Mount Everest: A Case Study in Deference & the Decision Nobody Stopped

Your organisation has a rule that everyone can recite and nobody has ever enforced. You know the one. It is quoted in onboarding, it appears on a slide once a year, and there is no moment on record when work stopped because of it.

That rule is not a control. It’s a shared belief about what a good version of your team would do. The difference between those two things is invisible on an ordinary day and total on the worst one.

The rule everybody knew

The two largest commercial expeditions on Everest in 1996 were led by men among the most experienced high-altitude climbers alive. Both talked constantly, and publicly, about the one decision that governs a summit day: the time you turn around whether or not you have reached the top.

One of them had a named rule. Six weeks before the climb, a journalist had written a story about it:

“If you aren’t on top by two, it’s time to turn around. Darkness is not your friend.”

The other lectured his clients on it. One recalled him “lectur[ing] us repeatedly about the importance of having a predetermined turnaround time on our summit day — in our case it would probably be 1:00 p.m. or 2:00 p.m. at the very latest — and abiding by it no matter how close we were to the top.”

Read that quote closely. Probably. Or. At the very latest. A lecture on the sanctity of hard deadlines, delivered without one.

That vagueness was not a general habit. Briefing the same clients on the same expedition, the same leader was perfectly capable of being absolute:

“I will tolerate no dissension up there. My word will be absolute law, beyond appeal. If you don’t like a particular decision I make, I’d be happy to discuss it with you afterward, but not while we’re up on the hill.”

The authority is specified to the letter. Only the threshold is left blank.

The sentence the case turns on

On the evening before the summit bid, both leaders stressed again that climbers falling behind schedule had to turn back. Then, from the same account:

At Base Camp before our summit bid, Hall had contemplated two possible turnaround times — either 1:00 p.m. or 2:00 p.m. He never declared which of these times we were to abide by, however — which was curious, considering how much he’d talked about the importance of designating a hard deadline and sticking to it no matter what. We were simply left with a vaguely articulated understanding that Hall would withhold making a final decision until summit day, after assessing the weather and other factors, and would then personally take responsibility for turning everyone around at the proper hour.

Deferring the number was not laziness. Conditions change, and a leader who can read the mountain should decide on the mountain. But it had a second effect nobody chose: it put the rule out of everyone’s reach but one man’s, at the altitude where that man would be least able to think clearly.

Here is what the undeclared rule looked like on the day itself.

Why nobody else called it

The other half of the case is who could have said something, and what it would have cost them.

By two o’clock most of one team’s clients had not reached the summit. That team’s third guide had serious reservations about anyone still climbing past midday. He did not raise them, and said why afterwards:

“I was definitely considered the third guide, so I tried not to be too pushy. As a consequence, I didn’t always speak up when I should have.”

The ranking was concrete, not atmospheric. His place in the chain of command sat below the leader and the senior guide, and his pay said so: he had agreed to guide Everest for $10,000, against the senior guide’s $25,000. A client’s place on the same mountain cost $65,000.

Beneath the guides the clients had no standing, and no expectation of any. One wrote that “we were a team in name only… we would ascend as individuals, linked to one another by neither rope nor any deep sense of loyalty. Each client was in it for himself or herself.”

A group with no mutual obligation does not produce a collective objection. It has no mechanism for one.

The asymmetry was also issued as equipment: on one team the leader and his two guides each carried a radio, and the eight clients carried none. The people with most at stake could not hear the decision being made, let alone contest it.

Both leaders knew the pull ran the other way — one said plainly that a client who sees the summit close and is dead set on it will laugh in your face and keep climbing. He knew the rule would need enforcing against motivated people, and still did not set the number that would have enforced it.

Why two failures are one

I had always filed “deference to authority” and “unclear requirements” as two different organisational problems, addressed by two different fixes — psychological safety on one side, better specifications on the other.

They are the same problem. An unspecified threshold is a delegation of authority to the most senior person present, made silently, at the moment the threshold matters. You can run every psychological-safety workshop in the catalogue; if the trigger condition is “when it seems too late,” a junior person raising it is still contradicting a superior’s judgment rather than reading out a number.

What I do differently

Nobody dies when a release of mine slips. That difference is not a footnote, and I want it said before I borrow anything from a mountain: this case is taught because its consequences were unmissable, which is the one advantage my version of the problem never has.

It is also why my version survives. On Everest the missing threshold killed people in a single afternoon, so it earned a decade of analysis. In my work the same defect costs nothing on almost every day — and a defect that costs nothing on almost every day is one nobody is ever assigned to fix. Low stakes do not make the flaw smaller. They make it permanent.

I write acceptance criteria, so here is mine. I have written it more than once:

The release will be held if quality is not acceptable.

Read it the way I read Hall’s. Will be held — by whom? Quality — measured against what? Not acceptable — in whose judgement, at what hour? Every slot where a number should sit has a person in it instead, and the sentence never says which person. So it resolves the way vagueness always resolves: upward. When the date is close and the pressure is on, the engineer who says “quality isn’t acceptable” is not reading out a threshold. They are making an accusation, and whoever outranks them decides whether it lands.

That is the same defect, in my own handwriting, with the stakes turned down far enough that I got to keep writing it.

So I now run three checks on any control that is supposed to stop work.

What I’d hold back from

It’s easy to read this case as a story about two men who should have known better, and I don’t think that’s the lesson — partly because they did know better and said so out loud for weeks, and partly because being certain of the right answer from a warm chair is not an achievement.

What makes it teachable is that the failure is structural and ordinary. A senior person reserved a judgment for themselves, reasonably. Nobody junior wanted to be pushy. The group had no obligation to each other. Every organisation you have worked in has that arrangement somewhere, and almost always it holds.

Thanks to an instructor who ran this as a discussion rather than a lecture, so the room had to argue about who could have stopped it — which is a much harder question than what went wrong.

If you have a rule at work that’s meant to halt things: when did it last halt anything, and who was allowed to say so?

Sources