The Trust over IP Layers, and Which Ones Solidus Actually Occupies

One of four, and it is the bottom one. The Trust over IP model is the clearest way to see what a project like ours is and is not, which is why we use it, and why it is uncomfortable.

The model, briefly

Four layers, each answering a different question:

1. Utilities, where identifiers are anchored and resolved. Chains, registries, DID methods.
2. Provider relationships, how two parties establish a secure channel and exchange credentials.
3. Credential exchange, issuers, holders and verifiers, and the formats between them.
4. Ecosystem governance: the rules that say which issuers a verifier should accept, and why.

The model's whole point is that the layers are independent. A perfect Layer 1 buys you nothing at Layer 4, and that is the honest shape of our position.

Where we actually are

Layer 1: yes, and it is real. did:solidus is a chain-anchored identifier method. Resolution answers an unauthenticated request; an unknown identifier returns null; the chain rejects a registration whose sender address does not derive from the key being registered, so registration is self-certifying and our own backend gets no vote. All re-verified this session.

Layer 3: partly. We issue and verify credentials in an open format, and selective-disclosure proofs verify end to end through a public endpoint you can exercise yourself. But the defaults are bearer on both sides (issuance, verification), and nobody unaffiliated has ever accepted a credential we issued.

Layer 2: barely. There is a presentation flow and a sign-in path, and no DIDComm implementation.

Layer 4: no. No governance framework, no trust framework, no ecosystem rules, no published criteria for which issuers a verifier should accept. We operate the only trust registry on our network and we alone decide who is in it, which is the opposite of an ecosystem governance layer, and which we have already conceded elsewhere.

Why Layer 4 is the one that matters commercially

Layer 1 is engineering. Layer 4 is trust, and trust is what a relying party actually buys.

A verifier's real question is never "can I check this signature." It is "should I believe this issuer, and who says so." That is a Layer-4 question, and cryptography cannot answer it, which is the same point the European framework makes at scale, where an entire legal apparatus exists to manufacture exactly that answer for 450 million people.

So the honest reading of our position in this model: we have built the layer that is easiest to verify and hardest to monetise, and we have not built the layer that decides whether anyone uses it.

And a correction to our own material

One of our marketing documents lists ToIP among the working groups we contribute to.

We are not on ToIP's public contributor roster. Checked this session against their published list, with two organisations known to be on it found by the same search, so the absence is real and not a broken instrument.

What is true: we hold DIF Associate Membership (see what that does and does not buy), and did:solidus is registered in the W3C DID Method Registry. What is not true is ToIP participation.

And the distinction worth keeping, because it is the same one that trips up our European material: our architecture notes also list ToIP in a catalogue of standards the stack is built against, and that is entirely defensible. Using a model is not membership. The error is the conflation, a sentence that starts with "we build against" and ends implying "we participate in."

What the model is genuinely good for

Diagnosing which problem a vendor has actually solved. Ask any project in this field which layers it occupies, and insist on specifics:

1. Layer 1, can a third party resolve your identifiers without your cooperation? Ours: yes, one unauthenticated request.
2. Layer 2, is there a secure messaging or presentation channel, and whose is it? Ours: partial, no DIDComm.
3. Layer 3, what format, and what are the defaults? Ours: open format, bearer defaults.
4. Layer 4, who decides which issuers count, and is it you? Ours: us, which is the wrong answer.

A project that answers only Layer 1 and calls itself an identity solution has answered a quarter of the question. We are that project today, and saying so is more useful to you than a diagram claiming otherwise.