Every Organization Has an Otter
Why organizations rarely know who they’re standing on
In the early 1960s, ecologist Robert Paine picked a short stretch of rocky shoreline on the Washington coast, at a place called Mukkaw Bay, and started removing every ochre sea star he could find from it. Not once. Every couple of weeks, by hand, so none could return.
Within months, mussels, the sea stars’ usual prey, took over the cleared rock. They crowded out barnacles, snails, algae, everything else that had been sharing that patch of shoreline. A stretch that had supported around fifteen species collapsed down to eight. One animal, never the most numerous thing in that ecosystem, had been holding the whole community together simply by eating enough of what would otherwise take over everything else.
Paine called it a keystone species. His 1966 paper is still the most-cited empirical article the journal that published it has ever run.
It was never about size
A few years later and a few thousand miles north, biologist James Estes found a version of the same pattern in the sea otters of the Aleutian Islands. Otters eat sea urchins. Where otters survived, urchin populations stayed small, and kelp forests grew thick, sheltering fish, invertebrates, an entire underwater community. Where otters had been hunted out by the maritime fur trade over a century earlier, urchins had multiplied unchecked and grazed the kelp down to bare rock. Ecologists call the result an urchin barren, and the name is accurate. Almost nothing else lives there.
The otter was never the dominant species by numbers either. Its impact on that ecosystem had nothing to do with how much space it occupied.
The term itself is still argued over inside ecology. Some researchers think “keystone species” has been stretched to cover too many species since Paine coined it, applied more as a compliment than a precise category. I don’t think that argument undermines what both of these cases actually showed. A small removal triggering a disproportionate collapse is real, whatever we end up calling it.
Two ecosystems, two scientists working decades apart, and neither pointed to the biggest or most numerous species in the system. That’s worth sitting with, but it isn’t the part that surprised me most.
Here’s the part I hadn’t appreciated
Paine didn’t identify the sea star as a keystone species by watching it. He identified it by removing it and recording what fell apart afterward.
Estes’ case worked the same way, just accidentally. Nobody set out to test what otters were doing for the ecosystem. The fur trade removed them first, across a huge stretch of coastline, and the kelp forests answered the question a century later, once ecologists finally thought to compare the coastlines where otters had survived against the coastlines where they hadn’t.
Both times, the removal came first.
The understanding came after.
Organizations don’t get to run that experiment on purpose. You can’t remove a person from a team to find out whether the team collapses, then put them back once you have your answer.
That means most of what we use to identify indispensable people, network analysis, informal surveys, asking around for who everyone goes to, isn’t an answer. It’s an educated guess.
The same goes for the instinct to look among the most visible people first. The highest performers, the biggest titles, the loudest presence in the room. That instinct feels like a shortcut around the removal problem, but it isn’t one. It’s just another form of watching, and watching was never how either of these cases actually got confirmed.
A high performer is not automatically a keystone.
Performance describes how well someone does their job while they’re in it.
A keystone is defined by something else entirely: what breaks, and where, once they’re gone.
What changed once I saw this
I stopped expecting to identify keystone people with certainty while they’re still there. I don’t think that certainty is available, not really, not from inside the system.
What I started paying attention to instead is smaller and less satisfying: who gets asked the same question by three different people in the same week. Whose absence from a single meeting quietly stalls three unrelated things. Who nobody remembers to loop in, but everyone eventually routes back to anyway.
None of that is proof. It’s the closest thing to a signal you get without running the experiment nature runs, which is removal.
Where the analogy breaks down
Ecologists can remove a sea star from a plot of shoreline and put it back. Paine did, repeatedly, for years, as a controlled experiment.
Nobody can ethically do that to a person. Organizational diagnosis is stuck with an asymmetry ecology doesn’t have: the one method that produces a confirmed answer is the one method that’s never actually available.
The uncomfortable part is that we rarely know who they are until they’re gone.
Field Questions
- If this person left tomorrow, what would I expect to break, and how confident am I in that answer?
- Am I identifying this person’s importance by their title, or by what quietly routes through them?
- Who gets asked the same question by multiple people in the same week?
- What am I doing right now to reduce how much the system depends on any one person staying?