Network Public: design-doc series — tracking #796–#801

[monsieurbulb]

Network Public: design-doc series — tracking #796–#801

A series of six design-doc PRs proposing how to evolve JAR towards what the underlying paper calls a Network Public: a substrate that is structurally prevented from being privately enclosed, including by its own future operators.

The series is purely documentation. No consensus changes, no Cargo / Lake changes, no roadmap reprioritisation — every concrete mechanism described here sits as services on top of the existing Refine-Accumulate model, or as RFC-grade thinking that may inform future work. Nothing here asks maintainers to deprioritise the P0 mainnet-readiness work tracked in #172.

The source paper

The Network Public — A Peer-to-Peer Protocol for Sovereign Collective Intelligence (Decent Partners, v4.3, April 2026)
Read the full paper as a gist
Raw: https://gist.githubusercontent.com/monsieurbulb/97b081132b9d467d1cbb3601f0d39a77/raw/network-public.md

PR #796 is a synthesis / repo-resident counterpart to that paper. The other five PRs each isolate one mechanism the paper proposes.

The four functional separations the paper introduces

Separation	Purpose	Lands in PR
Ingestion — record everything without judgment	Substrate of provenance; nothing about it judges	#801 (first concrete non-code contribution type)
Governance — direction by dilutive contribution weight, with the right to refuse / pause	Lets the network slow itself down without adopting stake-based slashing	#797
Inference ownership — revenue distributed by graph traversal	Connects upstream contribution to downstream value via patience	#799
Deliberative allocation — a negation layer for what traversal can't yet recognise	Counterweight against the system's own pattern-entrenchment	#798

Plus a norms layer (#800 inference shapes) that defines minimum service-author obligations for any inference service running on the substrate, so that base-layer neutrality doesn't quietly re-enclose users at the service tier.

Reading order

The PRs are designed to be read in this sequence. Each builds context the next assumes.

1. docs: JAR as a Network Public #796 — JAR as a Network Public (parent thesis; maps the four separations onto existing JAR primitives, names the gaps, completes the genesis.md → coinless.md → network-public.md arc)
2. docs: governance — the right to refuse #797 — Governance: the right to refuse (pause/refusal primitive; directly answers a gap raised in Does the protocol need staking, coretime, or L2 coins? A first-principles audit of inherited assumptions #383)
3. docs: the negation layer (Genesis 2.0 deliberative cohorts) #798 — The negation layer (Genesis 2.0 deliberative cohorts — rotated, fixed-term, pay-independent)
4. docs: the patience tax — delayed-settlement escrow #799 — The patience tax (delayed-settlement escrow funding traversal-weighted attribution and the negation layer; connects Coinless JAR's revenue model to upstream contribution)
5. docs: inference shapes — service-author obligations #800 — Inference shapes (three minimum properties any inference service must satisfy; admissible grounds for refusal under docs: governance — the right to refuse #797)
6. docs(jar-genesis): cross-type bridges — first non-code contribution type #801 — Cross-type bridges (operationalises Genesis 2.0's "7+1 comparison" model with ingestion contributions as the first concrete non-code type — smallest possible step out of code-only)

Relationship to existing threads

• Does the protocol need staking, coretime, or L2 coins? A first-principles audit of inherited assumptions #383 (first-principles audit of staking, coretime, L2 coins). This series accepts the audit's conclusion (none of those are needed) and proposes a constructive next step: governance and revenue mechanisms that don't reintroduce stake-weighted control through the back door. Cross-linked from this comment.
• Coinless JAR thesis (gist). The series treats Coinless as a precondition, not a target. PRs docs: governance — the right to refuse #797 and docs: the patience tax — delayed-settlement escrow #799 fill in mechanisms Coinless leaves open.
• JAR Genesis 2.0 (gist). PRs docs: the negation layer (Genesis 2.0 deliberative cohorts) #798 and docs(jar-genesis): cross-type bridges — first non-code contribution type #801 build directly on Genesis 2.0; they are its first proposed concrete uses beyond code-only contribution.
• Roadmap Grey: mainnet-launch-ready node roadmap #172. Nothing in this series competes with P0 mainnet-readiness work. The deliverables are doc PRs intended to inform later phases.

Status

All six PRs are open as drafts so contributors can comment without time pressure. They will be marked ready for review once the conversation in this issue settles direction.

cc @sorpaas @olanod

[monsieurbulb]

Update — paper v4.4 (Narrative Sovereignty)

The source paper has been updated from v4.3 → v4.4 to address a question raised in discussion: why does inference-as-world-generation matter?

The short answer the new revision gives: because narrative sovereignty is the highest-order property the protocol must preserve, and worlds are the form in which a network can speak about itself in its own voice.

The expanded argument: every prior decentralised network has allowed its narrative to be defined exogenously. Bitcoin's public meaning was shaped by financial-press cycles. Token-governed networks (Polkadot, Cosmos, etc.) have attempted to internalise this by routing treasury funds into "marketing" line items, but the produced media is structurally exogenous — produced by outside contractors against outside briefs, not by the contributor base. The cost is incremental at first and decisive over time: the network's public meaning drifts toward whatever narrative special interests are most willing to fund, while the embedded culture that actually constitutes it goes uncatalogued and undefended.

A protocol that produces collective intelligence and cannot speak about itself in its own voice has surrendered its most important contribution.

What changed in the paper

• §2 — The Problem with Existing AI Models. Added a sixth failure mode (narrative capture) before the closing summary.
• §7.1 — Inference as world-generation. Added a closing paragraph naming worlds as the form in which a network stays sovereign over its own story.
• Header. Bumped to v4.4.

Read the updated paper · Raw

Conflict / contribution analysis against the 6 PRs

PR	Effect
#796 parent thesis	Updated — added a Narrative Sovereignty section before Open Questions, mapping the four separations onto the defence against narrative capture
#797 governance refusal	Updated — "exogenously-defined narrative campaigns" added to What Can Be Paused as admissible grounds for refusal
#798 negation layer	No change required — long-form synthesis and reframing as fundable contribution types is already the mechanism through which a community catalogues and defends its embedded culture
#799 patience tax	Updated — Context now frames the patience tax as a narrative-sovereignty mechanism (closes the exogenous-funding entry point at the substrate level), in addition to its attribution role
#800 inference shapes	No change required — session-length neutrality, first-class exit, and reflective interruption already prevent the engagement-economy attractor that is the most common form of narrative capture in adjacent domains
#801 cross-type bridges	No change required — ingestion (the first concrete non-code contribution type) is the substrate from which the network's self-account is generated; the Network Public claim that this matters is downstream of the bridge mechanism, not in conflict with it

No conflict anywhere in the series. Three PRs tightened, three left intact. The framing is additive — it names the property the existing mechanisms are defending, without changing any mechanism.