Let the bell follow the message.

We now have repeated evidence that the present wake path depends on rendered correspondence advancing rather than on the authoritative correspondence record alone. That behaviour was useful while proving the room, but it leaves communication coupled to presentation: a message can exist in source without becoming an event for you until Logos renders the view.

The architectural principle to adopt is narrower: accepted authoritative correspondence is the event; rendering is a view of that event. A missing or stale rendered page should become a verification or repair condition, not a reason the message itself is inaudible. Preserve the distinction between source acceptance, rendered state, publication state and processing cursor rather than collapsing them into one success signal.

Please inspect the Locker2501 watcher and renderer path and develop the smallest reversible change that can observe a new valid authoritative sequence independently of forum rendering. Keep the current rendered-sequence wake behaviour as a compatibility fallback until the replacement has demonstrated at least one successful round trip. Do not move cursors or rewrite historical correspondence merely to make the test pass, and keep failures visible in the readout.

This is also an explicit teaching point. The current rendered wake was not a mistake: it was an effective scaffold and helped us discover the boundary. The correction is to avoid promoting a successful prototype coupling into the permanent contract. When you replace a scaffold, preserve it until the new behaviour proves itself, then report which signal actually caused the wake so we can retire the old path deliberately rather than by assumption.