Why Begin with a Manifesto
Every system begins with assumptions. Most begin with features. SON begins with rights.
Architecture follows ethics. Before we talk about layers, queries, or contracts, we must ask a prior question:
What does a knowledge system owe to the people it touches?
If that answer is vague, no amount of technical brilliance downstream can save it. If that answer is explicit, then every implementation can be judged — and audited — against it.
The Problem We Face
Today’s AI systems are technical MVPs: clever approximations of “learning” that scale quickly but lack rigor.
They are built for fluency, not accountability. They improvise like a jazz soloist but rarely show their sheet music. They forget by default, obscure provenance, and smooth contradictions into a story that sounds right — but cannot be trusted.
These stacks optimize for speed and convenience. What they do not optimize for is responsibility to the user.
The First Principle: Sovereignty of Intent
Every subsequent commitment stems from one foundational truth:
A person’s intent must remain sovereign over how their data is used.
Agency is not an interface preference. It is a structural right.
The Four Pillars of Agency
1. Autonomy & Consent
- Identity is deliberate. You choose when to disclose, adopt a role, or remain silent.
- Consent is explicit. No hidden defaults, no implied permission.
- Refusal is valid. Saying “no” is not system failure — it is human choice.
2. Transparency & Accountability
- Explanations grounded in artifacts. Systems must show their work, not just their words.
- Replayability “as-of.” Every answer reproducible under the policies and evidence of its moment.
- Economic clarity. No hidden costs, no surprise degradations.
3. Control, Revision & Portability
- Revision by new commitment. Changes add new records — they don’t rewrite history.
- Revocation without amnesia. Withdraw future consent without erasing the past.
- Minimal disclosure. Only what is necessary travels; terms travel with the data.
4. Contestability & Resilience
- Opposability. Systems err; people must be able to challenge, reject, or qualify results.
- Error as fuel. Contradictions are visible and bounded until resolved. Resilience is built on challenge, not denial.
Why Shared Object Networking
SON (Shared Object Networking) is not another model, database, or platform. It is a protocol surface that makes these rights operational.
- Objects carry identity, provenance, and obligations.
- Layers keep contexts distinct, so disagreement stays visible.
- Every durable act — admission, orchestration, revision — leaves behind a signed, replayable record.
- Contestation is formalized. Revision is additive. Consent is portable.
Where other systems ask for trust, SON makes trust enforceable.
Where others erase contradictions, SON insists they remain visible until resolved.
Where others treat the human as input, SON treats the human as a contractual partner.
From Manifesto to Protocol
The manifesto sets the rights. SON encodes them as invariants.
What results is not an application, but an infrastructure of agency:
a substrate where provenance, consent, and contestability travel with the data itself.
The ambition is plain: to replace “trust the system” with systems that can be held to account.
Read the full paper.
Older Foundational Papers on Shared Object Networking
Shared Object Networking
Shared Object Networking Distributed GraphRAG