I’ve been using Writing Blocks for long-form fiction, and I’m confused about the edit suggestions shown directly underneath them.
For example, after generating a Writing Block, I sometimes see suggestions such as:
“Combine short sentences into more natural paragraphs.”
“Make the shock of the character’s stalled growth more impactful.”
“Increase the tension of the battle scene.”
Because these suggestions appear directly under the Writing Block, the UI makes them look like editing actions for the existing document.
However, clicking one does not actually modify the current Writing Block. Instead, ChatGPT generates another version of the response from the same point.
For example:
Response A
→ Click “Combine short sentences into natural paragraphs”
→ Response A-2 is generated
The original A is not edited.
Actual behavior
The suggestion works essentially like an automatically generated follow-up prompt.
It creates another version of the response rather than editing the existing Writing Block in place.
Expected behavior
For something presented as an editing suggestion directly underneath a Writing Block, I would expect something closer to:
Writing Block A
→ Click editing suggestion
→ Writing Block A itself is updated
→ Previous version remains available through Undo / version history
This would make Writing Blocks behave more like an actual document-editing interface.
Why this matters for long-form writing
For short text, generating another version may be fine.
But for long-form fiction, repeatedly creating A-2, A-3, A-4, etc. is not very useful. I still have to decide which version becomes the canonical version before continuing with the next scene.
If the next scene is B, I now have to make sure I continue from A-2 rather than A. After several revisions, this becomes unnecessary version/branch management.
In practice, clicking the suggestion is not very different from manually typing:
“Please combine the short sentences into more natural paragraphs.”
That makes me wonder why these suggestions are attached specifically to the Writing Block UI if they do not actually edit the block.
Another confusing aspect
Sometimes the suggestion itself identifies a problem that I had already explicitly asked ChatGPT to avoid in my instructions.
For example, my prompt may already say not to use repetitive short sentences, but after generation the UI suggests:
“Combine the short sentences into natural paragraphs.”
If the system can detect that issue after generation, it would be preferable either to:
-
respect the instruction during the initial generation, or
-
let me apply the detected fix directly to the existing Writing Block.
Suggested improvement
Please consider adding an “Apply to current Writing Block” behavior for these suggestions.
Ideally:
Generate Writing Block → click suggested edit → update the same block → keep previous revision in version history
If generating a separate response is intentional, the UI should probably make that clearer, because the current placement strongly implies an in-place editing feature.
For long-form writing workflows, true in-place revision would make Writing Blocks much more useful.
