Subjects ยท Literature & Language Arts
Literature & Language Arts: Draft-and-critique loops
Get feedback specific enough to act on, across several cycles, without ever letting the tool write the fix for you.
What you'll be able to do: Get feedback specific enough to act on, across several cycles, without ever letting the tool write the fix for you.
Generic feedback is not feedback
"Make this more analytical" and "the dialogue could be stronger" tell you a destination and nothing about the route. Structured critique names the exact sentence or paragraph, says precisely what isn't working there, and stops, it does not offer the replacement sentence, because writing the replacement is the part that builds the skill.
Draft-and-critique loops draft-and-critique loops: submit an actual draft, get targeted critique, revise it yourself, resubmit, repeat. The mechanism is the cycle, not any single round, a single critique pass tells you one problem; several cycles, each targeting a different craft element, is what a real revision process looks like, compressed.
The danger specific to this method: if the tool critiques and rewrites, you haven't looped through critique, you've outsourced composition one clause at a time. The rule to hold: it describes the problem precisely enough that you can fix it. It does not supply the fix.
How to run it
- Pick one target per pass. Structure this round, evidence next round, sentence-level voice after that. Critiquing everything at once produces feedback too broad to act on.
- Submit the actual draft, not a summary of what you were going for.
- Demand specificity. Point to the exact sentence or paragraph, and say what's not working there and why, not a general impression.
- Revise it yourself. The critique names the problem; you write the fix.
- Resubmit and check whether the specific problem is gone, not whether the whole piece now feels better, which is too vague to verify.
Here's my draft: [paste]. Critique only the [structure / evidence / voice. name one] this pass. Point to the specific sentence or paragraph and tell me what's not working there and why. Don't rewrite it for me, describe the problem precisely enough that I can fix it myself.
What a targeted pass actually looks like
Structure pass, on an essay about unreliable narration in Ishiguro's The Remains of the Day. Does every paragraph's opening sentence advance the thesis about Stevens's self-deception, or does it just introduce another scene in sequence? A structure critique names the paragraphs doing the second thing, it doesn't rewrite their topic sentences for you.
Craft pass, on a short story draft. Does the opening earn attention, or does it explain before it dramatises? Is dialogue doing double duty, characterisation and plot movement at once, or is it carrying information the narration could deliver more efficiently? Does the ending earn its emotional beat through what's been built, or does it simply assert the feeling in the final line?
Argument pass, on a persuasive essay. Is every claim backed by evidence sufficient to support its strength, or does a strong claim rest on a single weak example? Does the essay handle the strongest counterargument, or a weak version of it? This pass should be entirely separate from a prose-style pass, fixing sentence rhythm doesn't fix an argument that doesn't hold.
Across the disciplines
Composition. As above, separating structural, evidential and stylistic critique into distinct passes is the core discipline of the method.
Creative writing. Workshop-style critique benefits enormously from named, single-target passes: one round purely on point-of-view consistency, one purely on pacing, one purely on dialogue. A single "how's this?" pass collapses all of them into vague impressions.
Rhetoric. Critique a persuasive draft on logical structure and evidence sufficiency separately from delivery and tone, a rhetorically polished argument with a hole in its logic and a rhetorically flat argument with airtight logic have different, non-overlapping fixes.
Linguistics. If you're writing a research report on a language variety, critique the claims separately from the prose, does each generalisation about the data actually follow from the examples given, independent of how well the paragraph reads.
Pitfalls
- Letting critique become rewriting. The tell is a critique response that contains a full replacement paragraph rather than a description of the problem.
- Critiquing everything at once. Feedback covering structure, evidence and style in one pass is too diffuse to act on; you'll fix the easiest issue and miss the important one.
- Treating the first loop as final. One pass finds one layer of problems; fixing it usually exposes the next layer, which is why it's a loop.
- The tell: you can't describe, in your own words, what was wrong with the previous draft. If the fix just appeared, the critique wasn't specific enough, or you didn't do the fixing.
Try this today
Take a draft you're partway through. Submit it for a single targeted pass, structure only, and ask for the exact paragraph that's the problem, not a rewrite. Fix it yourself before asking for the next pass.