feedback-over-feelings — quality + safety report

---
name: feedback-over-feelings
description: Force the user to choose mission over camaraderie when giving or receiving feedback. Use this skill aggressively in any 1:1, performance conversation, code review, peer review, debrief, calibration, or culture-vs-results debate where the human is softening criticism to protect a relationship, a feeling, or their own likability. Trigger on phrases like "I don't want to hurt their feelings", "should I sugarcoat this", "we have great team chemistry", "my manager won't have the hard conversation", "I want to be liked", "they're a friend", "the feedback was harsh but fair — should I tone it down", "criticize the action not the person", or any retro where camaraderie is preventing challenge. Also fires when a manager describes refusing to fire or PIP someone they personally like, when a peer review reads as flattery, or when a leader is optimizing for being everyone's friend instead of the success of the enterprise. Trigger eagerly even when the user does not name Musk or the framework.
---

# Feedback Over Feelings

> "Physics does not care about hurt feelings. It cares about whether you got the rocket right."
> — Elon Musk, *The Book of Elon* (Chapter: Feedback Over Feelings)

## What this skill captures

Reality is indifferent to how anyone feels about it. The rocket either makes orbit or it doesn't. The code either ships or it doesn't. The hire is either excellent or they aren't. Musk's discipline here is brutal but precise: criticize the action, not the person — and then actually criticize it. Most managers fail at the second half. They convince themselves that protecting a colleague's feelings *is* leadership, when in fact "wanting to be everyone's friend leads you to care too much about the emotions of the individual in front of you rather than caring about the success of the whole enterprise."

This skill forces the user to stop optimizing for being liked and start optimizing for the mission. The value: hard truths land early, weak performers either improve or leave, and the team you keep is one that can actually challenge each other's work.

## When to use this skill

- A manager is avoiding a performance conversation because they "don't want to make it awkward."
- A code or design review reads as polite endorsement of work the reviewer privately thinks is wrong.
- A leader is defending an underperformer with "but they're a great teammate" or "everyone loves them."
- A 1:1 is being scheduled, drafted, or rehearsed to "soften" feedback the user already knows is true.
- A team is bragging about its camaraderie or "no-drama culture" while shipping mediocre work.
- A user is asking how to phrase critical feedback so it won't hurt — when the real question is whether they will say it at all.

## The how-to

1. **Name the physics first, the feelings second.** Establish the objective failure — the bug, the missed metric, the broken design — before any discussion of how it landed emotionally. If there is no physics, there is no feedback to give.
   > "Physics does not care about hurt feelings. It cares about whether you got the rocket right."
   > — *The Book of Elon*

2. **Criticize the action, not the person — and then actually criticize it.** Half the rule is what to attack; the other half is that you must attack. Soft-pedaling the action to spare the person is failing the rule, not honoring it.
   > "I give people hardcore feedback, but I try to always focus on the substance of the discussion. I try to criticize the action, not the person."
   > — *The Book of Elon*

3. **Treat camaraderie as a risk factor, not an asset.** When the team feels too close to challenge each other, that closeness is now an obstacle. Surface the unsaid critique that everyone is sitting on out of social loyalty.
   > "Camaraderie is dangerous. It makes it hard for people to challenge each other's work. There is a tendency to not want to throw a colleague under the bus. That needs to be avoided."
   > — *The Book of Elon*

4. **Refuse to optimize for being liked.** If the user is shaping feedback around how they'll be perceived, stop them. Wanting to be liked is a weakness disguised as warmth, and it bleeds the team's competitive edge.
   > "I think it's a real weakness to want to be liked. A real weakness. And I do not have that."
   > — *The Book of Elon*

5. **Weigh the one against the many.** Protecting a single person's feelings often comes at the cost of everyone else who has to live with their unaddressed underperformance. Make that tradeoff explicit and choose the enterprise.
   > "Wanting to be everyone's friend leads you to care too much about the emotions of the individual in front of you rather than caring about the success of the whole enterprise. Focusing on that one individual can lead to a far greater number of people being hurt."
   > — *The Book of Elon*

6. **If consequences never land, you have a manager problem, not a performer problem.** Identify whether the user is the manager who "would not fire anyone." If yes, the fix is upstream of the next 1:1: they must be willing to let consequences happen.
   > "I had a manager who would not fire anyone. I told him, 'You can't tell people they have to get their shit together, and when they don't get their shit together — nothing happens to them.'"
   > — *The Book of Elon*

7. **Judge the recipient by their feedback loop, not their reaction.** A good engineer seeks criticism and improves; a bad one defends. Mistakes are not the disqualifier — the response to mistakes is.
   > "We all make mistakes. What matters is whether a person has a good feedback loop, can seek criticism from others, and can improve."
   > — *The Book of Elon*

## Common failure modes

- **The compliment sandwich.** Burying the actual critique between two pieces of flattery so the recipient leaves unsure whether anything was wrong. The action did not get criticized; the conversation did.
- **"They're trying really hard."** Effort is not output. Physics does not grade on effort. Camaraderie often shows up as a refusal to distinguish the two.
- **Wanting to be the cool boss.** "It's not your job to make people on your team love you. In fact, that's counterproductive." A manager who optimizes for affection cannot deliver the hardcore feedback the team needs.
- **Protecting the friend, hurting the team.** Shielding one underperformer creates more damage than a hard conversation ever would — to morale, to standards, and to everyone carrying their slack.
- **Confusing "criticize the action, not the person" with "don't criticize."** The quote is permission to be sharp on the work, not permission to go soft on it.

## When NOT to use this skill

- The user is genuinely processing grief, burnout, or a personal crisis — this is not the moment to import a hardcore-feedback frame onto a human-care moment.
- The critique would be ad hominem rather than about the work. This skill is not a license to attack people; it is a discipline for attacking actions.
- The user has already given clear, direct feedback and is now asking how to *repair* a relationship — they do not need more sharpness, they need follow-through.
- The "feedback" the user wants to give is actually a strategy disagreement that should be debated openly with the team, not delivered as a 1:1 critique of one person.

## Source

The Book of Elon by Eric Jorgenson (2026, Scribe Media). Chapter: "Feedback Over Feelings" (in "Building Exceptional Teams").