26.278 - Kindness That Won't Say the Hard Thing

“Psychological safety is not about being nice; it's not about letting people off easy and being comfortable.”

— Amy Edmondson, 2012, quoted in Teaming in the Twenty-First Century

✏️🗣️✨

Amy Edmondson studies how people learn together when speaking up carries an interpersonal risk.

Psychological safety concerns whether people believe they can ask questions, acknowledge mistakes, and raise concerns without humiliation or punishment. It does not mean that those conversations will always feel comfortable. In a 2012 Harvard Business School interview, Edmondson emphasized that speaking directly and holding high expectations can coexist with a climate that allows people to learn.

That distinction helps us examine a familiar difficulty: wanting someone to feel supported while noticing something they need to reconsider. Care involves the conditions in which a concern is raised, as well as the information it contains. For today’s reflection, Edmondson’s work invites us to ask whether both people have room to contribute: the reviewer to name a concern, and the author to explain, disagree, or point out what the reviewer missed. The willingness to offer feedback needs a corresponding willingness to revise our own view.

The Comment You Delete

Two-week delivery / staffing assumption?

Imagine typing that note beside the first page of a colleague’s proposal. The opening promises delivery within two weeks. Farther down, the delivery section describes the work but does not explain who will be available to do it. You read the passage again. Perhaps you missed something. Perhaps a separate schedule settles the question.

Your colleague has asked, “Is this ready to send to the client?”

There is plenty to admire. The proposal explains the client’s problem clearly. The examples are relevant. You know the author has spent several evenings bringing it together, and you want that effort to be recognized. You also know what another question might mean: another revision, another conversation, another delay at the end of a long day.

So you delete the annotation.

“This is strong. The opening is especially clear. You’ve done a great job with it.”

Every sentence is sincere. The concern remains.

The small decision is worth staying with. No client has complained. No deadline has been missed. The author may have a perfectly sound plan. The issue is that a question relevant to the decision has disappeared from the review, and the person making that decision has no way to know it was there.

From the reviewer’s side, the omission can feel considerate. You did not want to turn someone’s accomplishment into another problem to solve. You did not want a tired colleague to hear your question as a judgment on their competence. Perhaps you remember receiving criticism when you most needed encouragement, and you are trying to spare someone that experience.

These are understandable concerns. Delivery matters. So does the amount of attention a person has available. Yet a change takes place when “I want to raise this carefully” becomes “I will let them proceed without knowing I noticed it.” The author’s possible discomfort becomes the reason for withholding information the author explicitly asked you to help assess.

There is also your own discomfort to consider. Asking about the schedule might make you sound distrustful. It might reveal that you failed to understand the document. Your colleague might be irritated. Warm praise avoids those possibilities, at least for the moment.

You do not have to decide which motive is the real one. Care and relief can occupy the same reply. The more useful question is what the reply enables the other person to do.

The proposal is still open. The author is deciding whether to send it. This is the moment when a specific question could make a difference, including the difference of discovering that no change is needed. Once you remove the question, the author must make that decision with less information than you could have offered.

The Comfort of an Agreeable Reply

An agreeable exchange can feel like immediate evidence of success. A warm reply may suggest that someone feels encouraged, and we can return to our work without an unresolved disagreement. Whether we helped them assess the proposal is a separate question, easier to overlook when the conversation ends pleasantly.

The value of an uncomfortable question is harder to see. It may require time before it proves useful. It may lead to an answer we should already have known. Even when it identifies a genuine problem, the first response may be annoyance rather than appreciation. If we judge an interaction only by how it feels when it ends, reassurance has an advantage before the substance has been considered.

This creates a particular difficulty when someone asks for a readiness check. In that setting, praise can communicate more than admiration. “This is strong” may be heard as “I found no material reason to hold it back.” The reviewer may intend encouragement while the author reasonably receives an assessment.

We cannot solve that mismatch by expecting the author to read what we left unsaid. They do not know that the compliment replaced a question. They cannot distinguish an uneventful review from a review in which something important was noticed and then withheld.

Still, an ordinary readiness check does not usually call for a complete inventory of imperfections. A proposal can always be rearranged. A sentence can almost always be tightened. The reviewer’s preferred wording may add little. Flooding the author with observations can make the consequential issue harder to find and the work harder to finish.

The purpose of the review provides a boundary. If the author is testing an early idea, detailed corrections may arrive too soon. If they are deciding whether to make a commitment to a client, the basis for that commitment belongs in the conversation. The question is whether the observation bears on the decision at hand, rather than whether the reviewer can defend it as technically true.

That boundary also helps separate an observation from a verdict. “I could not find the staffing assumption behind the two-week promise” identifies something the author can locate, explain, or revise. “You haven’t thought this through” reaches much further. It turns a gap in the reviewer’s understanding into a conclusion about the author’s thinking.

Perhaps the author did think it through. Perhaps the schedule sits in a document the reviewer did not receive. Perhaps the opening assumes a condition that both colleagues understand but the client will not. Each possibility calls for a different response. A useful review leaves those possibilities available long enough to investigate them.

Feedback itself offers no guarantee of improvement. In their meta-analysis, Avraham Kluger and Angelo DeNisi found that feedback interventions improved performance on average, yet more than one-third reduced it. Their analysis suggested that feedback became less effective as attention shifted away from the task and toward judgments about the self. That is a reason to examine the form and purpose of a comment, not simply applaud the willingness to give one. Kluger & DeNisi, 1996

Power changes the exchange as well. An invited review between peers differs from a manager’s evaluation. A manager may regard “Would you like some feedback?” as an open invitation, while an employee hears a question they cannot safely decline. Warm language does not erase that difference. A person with authority needs to make the scope of the discussion clear and allow questions about the assessment itself.

The reverse situation matters too. An employee who has seen people punished for questioning a superior may have good reason to remain quiet. Asking that employee to be more candid does not resolve the conditions that made candor costly. Responsibility includes how those with power receive inconvenient information and what happens to the person who brings it.

Edmondson’s study of 51 work teams found an association between team psychological safety and learning behavior. It gives us grounds to take those surrounding conditions seriously. It cannot tell us how this colleague will react to this comment. Edmondson, 1999

In our proposal review, the invitation is explicit and the concern is relevant. Even here, the author owes the reviewer neither gratitude nor immediate agreement. The aim is to make a material question available for consideration. Keeping that aim modest helps prevent a caring intervention from becoming a demand that the other person recognize our superior judgment.

Give the Concern a Usable Form

Return to the deleted annotation. Its first version was shorthand for your own thinking. Before sending it, you can do a little more work so the author does not have to reconstruct what troubled you.

“The opening promises delivery within two weeks, but I could not find the staffing assumption that supports it. Is that covered elsewhere, or should we make the condition explicit before this goes out?”

The wording locates the issue and makes your uncertainty visible. It also connects the concern to something the author is deciding now. A client might read the opening as an unconditional commitment. Checking its basis could confirm that the proposal is ready or reveal a condition worth stating.

“I could not find” matters only if you mean it. It gives the author room to show you something you missed. If the schedule is already included and clearly linked, the next responsible sentence may be, “You’re right. I overlooked that.” A question used merely to disguise a settled accusation offers little room for a real answer.

Suppose instead that the staffing plan exists but the proposal does not make the dependency clear. Then the conversation can become more precise: would the client understand when the two-week period begins and what it depends on? The author may decide that a short qualification is enough. You may agree. The issue need not expand into a general discussion of diligence, communication style, or the quality of the entire project.

There is another possibility: the author understands your concern and decides against changing the draft. They may have information you cannot evaluate. They may judge the risk differently. They may be wrong. An invited reviewer can explain why the issue matters and check that the concern has been understood without assuming ownership of the final decision.

If you share responsibility for approving the commitment, that responsibility should be stated directly. “I need us to resolve the delivery assumption before I can approve my part” is different from pretending to offer optional advice while expecting obedience. Clarity about your role is part of clarity about the work.

Timing deserves the same honesty. Someone who says, “I’ve had an exhausting day and need a little encouragement,” has made a different request from someone asking, “Is this ready to send?” Listening to the first request does not require searching for a flaw. Encouragement can be the appropriate and complete response.

If a material issue in shared work also needs attention, you can identify that need without turning the moment into an unannounced review: “There’s one delivery question we need to settle before tomorrow’s submission. Would later this afternoon work, or should we look at it in the morning?” The available choices need to be real. Offering a choice of times while knowing the work must leave in ten minutes creates another misleading reassurance.

These judgments take more effort than either automatic praise or automatic correction. They require attention to what was requested, what is consequential, what you know, and what authority you have. They also require accepting that a carefully offered concern can still disappoint someone. Consideration can reduce needless injury; it cannot promise a conversation without feeling.

The intention to protect a relationship can narrow our view to the discomfort immediately in front of us. Beyond it sits another person’s need to make a decision with the relevant facts and questions available. Their present ease matters. So does their ability to act before a commitment becomes harder to change.

We can take both seriously without treating the recipient as too fragile for information or treating ourselves as its unquestionable authority. The author can be tired and capable. The reviewer can be caring and mistaken. The proposal can be strong and still need one condition clarified.

Your praise therefore does not need to disappear. “The way you connect the client’s problem to the proposed work is especially clear” remains true. It names something worth preserving. The delivery question names something worth checking. Neither statement needs to cancel the other, and the concern should not be softened until it sounds optional when you believe it is material.

This time, you keep the comment beside the two-week promise. You replace the shorthand with the question the author can answer. You send the review with the specific praise that you meant from the beginning.

Beside the praise for the opening, the delivery question is now visible. The author can see what you admire and what you still need to understand.

✏️🗣️✨

Practice: The Comment Worth Sending

Estimated time: 8 minutes.

Choose a low-stakes review, real or imagined, in which you noticed something relevant but hesitated to mention it. Identify what the person asked you to assess, the specific issue you noticed, and why it matters to their current decision. Then name one thing you may be missing.

Draft a brief observation and a question the person could answer. Check whether your role and the timing make the comment appropriate, and whether you have left room to be corrected. The intended outcome is a concern expressed clearly enough to consider without turning it into a judgment of the person. You do not need to send it.

Scientific Summary

Edmondson’s (1999) multimethod field study examined 51 teams within one manufacturing company. Team psychological safety was associated with learning behavior. This is a team-level finding from an organizational setting, not proof that any particular wording will make an individual feel safe or improve a relationship.

Kluger and DeNisi’s (1996) meta-analysis included 607 effect sizes and 23,663 observations. Feedback interventions improved performance on average, while more than one-third reduced performance. Those results concern the interventions included in the analysis; they do not establish a one-in-three probability that an ordinary conversation will cause harm.

Together, these sources support examining the conditions and effects of feedback rather than treating either comfort or candor as sufficient. The proposal scene, suggested language, and eight-minute practice are editorial applications, not a tested intervention. The article’s central ethical claim is that a relevant, invited concern deserves consideration while the recipient can still use it, alongside careful attention to uncertainty, timing, power, and choice.

References

• Edmondson, A. (1999). Psychological safety and learning behavior in work teams. Administrative Science Quarterly, 44(2), 350–383. https://doi.org/10.2307/2666999

• Kluger, A. N., & DeNisi, A. (1996). The effects of feedback interventions on performance: A historical review, a meta-analysis, and a preliminary feedback intervention theory. Psychological Bulletin, 119(2), 254–284. https://doi.org/10.1037/0033-2909.119.2.254

• Starvish, M. (2012, December 17). Teaming in the twenty-first century. Harvard Business School Working Knowledge. https://www.library.hbs.edu/working-knowledge/teaming-in-the-twenty-first-century

Legal Disclaimer: The content published on Lucivara is provided for informational, educational, and reflective purposes only and is not intended to constitute medical, psychological, legal, or professional advice. Lucivara does not diagnose conditions, prescribe treatments, or provide therapeutic or professional services. Readers are encouraged to consult qualified professionals regarding any personal, medical, psychological, or legal concerns. Use of this content is at the reader’s own discretion and risk. Copyright Notice: © 2026 Lucivara. All rights reserved. This content is protected by copyright laws and is intended for individual, personal, and non-commercial use only. No part of this publication may be reproduced, distributed, transmitted, displayed, published, or broadcast in any form or by any means without prior written permission from Lucivara. Acceptable Use: The content published on Lucivara is intended for individual, personal, and non-commercial use only. Readers may access, read, and engage with the content for their own reflective, educational, or informational purposes. Except for such ordinary human use, no portion of this content may be copied, reproduced, redistributed, republished, transmitted, stored, scraped, extracted, indexed, modified, translated, summarized, adapted, or incorporated into derivative works without prior written permission from Lucivara. This restriction expressly includes, without limitation, the use of Lucivara content for training, fine-tuning, prompting, testing, benchmarking, or operating artificial intelligence systems, machine learning models, automated agents, bots, or any other computational or data-driven systems, whether commercial or non-commercial. By accessing or using this site, readers acknowledge and agree to Lucivara’s Terms and Conditions.