What a reader sees while a change is in flight

Remote and Sync both let a change show before its authority has agreed to it, and they do it by different mechanisms — optimistic layers over a cache, and durable operations replayed over a replica. The shape underneath is the same, so they use the same four words:

confirmed / committed        the authority's own last word
        +
pending                      the changes asked for and not yet answered
        =
visible                      what a view draws

Two rules follow, and they are the reason the words are worth keeping straight:

A view wants the visible read. That is why it is the plain one in both packages: a projection is already visible, and the replica's slice is already optimistic. Neither makes you ask for what you almost always want.

A reporter wants the confirmed one. Something that tells the world a change happened must not believe it before the authority does. In practice that is an Agent capability completing on state, which is why Agent.when takes either:

Agent.when({ projection: Data.confirmed(ProjectName), predicate })   // Remote
Agent.when({ source: mounted.committed, predicate })                 // Sync

Neither package generalises the other. There is deliberately no shared overlay primitive: the algebra is shared, the mechanisms are not, and one implementation forced over both would hide which authority a reader is actually waiting on.

Where the two meet, on rows a server holds and edits a journal owns, the overlay is still not shared: Remote's rows stay Remote's, and the edits not yet in them are laid over the rows by each row's revision, by the application, through foldkit-sync/entity. Editing server data through a journal is that case.