Easy to Work With Beats Brilliant

Your output is scored by the org; your interactions are scored by your peers — and the org reads the peers' score first.

Technically strong engineers run on a simple model: the work speaks for itself. Ship good code, solve hard problems, be right in design reviews, and the score takes care of itself. The model is wrong — not because quality doesn't matter, but because quality is only one of two channels the org listens to. The other channel is what people say about working with you, and it carries more weight than most engineers ever realize.

The hidden channel: peer feedback

A surprising share of career outcomes flows through sentences you never hear:

• Peer feedback in performance reviews. Your manager sees a small slice of your work directly. The rest of their picture is assembled from what your collaborators write about you. Three reviews that say "brilliant but exhausting" outweigh a quarter of great commits.
• The informal poll before staffing. Before you're put on a high-visibility project, someone asks around: "would you work with them again?" This poll is never announced, has no rubric, and is decided by the accumulated residue of every interaction you've had.
• The pre-transfer DM. Before a manager accepts you onto their team, they message someone who knows you: "what's she like to work with?" One sentence comes back. That sentence was written by you, over months, without your knowing it.

The mechanism underneath all of these: the org's real question is never just "are they good?" It is "does adding this person to a team make the team better or worse?" A team is a multiplier system. An engineer who produces 1.3x output but drags three teammates to 0.9x each is a net negative, and orgs — the healthy ones — do the math.

The abrasiveness tax

Here is the part that stings for people who are usually right.

Being right in a way people can't accept produces the same org outcome as being wrong. The idea dies either way — your correct design gets ignored because accepting it now feels like losing — and you additionally pay reputation for the interaction. You lose twice. The engineer who was wrong but pleasant paid once.

Every collaboration leaves a one-line residue. Nobody remembers your argument's structure six months later. They remember one compressed line: "helpful." "Condescending." "Made my code better." "Made me feel stupid." "Always delivers." "Landmine." These residues, aggregated across everyone you've touched, are your peer-feedback layer. There is no appeal process. There is no place to attach a footnote explaining you were technically correct.

Brilliance amplifies the tax rather than offsetting it. This is the counterintuitive core. The more habitually right you are, the more your delivery gets scrutinized — because when you're right, the only degrees of freedom left for people to evaluate are how you were right. And a strong engineer who is abrasive does more damage than a weak one: their comments carry authority, so their contempt lands harder, and people reorganize their behavior to avoid them — routing around the very person who has the most useful information. This is why "brilliant jerk" is a named firing category at healthy companies. It isn't moralizing. It's arithmetic: their net team effect is negative, and the brilliance is exactly what makes the negativity load-bearing.

What easy-to-work-with actually means

This is where engineers most often mishear the lesson. Easy to work with does not mean agreeable, deferential, or conflict-avoidant. A pushover is also hard to work with — you can't trust their yes. The actual skill set:

• Disagree with the idea while visibly respecting the person. "I think this design breaks under concurrent writes — walk me through how you're thinking about that?" attacks the design and enlists the designer. "This won't work" attacks both.
• Code review comments that teach, not score points. Ask questions instead of issuing verdicts. "What happens if X is null here?" gets the bug fixed and leaves the residue "made my code better." "This is broken" gets the same bug fixed and leaves "made me feel stupid." Identical technical outcome; opposite reputational outcome.
• Make others' work better and let them keep the credit. You quietly fixed the perf issue in their PR, or suggested the framing that saved their doc. Resist narrating it. The credit you release comes back as advocacy; the credit you claim comes back as resentment.
• Be predictable. No moods. No landmines nobody warned the new hire about. You deliver what you said, or you say early that you won't. Predictability is what lets people build on top of you, which is the definition of being valuable to a team.
• Make it cheap to ask you dumb questions. The engineer people are afraid to ask questions of has amputated themselves from the information flow. Problems route around them, they hear about incidents late, and their expertise stops compounding because nobody brings them raw material.

The compounding

The payoff structure is nonlinear. People who are easy to work with get pulled into more rooms — and rooms are where interesting problems, visibility, and sponsors live. They get staffed on better projects because the staffing poll comes back clean. Most valuably, they get defended in calibration by people they never asked for anything — a peer volunteers "actually, he unblocked us twice last quarter" in a meeting you're not in. That is sponsorship emerging organically, and it's how the shadow org chart routes opportunity. Abrasive brilliance compounds too — in the wrong direction. Each residue makes the next collaboration start from a deficit.

Cautions: this is not agreeableness

Do not overcorrect into silence. Swallowing your objection to preserve smoothness is its own failure mode — the org loses your information and you lose your integrity as a signal source. The skill being described is narrower and harder: separate the strength of your position from the friction of your delivery. Hold the position at full strength; sand the delivery down to zero unnecessary friction. You can be immovable on the technical point and gracious about the human across from you at the same time. That combination — firm and easy — is rare, and it is what the informal poll is actually screening for.

Diagnostic test

If your teammates were anonymously asked "would you fight to have them on your next project?" — do you know, honestly, what they'd say?