On the Last Attempt to Centralize the Internet—the ICP-2 Revision Draft, Implementation Power, and Structural Risk

Written by Lu Heng

|

29 December 2025

CEO of LARUS Limited and founder of the LARUS Foundation, he works at the intersection of Internet infrastructure, the IP address market, and global Internet governance, drawing on direct experience with all five Regional Internet Registries. These notes seek to clarify digital resource governance and promote a more responsible and resilient framework for critical IP assets.

Every system reaches a point when it must adapt to reality or try to freeze itself in place.

That point is not determined by debate.

It is determined by scale.

The Internet has reached that point.

The ICP-2 revision draft is presented as a promise of protection, stewardship, and order.

In reality, it is an attempt to preserve a structure that no longer fits the scale or nature of the system it claims to govern.

The fundamental error is simple.

The problem is not that global trust has weakened.

Global trust never existed.

If centralized trust were possible on a planetary scale, we would already have a world government.

We do not.

We never will.

The Internet did not thrive because of trust.

It thrived because trust was unnecessary.

When the Internet was small, abundance concealed fragility.

When IP addresses were plentiful, governance mattered little.

When failures were local, mistakes were manageable.

That era is over.

The Internet is now critical infrastructure.

IP addresses are now productive capital.

Failure is global.

Fragility is no longer acceptable.

At this stage, systems face a choice.

They can remove chokepoints.

Or defend them.

They can permit exit.

Or try to prevent it.

They can decentralize.

Or concentrate authority and claim to guarantee safety.

The ICP-2 revision draft chooses concentration.

It blocks entry.

It restricts exit.

It makes the creation of new institutions practically impossible and existing institutions effectively irreplaceable.

It turns coordination into permission.

And permission into power.

Worse still, it does so without meaningful community consent, entrusting decisive authority to a private corporation and calling it governance.

This is not stabilization.

It is entrenchment.

The structural flaw is decisive.

ICANN and the NRO have no coercive force.

They command no army. They control no borders.

They cannot compel obedience.

Their authority rests entirely on voluntary cooperation.

That is precisely why this approach is fatal.

A system based on voluntary participation collapses as soon as it restricts exit options, blocks competition, and concentrates power, because it destroys the only foundation on which it depends.

Consent is the system.

Those who linger too long in institutional comfort eventually mistake proximity for real power.

They confuse a self-selecting community with the world it claims to represent.

This is not naïveté.

It is structurally unstable.

The moment obedience shifts from voluntary alignment to perceived coercion, legitimacy disappears.

There is no control mechanism to absorb the shock.

No buffer.

No fallback plan.

Unlike historical empires, this system cannot impose order.

It can only persuade.

And when persuasion gives way to entrenchment, cooperation does not endure.

It disappears.

That is why centralization is especially fragile here.

In trying to protect itself, the system destroys the very conditions that made it viable.

History repeats this pattern relentlessly.

When threatened, old systems close themselves off.

They block every exit.

They defend privilege.

They confuse control with stability.

Kings did this.

Authoritarian regimes did this.

Planned economies did this.

All believed they were preserving order.

All accelerated their collapse.

Systems do not collapse because challengers attack them.

They collapse because, in seeking to protect themselves, they obstruct adaptation, defend strategic gateways, and prohibit exit.

The ICP-2 revision draft follows exactly this path.

It presents a structural transition as a moral dilemma: good institutions versus bad actors, order versus chaos.

This view is not merely wrong;

it is fatal.

Complex systems do not collapse because someone is malicious. They collapse because assumptions stop evolving, incentives diverge, and chokepoints are defended instead of removed.

In hindsight, accelerating the old model’s collapse may not be a tragedy.

It may even be necessary.

A system that refuses to evolve and lacks force cannot stabilize itself.

It can only lose consent.

And when consent disappears, the system does not collapse violently.

It is quietly bypassed.

Nations do not trust one another.

They never have.

But nations cannot afford to lose access to the Internet.

This contradiction leaves only one outcome.

Critical Internet resources cannot depend on centralized permission, not because trust has disappeared, but because centralized trust was never possible.

When coordination fails, systems fracture or decentralize.

The Internet has already chosen decentralization at the network layer.

History now demands the same choice at the resource layer.

The ICP-2 revision draft does not prevent this future. It confirms it.

The conflict will subside.

The names will fade into obscurity.

But the direction is immutable.

When a system founded on consent tries to survive by closing itself off,

it ensures that whatever replaces it will replace it completely.