A commons gives reasoning somewhere to live without requiring everyone to believe the same thing. The distinctions below keep that shared record useful.
Source words and interpreted meaning
A person authors what they say. AI may propose what those words mean in a structured argument. That interpretation remains a proposal until the appropriate person confirms it; another reviewer cannot retroactively decide what the speaker meant.
Acceptance and reliance
A group may accept a statement into its model while still treating it as uncertain. Deciding to rely on it for a particular action is a further commitment, with its own scope and authority.
“Let us try Sunday opening” does not mean “Sunday opening will work.”
Disagreement and membership
A reservation attaches to something specific: a claim, an assumption, a causal relationship, or a proposed action. The group can act while keeping that reservation visible. Belonging does not require unanimous agreement.
Predictions and observations
A prediction records what you expect before seeing the outcome. An observation records what happened, with its source. A retrospective explanation cannot substitute for a prediction made in advance.
Actions and outcomes
Opening the garden is an action. More households visiting is an expected effect. Completing the action does not prove that the effect occurred.
Revision and erasure
When an expectation fails, the record should help locate what needs reconsideration. Changing your mind leaves the earlier reasoning available, so someone else can understand what changed and why.
A shared memory, governed by people
AI can help propose, connect, and retrieve reasoning. People remain responsible for what they confirm and authorize. Evidence can challenge what the group has chosen to rely on.
The Reason Commons Protocol specifies these distinctions in detail. It is a specification; it is not a claim that every part is already implemented.