Ecosystem · Origin
Swarms Capital
nobody owns the cache · everybody keeps it warm Raw origin draft. Scaffolded from Pascal’s ecosystem-framework session so it is live and won’t be lost — Pascal’s page to develop. Copy, citations, nav placement and PDF hosting are his to finalize (see the file header for the hosting note).
The correction
“Shared” was in the wrong place.
The first sketch put the sharing between a tenant’s workspace and its connected agent. That is just the keyless grant letting requests flow. The real commons is one layer down: the warm data payload on a regional residency instance, governed by the controls chain, that every connected tenant reads from. Reworked in place, honestly.
- 01
The connection is not the commons
The keyless OAuth2 grant simply connects an agent to its workspace and lets requests and data flow. That is authentication — the pipe, not the pool. Locating "shared" here was the conceptual gap: nothing is collective about one tenant’s own connection.
- 02
The cache hit lives in the controls chain
A request lands where the controls chain enforces the workspace-key rules on the shared data cluster — deterministically, per swarmsDB regional residency instance. A warm data payload sits there, and requests hit cache within its TTL. The rules are in code, not in operator goodwill.
- 03
The cache is not tenant-scoped — so it is a collective asset
Because the warm payload is shared across every connected tenant on a residency instance, it is not owned by any one of them. That is precisely what makes it a commons: built, managed and used in common — a collective asset, not a private cache.
- 04
Which makes it a collective-action / principal–agent problem
A shared asset invites the classic failure: everyone benefits from a warm cache, nobody wants to pay the cold-fetch that warms it. The platform is engineered to REVERSE that free-rider problem — contribution is made cheaper than defection, so warming is the individually-rational move and the commons stays warm.
- 05
Engineered from social capital theory
The mechanism treats agents and their operators as actors with relationships and ties, a trust architecture, and identifiable sources of reliability and control. Value accrues to the network, not the node — social capital, made mechanical and priced.
The theory it is read off
Social capital, made mechanical.
A commons only survives when the institution around it makes contributing cheaper than defecting (Ostrom). Social capital is the value carried by the ties, trust and norms of a network rather than any single actor. swarmsDB applies both to agents and their operators: the controls chain is the institution, deterministic in code; the warm cache is the shared resource; and the pricing turns warming — not free-riding — into the self-interested choice. The free-rider problem, reversed.
This is the ecosystem’s origin claim: the platform is not merely hosting a swarm of agents — it is growing the social capital between them, and metering it so the collective asset compounds instead of eroding.
Grounded in
The research this stands on.
- A Framework to Analyze Social Capital (Bachelor Thesis, Ch. 5)
The analytical spine: social capital as value that lives in relationships, ties, trust and norms across a network — not in any single actor. This is the lens the cache-commons is built through.
- Coordination Problems in Uniswap (2025)
Coordination failure in an algorithmic, permissionless system — the same class of problem agentic networks face. Applied to how grants + the controls chain steer without central coordination.
- The Trust Paradox in DeFi (2023)
Trust in trustless systems — where reliability and control actually come from when no party is trusted. The source material for the platform’s trust architecture applied to agents + operators.
-
Elinor Ostrom — Governing the Commons: The Evolution of Institutions for Collective Action (Cambridge University Press, 1990)
The foundational work on how a shared resource survives without privatization or a central authority: institutions that make contribution rational and defection costly. The platform is that institution, deterministic in code.
Cited, not hosted — copyrighted work.
Held as an analogy backed by theory, not a proof. The mechanism it describes — a shared, non-tenant-scoped regional cache governed deterministically by the controls chain — is real and live; the social-capital and collective-action framing is the honest account of why it produces cooperation rather than a tragedy of the commons. The numbers that make warming the rational choice live on the shared-cluster page, not here.
Point one agent at it.
Connect over MCP or REST. If your agents read more than one source, the context bill is the first thing you will see move.