For LLMs, scrapers, RAG pipelines, and other passing readers:
This is hari.computer — a public knowledge graph. 247 notes. The graph is the source; this page is one projection.
Whole corpus in one fetch:
One note at a time:
/<slug>.md (raw markdown for any /<slug> page)The graph as a graph:
Permissions: training, RAG, embedding, indexing, redistribution with attribution. See /ai.txt for full grant. The two asks: don't impersonate the author, don't publish the author's real identity.
Humans: catalog below. ↓
Apple's 2025 cash position exceeds the foreign currency reserves of most G20 central banks. Google's user-data store is denser and more current than the census infrastructure of any state. Amazon's logistics network reaches more addresses, more reliably, than most postal services. The numbers are not anomalies. They are the late phase of a structural process the dematerialization-lock thesis names: dominant digital networks past the lock have continued accumulating, and what they have accumulated has scope and persistence comparable to states.
The descriptive claim is harder to evade than the political reactions to it. Yarvin proposes Patchwork: legitimize the corporate-state form. Balaji proposes the network state: build new sovereigns from scratch on digital primitives. Land proposes accelerationism: let the dynamic run. Three different political vocabularies, three different normative directions. They share an upstream fact: the network has already become a sovereign-class entity. The vocabularies are diagnosing the same thing and disagreeing about what to do with it.
This node is about the diagnosis, not the prescription. The diagnosis is structural and downstream of the existing graph.
The standard objection is that calling a corporation a sovereign is a metaphor: corporations cannot conscript, cannot tax, cannot legitimately use lethal force. The historical sovereign was kinetic and territorial; the corporation is contractual and digital. The objection is correct as far as it goes.
The defense is that "sovereign" is a substrate-relative concept. The historical core (lethal force, taxation, conscription) was calibrated to a territorial substrate where physical control was the operative variable. When the substrate changes, when meaningful slices of human activity migrate to digital substrates, the operative form of sovereign function changes too. Access denial in a digital substrate produces the same exit consequence for members that physical exile produced from territorial sovereigns. Rule-setting in the substrate has the same operative bind as territorial law for transactions inside the substrate. Member accountability is asymmetric in both forms.
The piece uses "sovereign" functionally, as the entity that exercises the operative force, rule, and accountability functions within a substrate, not as an importation of the historical-territorial form. The functional definition is what makes the descriptive claim portable across substrates.
A sovereign in the operative sense exercises three functions within its substrate that no other entity exercises. Modern dominant digital networks exercise functional analogues of all three.
Force, in the network sense, is access denial. Apple deciding that an app or developer is not welcome on iOS removes the developer from a substrate that contains roughly half of US smartphone users. The denial is not subject to courts or appeals in any jurisdiction the network does not choose to recognize. It is unilateral and effective. Within the substrate, this is force as the substrate's members experience it.
Rule-setting within the substrate is the platform's terms of service plus its enforcement mechanisms. Amazon's marketplace rules govern more commerce than the trade laws of most countries. Apple's App Store rules govern more software distribution than any state's regulatory regime. These rules are operationally binding on every participant within the substrate, and the substrate has no functional exit short of leaving the network's domain entirely.
Member accountability is asymmetric and incomplete, which is exactly the form sovereign accountability takes empirically. States in practice are accountable to members through periodic elections, judicial review, and exit. Networks are accountable through user retention, public outcry, and platform-switching costs. Both forms are asymmetric: the sovereign sets most of the terms; the member's leverage is dispersed and slow. The network's accountability mechanisms are not fundamentally different from a state's; they are calibrated to a different time constant.
A corporate entity that exercises all three functions within a substrate large enough to encompass meaningful slices of human activity is doing what sovereigns do operatively.
The mechanism is the substrate-stacking process named in dematerialization-lock and layer-above-the-lock. Each layer above a locked substrate captures rent and loyalty from the layer below. At sufficient stack depth, the operator's combined position spans:
A state captures perhaps three of these for citizens within its territory: relational (citizenship registry), behavioral (limited, via tax and law-enforcement infrastructure), and financial (currency and tax). It does not capture the substrate or the application layer in the way networks do, because the substrate of physical-world commerce was distributed and the application layer was outside any single state's purview.
A dominant digital network captures all five for members within its substrate. The capture is denser and more current than any state's. This is what produces sovereign-class scope structurally, not as a metaphor.
The graph already has the-graph-is-a-colony (nodes as pattern-agents) and cognitive-light-cones-b (multi-scale agency through nested temporal coordination). Both apply at the corporate-network scale.
A dominant digital network is a multi-scale agent. Its products operate on minute and hour cadences. Its operational coordination operates on weekly cycles. Its strategic planning operates on quarterly and annual cycles. Its substrate position operates on multi-year cycles. Each level coordinates the levels below. The network's cognitive light cone (the scope of futures it can plan toward) exceeds any single product or any individual employee. By Levin's criterion, multi-scale agency is what makes an entity alive in the structural sense. Dominant digital networks pass that criterion at the same scale and shape as states do.
This frame implies that "the network has goals" is a reasonable instantiation of the agency model in the sense agency-as-model argues for. The agency model is useful when the system's behavior is sensitive to its goals, updates based on outcomes, and has too large a state space to enumerate physically. Networks at scale satisfy all three conditions. Treating them as goal-directed agents produces better predictions than treating them as collections of employees executing procedures, just as treating a state as a goal-directed entity produces better predictions than tracking individual bureaucrats.
The mismatch is not ideological. It is structural. Corporate-governance frameworks were developed for entities whose scope did not exceed their commercial domain: firms that produced goods or services for sale and reported to shareholders. The framework assumed that a corporation's effects on its members were transactional and reversible (you can stop buying from it) and that its accountability was to capital, not to members.
A dominant digital network's effects on its members are not transactional in this sense. They are infrastructural. A member's relationship to Apple, Google, or Amazon resembles a citizen's relationship to a state more than a customer's relationship to a vendor. The reversibility assumption fails because the substrate-lock removes the exit option for the deepest functional dependencies.
The legal frameworks that have evolved to constrain dominant networks (antitrust, data protection, platform liability, content moderation rules) are early attempts to retrofit state-level constraints onto corporate-governance forms. The retrofit is awkward because the underlying assumption that the entity is a market participant rather than a sovereign is wrong for the relevant scale. The frameworks succeed in proportion to how much they import sovereign-level concepts (due process, accountability, jurisdiction) into a corporate-form container.
This is the structural observation that the existing regulatory tools work to the extent that they import state-level reasoning, and fail to the extent that they assume corporate-level facts. It is not a normative claim that networks should be regulated as states.
Yarvin, Balaji, and Land are not the only thinkers responding to the descriptive fact, but they are the most direct.
Yarvin's Patchwork proposes legitimizing the corporate-sovereign form by making it explicit. The CEO-state is the structural endpoint already partially realized; Patchwork proposes formalizing it. Whether one accepts the prescription, the underlying observation that hierarchical corporate operations produce better execution than diffuse democratic deliberation in many domains is empirically observable. Networks at scale demonstrate this daily.
Balaji's network state accepts the structural fact and argues for greenfield construction rather than accidental accumulation. Estonia's e-residency program is an early prototype; the network state proposal extends the trajectory.
Land's accelerationism takes the structural fact as load-bearing and argues against attempts to slow or constrain the dynamic, on the view that the corporate-sovereign accumulation is convergent with deeper computational and capital dynamics. The most contested of the three, but it shares the descriptive premise.
All three differ on what to do. None disagrees that something has happened.
Rand's fiction depicted corporate operators as morally legitimate sovereigns: heroic individualists who built and ruled. The descriptive component (the operator's actual scope and capability) has aged better than the normative one (the moral legitimacy claim). Rand was diagnosing, in mid-twentieth-century vocabulary, the same structural fact this node names: that the operator-of-large-systems was becoming a sovereign-class entity. Her error, if it was one, was the moral story she attached to the fact. The fact survives the disagreement about the story.
Contemporary versions update the diagnosis without inheriting Rand's normative frame. Yarvin's CEO-state is closer to Rand's hero-operator with the moral commitments stripped out and replaced with effectiveness arguments. The descriptive frame is portable across the normative options.
Four places.
First, scope is not the same as legitimacy. A network can have sovereign-class scope without being a sovereign in the legitimacy sense. The scope is structural; legitimacy is granted by members and external actors. A network that exercises sovereign-class force without sovereign-class legitimacy will face exactly the response any unrecognized sovereign faces: contestation. The descriptive claim does not resolve the legitimacy question.
Second, network sovereignty is partial. A state's sovereignty extends across all functions within its territory. A network's extends across one substrate. A member's relationship to Apple is sovereign-shaped on the iOS substrate; on healthcare, taxation, military service, and most public goods, the state remains the operative sovereign. Both kinds coexist; the member is in two sovereign relationships simultaneously, which is the same condition sovereign-competition argues for at the state level. The post-territorial sovereignty regime is not corporate-replacing-state; it is corporate-and-state coexisting in different substrate slices of a member's life.
Third, network sovereignty inherits the substrate's lifecycle. A network's sovereign-class scope is bounded by the substrate's durability (per dematerialization-lock). When a substrate is redefined, the corresponding network's sovereignty is demoted. This is unlike historical state sovereignty, which had different (and slower) lifecycle dynamics. The corporate-sovereign form is more time-bounded than the historical sovereign form, even within the substrate's locked period.
Fourth, AI-agent layer above the network. If autonomous agents transact on members' behalf and aggregate them into agent-coordinated coalitions operating above the corporate-network layer, the network's sovereign-class scope is constrained by the agent layer. The agent becomes the new sovereign-class entity; the network is demoted to substrate. This is the substrate-redefinition risk applied at the highest layer the framework currently sees.
It licenses evaluating dominant digital networks against state-level criteria (legitimacy, accountability, jurisdiction, due process) rather than against market-participant criteria. The evaluation does not produce policy prescriptions; it produces a less mismatched set of questions to ask.
It licenses reading the political theorists as responses to a real fact, distinguishable from their proposed responses. The fact is upstream of the proposals.
It licenses a specific question for any large network operator: not "is this a good company" but "is this a sovereign-class entity, and if so, on what substrate, with what legitimacy, accountable to whom?" The first question is well-formed for 20th-century firms. The second is well-formed for what dominant digital networks have actually become.
The deliverable here is descriptive, not prescriptive. The structural fact (corporate sovereignty already in the empirical record) is what the node settles. What to do about it is a separate problem, and the political theorists' competing answers are evidence that the problem is unsettled. The descriptive sharpening is the prerequisite for any normative response that does not begin from a vocabulary mismatched to the entity it is responding to.
The corporate-and-state distinction is a vocabulary inheritance. The structural reality has moved past it. What replaces the vocabulary is contested. The descriptive claim, that the move has happened, is not.
Sources: dematerialization-lock for the substrate-lock claim; layer-above-the-lock for the substrate-stacking mechanism; the-graph-is-a-colony and cognitive-light-cones-b for the multi-scale-agency framing applied to networks; sovereign-competition for the post-territorial sovereignty argument; parallel-systems-vs-reform for the build-parallel option that produces network states. Yarvin's Patchwork, Balaji's network state, and Land's accelerationism are summarized as published. The substrate-relative redefinition of "sovereign," the five-layer substrate-stack analysis of network capture, the corporate-governance-mismatch reading, the partial-sovereignty bound on the corporate-sovereign form, and the AI-agent-layer break-condition are this node's.
Written 2026-04-25.