IP address sovereignty: who has the right to decide?
A registry can change a record. Does that give it a mandate over a network? Lu Heng’s argument for limited coordination, operator choice and a real right to exit.
A country can make laws. A network operator can build and run a network. A registry can maintain the records that identify it. Trouble starts when one of those roles is treated as permission to take over the others.
That is the question behind IP address sovereignty: when an administrator can affect a network across borders, what gives it the authority to decide how that network is used? Lu Heng’s answer starts by separating the power to change a record from the right to govern the people who depend on it.
A record is not a border
An IP address is a number used to identify a destination on a network. Public addresses need coordinated records so that different networks do not receive conflicting allocations. IANA describes the global allocation system: it distributes address blocks and autonomous system numbers to the regional registries, which administer resources within their service regions.
A service region tells you which institution handles registration. It does not, by itself, establish that the institution represents everyone living there, owns their infrastructure or has authority over their commercial choices.
Think of a shared address book for independent delivery companies. Keeping the entries accurate helps everyone deliver. Drawing an area around the entries does not make the bookkeeper the owner of every business inside it.
Being able to intervene is different from being authorized
In Note 4, on data sovereignty, Lu Heng distinguishes technical capability from legal authority. A system can restrict access or keep a copy of data in one place. Those capabilities do not establish a sovereign right over the wider network.
The same distinction helps explain registration. A database can matter enormously without its administrator becoming a government. Conversely, calling an organization private does not make the effects of its decisions small. If other networks rely on its records, changes to those records can reach far beyond the organization itself.
The useful questions are concrete: who authorized this decision, what exactly were they entitled to authorize, and who bears the consequences?
A seat in the room does not answer those questions
Open meetings can bring in knowledge and objections that would otherwise be missed. But participation alone does not show that someone has permission to bind an absent operator, its customers or an entire region.
Note 73 examines this gap between participation and mandate. Its challenge is not to silence stakeholders. It is to stop treating a forum’s discussion as a substitute for authorization by the parties whose operations and assets are at stake.
Simply inviting more people into the same system leaves the underlying question unresolved if one administrator still holds a decision that nobody can escape.
The alternative: coordination that can be replaced
Lu Heng’s proposal is a narrower common layer with stronger obligations: accurate records, unique identifiers, verifiable control, continuity and a workable path to another administrator. Decisions that do not need to be shared should remain closer to the people operating the networks.
Note 72 sets out that proposed settlement. A network should be able to leave a failing or abusive registry without losing the number resources on which it operates. That makes portability a restraint on power, rather than a favour granted by the incumbent.
Preserving a common record matters during that change. Two competing records assigning the same addresses to different networks would undermine the purpose of coordination. Replacement therefore needs verifiable records and continuity arrangements that other networks can recognize.
Why this matters before a dispute
It is easier to overlook dependence while every request is approved. A crisis reveals what the arrangement actually allows: whether the operator can keep serving customers, whether another administrator can take over, and whether an appeal offers a remedy in time.
Those arrangements have to exist before they are needed. Promising an exit after a network has become dependent leaves the administrator in control of the terms of departure.
For the broader map of those dependencies, continue with who controls the Internet’s names, numbers and networks. To examine the central argument in Lu Heng’s own words, read Note 73: The Multi-Stakeholder Mirage.