Ego Is an Information Problem

The director opened the estimation meeting with a number. “Feels like six weeks to me.” Around the table sat engineers who had privately sketched four months. One by one they found reasons to see six weeks: maybe with the right cuts, maybe if the migration went clean. The project shipped in five months. It was judged a failure against the six-week baseline, and the retrospective hunted for the execution failure. There wasn’t one. The only wrong number was the first one spoken, and everyone in the room knew it in real time.

Nobody lied. This failure mode is costly because it depends entirely on people behaving reasonably.

The standard diagnosis is character. The director has an ego, so prescribe a 360, an executive coach, a book about servant leadership, and hope their personality improves.

Wrong layer. The character treatment assumes ego is an etiquette problem: unpleasant but survivable if results are good. Ego is an instrumentation problem. It doesn’t just bruise people nearby. It bends each reading the organization sends upward until the dashboard is a mirror. Humility isn’t a virtue you perform at the offsite. It’s the maintenance that keeps your instruments honest.

The Most Expensive Opinion in the Room

The most senior opinion in any meeting costs more than the others because everything downstream adjusts to it. Speak first, and every later estimate becomes a negotiation with yours. Authority alone does that much; anchoring doesn’t check for humility first. Ego compounds this by raising the price of disagreement. Signal the conclusion you prefer and the analysis arrives pre-shaped. React badly to bad news once, and the next bad news comes late, laundered, or attached to a resignation.

None of this requires malice or cowardice. It only requires ordinary people to run the ordinary calculation: contradicting the person who writes their review can wait for a better day. From the inside, none of it feels like distortion. It feels like maybe I’m being pessimistic, like they probably know something I don’t, like this isn’t the meeting to derail. The channel corrupts one reasonable-sounding thought at a time.

The quality of what you’re told degrades in proportion to how costly you make honesty. Flinch once and entire categories of truth go quiet, and you never learn it, because the silence is the one piece of news nobody will deliver. From your seat, the degradation is invisible.

The dashboard still fills with numbers. They’re just yours, reflected back: optimistic estimates, filtered status reports, green reviews two weeks before a red outage, retrospectives sanded smooth. Not separate pathologies. One distortion, read on different instruments. The tell is direction: when numbers run consistently wrong the way you implied you wanted them, that isn’t estimation error. It’s instrument bias.

That’s the real cost of ego. Not morale, though it takes that too. Fidelity. You make live decisions on laundered data while the organization watches and adjusts.

Where Humility Is a Procedure

The military ran into this information problem earlier and harder, and attacked it the way it attacks most things: with a format, not a feeling. In an after-action review, junior participants speak first before anyone senior can anchor the room. Years in Airborne and Special Forces units taught me the AAR isn’t courtesy, and rank never fully leaves any room. The sequence doesn’t need it to. It just gets the cheapest opinions on the record before the most expensive one speaks. Units that debrief honestly stay alive, so the format assumes distortion and builds the order of speech against it. Nobody is asked to become humble. The agenda does it for them.

Engineering organizations face the same distortion with lower stakes and worse discipline. We hold the retrospective, then let the most senior person open it.

Three Repairs, No Personality Transplant

Speak last, on purpose. In estimates, design reviews, and incident calls, the senior-most voice goes last. The room knows it’s a rule rather than a mood. Your opinion doesn’t get less influential. It gets unanchored data to be influential about.

Juniors report first. Import the AAR sequence: the person closest to the work opens, facts before conclusions, and seniority weighs in reverse order. The closest reading is also the one most easily overwritten, so it goes on the record before anything can overwrite it.

Pay publicly for corrections. Reverse a decision out loud, cite the data, and name the engineer who surfaced it. In practice, that’s one message in the channel where you announced the original call: reverting the rollout plan; her capacity numbers were right; mine weren’t. Channels reopen when the organization sees truth rewarded at your expense. The first time you thank someone by name for proving you wrong, you buy back more signal than a quarter’s worth of skip-levels.

To be clear about what this isn’t: consensus leadership. Uncorrupted inputs don’t dilute the call. Someone still decides, alone if necessary, and the decision can still overrule the room. Humble about information, decisive about action. The sequence protects the first so the second is worth something.

The Instruments Are Talking

Give the director back that meeting. The engineers speak first, and four months go on the record. The director still gets the final call and can still choose six weeks and knowingly eat the risk, which is a different act from never hearing the risk at all. Maybe the project still ships in five months. It ships against an honest baseline, and the retrospective studies reality instead of a wish.

Your organization already knows what’s about to break. It knew in the room, in real time, the way that room knew. The only variable is what it costs them to say it to you.

Lower the price. Then make the call.