Habit, Not Hack: Giving Feedback That Stays About the Work (Mentor)

Feedback is about the work. Whether it lands that way is decided before you send it — and that part is yours.

The comments weren't harsh.

"Needs more clarity." "Logic jumps here." "This isn't convincing yet."

Dr. Rao wrote them the way she always did — quickly, precisely, in the margins of her trainee's draft, one she'd read carefully at eleven at night, because that was when the time existed. Twenty-three comments. Every one of them accurate. Every one of them, she would have said, useful.

The revision came back three weeks later. Technically responsive, strangely flat. The sections she'd flagged were fixed, but the argument hadn't grown — it had retreated. Safer claims. Hedged language. The interesting idea from the first draft, the one that needed work but was worth the work, had quietly disappeared.

At their next one-on-one, she asked about it. There was a pause that lasted slightly too long.

"Honestly? The feedback felt like maybe this project isn't going well. So I tried to make it... less wrong."

Dr. Rao was genuinely surprised. Not going well? The project was going fine. The draft was a normal first draft. The comments were the same comments she'd write on a colleague's manuscript, on her own manuscript. They were about the document.

But they hadn't landed as information about the document. They had landed as a verdict — and the revision she'd received wasn't a response to her comments. It was a response to the verdict.

How This Happens

The gap is not carelessness. It has a logic to it, built over years of operating on the other side of the feedback exchange.

By the time you're supervising, critique has become a neutral tool. You've received thousands of comments, written thousands more, watched brutal review cycles produce good papers. Somewhere along the way, "this isn't convincing yet" stopped carrying any charge for you. It's shorthand. It means: strengthen this and the paper improves. You hear the full sentence even when you only write the fragment.

The trainee doesn't have that translation layer yet. They haven't accumulated the evidence that critique and belonging are separate things — that people whose work gets marked up heavily still succeed, still get respected, still stay in the field. For someone still establishing whether they belong, every comment arrives carrying a second, unwritten question: and what does this say about me?

Written feedback makes it worse. In person, the frame travels with the critique — your tone, the offhand "this is normal for a first draft," the fact that you're clearly engaged rather than disappointed. In a margin comment, all of that is stripped out. What remains is the substance without the frame. And when the frame is missing, the reader supplies one. Early-career readers, reliably, supply the harshest available.

The efficiency that makes margin comments practical for you is exactly what makes them expensive for them.

What Dr. Rao Tried

The conversation surfaced the problem. It didn't solve it — Dr. Rao wasn't going to start writing feedback essays at eleven at night, and softening her standards would have been a disservice dressed up as kindness.

Instead she tried something structural, on the next draft. Before sending her comments, she read back through them one time with a single question: is this a task, or does it read like a verdict?

She found the comments sorted into two kinds.

Verdict-shaped: "This isn't convincing." "Logic jumps here." "Unclear." Accurate — but they describe a failure without naming the work. They tell the reader something is wrong and leave them alone with what that means.

Task-shaped: "The transition between paragraphs three and four loses the thread — the causal claim appears before the mechanism is introduced. Move the mechanism up." Same standard. Same directness. But the subject of the sentence is the document, and the comment comes with its own exit: here is the thing to do.

Converting the first kind into the second took her eleven extra minutes. Then she added two sentences at the top of the file — not praise, frame:

"This is a normal first draft with a genuinely interesting core idea — the comments below are where I'd focus the revision. None of this is a concern about the project."

Two sentences. The kind of thing she'd always assumed was understood.

What Changed

The next revision came back different. The flagged sections were fixed — but the interesting idea was still there, and it had been developed rather than defended. The one-on-one about the draft spent its time on actual questions: whether the framing in section two was right, whether the third analysis was worth adding. Less energy managing an unspoken emotional weather system, more energy on the work.

And something Dr. Rao hadn't expected: the trainee started disagreeing with her. Pushing back on one comment, defending a structural choice in another. Which is what a researcher in training is supposed to do — and what someone processing feedback as a verdict on their belonging almost never does. You don't argue with a verdict. You only argue with information.

The eleven minutes weren't a cost. They were the difference between feedback that produced compliance and feedback that produced a colleague.

The Habit: Send the Frame With the Critique

The read-back pass is one version of the habit. The underlying practice can take different forms, but the core move is always the same: before feedback leaves your hands, make sure the separation between the work and the person is carried in the feedback itself — not left for the reader to construct.

Before sending a marked-up draft or delivering critique in a meeting, try one of these:

Open with one sentence of frame. Not a compliment — a frame. Where does this draft sit in the normal arc of a project? What is the feedback for? "This is where I'd focus the revision" positions everything that follows as a work plan. Silence positions it as a judgment, because silence always does.

Convert verdicts into tasks. Reread each comment and ask: "Does this name what the document should do, or only what it fails to do?" "Unclear" is a verdict. "Define the term before using it in the claim" is a task. Every comment that names the specific move is one that cannot easily be generalized into an identity statement.

Say the quiet part once. The sentence you assume is understood — this is normal, the project is fine, heavy markup means I'm engaged — is understood by you, because you have fifteen years of evidence for it. Your trainee may have none. Say it explicitly, once per feedback cycle. It costs one line. It reframes twenty-three.

A Note on What This Isn't

This is not a habit about softening standards, sandwiching every critique between compliments, or protecting trainees from the experience of having their work marked up. Research runs on rigorous critique, and a trainee who never receives it is being prepared for a field that doesn't exist.

The habit is about addressing, not diluting. A comment can be blunt, demanding, and entirely unsoftened — and still be unmistakably about the document. Precision does the work that padding pretends to do. "This argument fails, and here is exactly where and why" is harder feedback than "hmm, not quite convincing" — and it is also far less likely to be read as a statement about the author, because it leaves no vacuum for the reader to fill.

The separation also runs both ways. Sometimes the feedback genuinely is about calibration — about instincts and judgment, not just this document — and pretending otherwise would be its own failure of mentorship. The habit is not about avoiding those conversations. It is about making sure they happen deliberately, named as what they are, rather than leaking accidentally through the margins of a draft.

For Trainees

The frame Dr. Rao learned to send is something you can also learn to ask for — and to build yourself when it doesn't arrive.

If feedback lands without context, the highest-leverage question you can ask is not about any individual comment. It's: "Where does this draft sit, overall? Is this normal markup, or is something off track?" Most supervisors can answer that in one sentence, and most are surprised the question needed asking. And when the spiral starts before you can ask — I should have known this, everyone else gets it — the companion piece to this post walks through a two-column practice for separating what the feedback says from what it doesn't.

The comments in the margin are about the work. Learning to write them so they stay that way is not a lower standard.

It's a higher one.

That's not a hack. That's a habit.

✨ The GLC Feedback Reflection Guide includes a feedback framing section for structuring what you're saying, why you're saying it, and what happens next — because feedback that gets acted on is feedback that was delivered clearly.