Copy a block, paste it somewhere else, and it still feels like the same thing. The visible action is simple. The interesting question is which representation crossed the boundary and which stayed inside the application.
The published investigation
The REA project’s case study traces a clipboard call in Notion Desktop 7.6.1 through the preload and IPC wrappers to a main-process receiver. The documented channel is notion:clipboard:write, and the receiver includes a sender check before the clipboard write.
The study also explains a separate mechanism in saved web assets: an HTML marker links a copied item to locally retained structured block data. Its sources and probes have their own version context. REA AI has not repeated that package inspection or those probes.
Why “copy text” is an incomplete description
A clipboard operation can carry several representations of the same selection. Plain text is useful outside a rich editor; HTML can preserve formatting. Application-specific structure may need a different path.
That observation gives you better questions for any editor investigation. What data is sent to the system clipboard? What remains in application state? How does the receiving path decide whether the richer representation still corresponds to the copied item?
The answers need evidence from that application. A generic diagram of an Electron bridge does not establish its clipboard format.
Wrapper code is part of the explanation
In a production package, the method exposed to the page may be created by a helper. A name such as write tells you little until you inspect the helper that constructs it and the transport operation it uses.
The useful habit is to preserve the exact call site, then resolve the wrapper before drawing the next arrow. If a channel is dynamic, record the expression and the unresolved value. A chain with one honest gap is more useful than a complete chain built from assumptions.
Our small IPC lab removes most of that production complexity so you can practice the same reasoning on readable files.
Questions for your own editor
Consider copying formatted text into a plain text field, then pasting within the same editor. What changes? Does a new app session preserve the richer behavior? Does an expired local item change the result?
These are proposed experiments, not claims about every Notion version. Their purpose is to separate observable representations and the state each depends on.
For your own implementation, define a simple fallback first. Rich structure is useful, but a reader should still get sensible text when a private representation is unavailable. That product decision needs its own tests, independent of the analysis.
Keep the scope of the source
The original case study distinguishes packaged desktop JavaScript from cached web assets and extracted-module tests. Keep those distinctions when discussing it. A saved module probe is not a captured desktop UI session, and an older package is not proof of current behavior.
Read the original study for its exact excerpts, version notes and limitations, then use the evidence checklist to structure your own report.