Application Semantics Over Bounded Verification
The original can be found at Zenon Developer Commons .
Status: Draft / Notes Non-normative Builds on: Multi-Verifier Consistency & Refusal
Motivation
Up to this point, verification has been described purely at the protocol level.
Applications must operate on top of partial, bounded, verifier-local truth.
This note defines how applications interpret and act on verified state without assuming global completeness or canonical ordering.
Fundamental Constraint
An application must never assume:
- global state completeness
- universal finality
- synchronized verification across users
Applications operate on locally verified facts only.
Verified Fact Model
A verified fact is:
- a statement proven against the verifier’s frontier
- derived from lineage-consistent commitments
- accepted within bounded resources
Examples:
- an account balance
- ownership of an asset
- inclusion of a state transition
- a resolved cross-chain proof
Facts may expire as frontiers advance.
Fact Scope
Each fact has an explicit scope:
- verifier-local
- frontier-bounded
- time-relative
Applications must treat facts as contextual, not absolute.
Application Read Semantics
When reading state, an application must:
- specify required verification depth
- specify acceptable frontier age
- handle missing or unverifiable facts
Absence of data is not an error.
It is a valid outcome.
Application Write Semantics
When submitting actions, an application must:
- reference the facts it depends on
- tolerate refusal or delay
- avoid assuming immediate global effect
Writes are proposals, not guarantees.
Refusal-Aware Logic
Applications must explicitly handle refusal states:
- retry later
- request alternative proofs
- degrade functionality
- halt execution safely
Refusal is a first-class outcome.
Multi-User Interaction
When multiple users interact:
- shared state is inferred from overlapping verified facts
- disagreement is expected
- reconciliation is explicit, not implicit
Applications cannot rely on implicit consensus.
Eventual Consistency Without Global Truth
Applications may converge over time if:
- frontiers advance
- proofs propagate
- refusals resolve
Convergence is opportunistic, not required.
Trust Surfaces
Applications must expose:
- what is verified
- what is assumed
- what is pending
- what is refused
Opaque trust assumptions are prohibited.
Offline Operation
Applications must support:
- cached facts
- degraded modes
- delayed verification
Offline correctness is local correctness.
Failure Modes
Applications must assume:
- proofs may never arrive
- some facts may never verify
- some actions may never complete
Safety over liveness.
Design Implication
This model favors applications that are:
- state-light
- fact-driven
- tolerant of delay
- explicit about uncertainty
Applications designed for monolithic global state do not translate.
Boundary Statement
Bounded verification does not limit applications.
It forces applications to be honest about what they know.
What Follows
Once applications operate on bounded facts, the system must address how economic and incentive mechanisms behave without global finality assumptions.
The next note introduces incentive-aware design:
0x08 — Incentives Under Partial Verification