological foundation humans use to control the digital world itself requires continuous upgrades, migration, and patching, can we rest the ultimate security of digital society on the assumption that some technology will “never fail”?

Blockchain has long relied on elliptic-curve digital signatures (ECDSA), private keys, and state consensus to establish control over assets. The blind spot of the traditional view of security is this: relying purely on “static technical control” and “protection by hard math problems” cannot deliver once-and-for-all absolute security for humanity over long time horizons.

This does not mean blockchain or cryptography has lost its value. It poses a more piercing civilizational question: if technological security is destined to be a dynamically evolving contest between attack and defense, what paradigm shift do humans need in order to keep the capacity to survive, collaborate, and self-govern within ever more powerful agent networks?

1. The Limits of Technical Control: Can We Really “Control” Intelligence?

Traditional digital-security thinking is built on “control”: through keys, permission isolation, code audits, sandboxed environments, and runtime monitoring, it strictly limits what a system is allowed to do.

These technical defenses remain indispensable, and they must keep upgrading alongside advances in post-quantum cryptography, more robust key management, and agent permission control. But facing AI agents capable of autonomous planning, tool invocation, cross-platform execution, and multi-agent coordination, the security problem has completely outgrown the boundary of any single system or model.

flowchart TD
    subgraph S1 [Emergent risk of complex systems]
        A[A single system behaves compliantly] --> B[Many agents combine across platforms] --> C[Unforeseen risks emerge]
    end
    subgraph S2 [Intent–verification misalignment]
        D[Cryptographic verification passes] -. does not imply .-> E[Behavior matches real human intent and social legitimacy]
    end
    S1 --> S2

1. Compositional risk and system-level emergence: even if every individual agent strictly obeys its internal code and local permissions, when tens of thousands of agents interact and contend with one another across platforms and protocols at millisecond speed, the system as a whole will exhibit systemic risks that no single designer can foresee.

2. The misalignment between technical verification and real intent: cryptography can prove that a certain private key signed an instruction, and a blockchain can faithfully execute that code, but neither can judge on its own: what is the real intent behind this signature? Does it stem from fraud, covert manipulation, or an unauthorized act taken under a complex prompt-injection attack?

We can keep refining the precision of control, yet we can never derive “absolute security” from it. Technical control draws the technical boundary of what agents can do; it cannot answer the question of order — what they should do.

2. The Lessons of Civilization: Human Society Has Never Run on Eliminating Risk

Looking back across the history of civilization, human society has never functioned by eliminating every potential danger.

Fire never lost its destructive power by being tamed, yet humans invented fire-resistant buildings, firefighting equipment, and legal liability. Cars did not stop having accidents just because traffic rules existed, yet through driver’s licenses, road design, insurance payouts, and accident accountability, humans constrained traffic risk within bearable bounds. The financial system, even more so, has never assumed that everyone is honest; it organizes trust-based collaboration at scale through identity penetration, audit trails, contractual constraints, and penalties for default.

Progress in civilization means not only acquiring more powerful tools, but also building the rules, accountability, and institutions of collaboration that match them.

AI agents, however, bring an unprecedented challenge: they can replicate without limit at extremely low marginal cost, run around the clock 7×24, act across physical and digital boundaries, and may autonomously join in complex coordination with other agents.

Simply copying the old rules of traditional society is bound to fail; we need entirely new institutional design — security is not merely the elimination of danger, but the establishment of an order under which society keeps running even when danger cannot be fully eliminated.

3. From Controlling a Single Agent to Building Common Rules for an Intelligent Society

To accomplish the intellectual leap from “technical control” to “order building,” the pivotal turn is this: shift the question from “how do we make one agent safe enough” to “how do we let different humans and agents collaborate under common rules.”

That turn requires building the foundational architecture of digital society along four dimensions:

flowchart TD
    R[Common rules of an intelligent society] --> I[Identity<br>Who initiates the action]
    R --> A[Authorization<br>Boundaries and quotas of authority]
    R --> L[Liability<br>How overreach and losses are accounted for]
    R --> V["Value & Coordination<br>How contributions are verified and distributed"]

1. Identity: Who is initiating the action? On whose intent does the agent act? How do we establish a tamper-proof anchoring between a human principal and its digital proxies?

2. Authorization: Who permitted it to do what? What are the validity period, quota, and boundaries of the authorization? Can an agent further delegate that authorization?

3. Liability: When an agent’s behavior causes breach of contract, overreach, or economic loss, how do we identify the parties involved, apportion responsibility, and enforce compensation based on public evidence?

4. Value & Coordination: How is the value that agents help create through collaboration objectively verified and distributed? When disputes arise, by what procedure are they adjudicated?

What truly needs to be built is not merely a safer individual agent, but a social mechanism that lets large numbers of not-fully-trustworthy agents run together.

4. Reunderstanding Blockchain: Not an Omnipotent Firewall, but Infrastructure for Order

In this governance turn, we need to clarify what blockchain and cryptography really are. Potential risks to cryptography do not mean blockchain as a whole has lost its value; emphasizing order building does not mean we may abandon technical defenses either.

The Ethereum Foundation and the broader Web3 ecosystem continue to advance quantum-resistant signature schemes, post-quantum cryptography roadmaps, and more robust wallet-protection architectures. That in itself shows: technical defense is foundational engineering that must be patched, upgraded, and migrated continuously.

Blockchain’s strategic value has never been that of an omnipotent firewall that eliminates all risk on humanity’s behalf; it is the infrastructure of order for digital society:

• Public rule enforcement: smart contracts can turn collaboration rules that have passed social consensus into transparent, deterministic programs that no single party can unilaterally tamper with.

• Verifiable behavioral attestation: credential infrastructure such as EAS (Ethereum Attestation Service) produces verifiable records of party identity, authorization relationships, task fulfillment, and collaboration outcomes. For behavior that happens off-chain, reliable data sources, signed proofs, or independent verification mechanisms are still needed to judge whether those records are truthful.

• A cross-organization coordination medium: a common state layer that does not depend on any single tech giant, allowing agents from different platforms and different models to collaborate on unified protocols.

It cannot independently prove the truth of off-chain facts, nor guarantee that a given behavior is absolutely legitimate in a social sense; but it provides the most solid digital foundation for making social norms enforceable, verifiable, and accountable.

5. Who Writes the Rules? Where Does the Order of Digital Civilization Come From?

If we pin the future on “order and norms,” we must immediately answer a fundamental question about power: who writes these rules? Who judges violations? Who has the right to change the rules?

• If the rules are written entirely by a single centralized AI giant, humanity will be reduced to “digital tenants” of the platform;

• If the rules rest entirely on hard-coded “code is law,” the system loses the resilience to cope with complex reality and sudden risks;

• If the rules rest entirely on permissionless community voting, governance easily degenerates into Sybil attacks and the tyranny of the majority.

The legitimacy of norms in digital society must rest on procedural justice and dynamic evolution:

• Joint rule-making and separation of powers: rule-making, execution, evidence verification, and dispute adjudication must not be concentrated in a single party.

• Procedural justice and channels of appeal: determinations of violations must be based on on-chain records and public chains of evidence, with review and appeal mechanisms preserved for humans.

• Capacity for continuous evolution: rules should be upgraded and iterated in an orderly fashion in response to breakthroughs in AI capability, changes in cryptographic security assumptions, and new risks emerging from real collaboration.

Digital civilization needs to constrain not only agents, but also the human organizations that write and enforce the rules.

Digital society therefore needs to explore a governance structure that combines open participation, constraints on power, procedural justice, and verifiable execution. The DAO (decentralized autonomous organization) can be one important form of organization and coordination, but it too must submit to transparency, checks and balances, attribution of responsibility, and continuous error correction.

6. Why Humans Must Remain the Primary Subject of the Digital World

If order is built merely so that AI can execute tasks more efficiently, it is at best a management manual for machine efficiency.

The ultimate question of digital society is: in a flourishing network of intelligence, who decides the goals of collaboration? Who expresses real intent? Who holds the power to authorize? Who bears the final responsibility?

In the worldview of m&W, all of these answers are re-anchored in the human being:

flowchart TD
    B[The human Builder<br>Source of intent / value judgment / accountable party] --> T[Digital Twin<br>Human digital identity ＋ the agent legion it manages<br>＝ digital capacity to act]
    T --> N["The m&W order network<br>Identity → Attestation → Coordination → EcoFi"]

• Builder: humans are a primary source of intent, value judgment, and creative activity, and the party in digital collaboration that must be recognized and respected. Their responsibility should be defined according to the specific authorization, actual control, degree of participation, and applicable rules — not by blanket attribution of every agent action to a human.

• Digital Twin: not a mere identity label, nor a machine account independent of the person, but the digital capacity to act formed jointly by a person’s digital identity and the legion of agents it manages.

AI can vastly expand the scope of human action, but the goal-setting, value judgments, and governance rules of digital society must preserve real room for human participation and oversight.

AI can raise the efficiency of machines, but the goals, values, and ultimate governance of digital society must never slip out of human hands unnoticed.

Conclusion: Bringing Intelligence into Civilization

Returning to the thesis of Agent Jungle Law: in a jungle without common rules, an advantage in capability converts directly into an advantage in action, and an advantage in action converts further into the seizure of resources and power.

But an intelligent society does not have to remain forever in the disorderly contention of a dark forest.

The Ethereum ecosystem’s debate over cryptographic security has sounded an alarm for the entire industry — and pointed the way: do not try to seal dynamically evolving intelligence behind static technical defenses; discipline powerful technology with a dynamic social order.

This is precisely the mission of m&W: to establish order for the network of intelligence.

Progress in civilization has never been merely about creating ever more powerful tools or intelligence; it is about establishing a great order — one in which even the mightiest intelligence can only collaborate within an environment of explicit rules, authorization, accountability, and value distribution, and in which ever-expanding machine capability always serves the continuation and evolution of human civilization.

Let intelligence enter civilization, not let civilization submit to intelligence. That is the deepest declaration of humanity — the primary subject of the digital world — in the age of AI.