COUPLING · AND WHAT IT COSTS
On adjusting to the relationship rather than to your own part — and on the research that is oversold.
Anybody who has been in a group doing something demanding together, well, has a report about it that they cannot quite finish.
It comes out as a boundary going soft. They stop tracking their own contribution separately. There is a sense of the thing being done by the group rather than by several people at once, and it is usually described as the best hour of the season and as slightly embarrassing to talk about.
The reports are consistent across activities with nothing else in common, and they share one structural feature: attention moved from a person’s own part to the relationship between their part and everyone else’s.
A group can fail on either side of this, and the second failure is the one nobody warns about.
Everyone optimizing their own contribution, correctly, in isolation. It produces a group of good individual performances that do not add up, and everybody involved can defend their own part.
The other ditch. Difference gets suppressed in the name of togetherness, dissent becomes disloyalty, and the group converges on whatever the confident people already thought. Merging is not automatically good.
That line is mine, because it is a description of what people report rather than a doctrine anybody taught.
The practical content is a change in what you are attending to. Not your contribution, monitored and optimized — the gap between yours and the next person’s, which is a different object and is outside you.
Saṅgha is the Buddhist term for a community of practitioners, and in early usage it is a practical arrangement rather than a mystical one — people who make the practice possible for each other.
The Confucian material makes a different claim entirely: that harmony is not sameness, and that the exemplary person harmonizes without conforming. That is an argument for preserved difference inside a functioning group.
These are not the same idea and neither is about the state this essay describes. I am using the first for the structure and the second as the correction to it, which is a use rather than an exposition.
Interpersonal synchrony research finds that people moving in time together subsequently report more liking and more willingness to cooperate, and the basic effect has replicated reasonably well.4
Two honest qualifications. The effects are modest, and the popular version — that synchrony produces bonding through oxytocin — runs well ahead of the evidence; the oxytocin literature in humans has had substantial replication problems and should not be leaned on.5
The collective-intelligence work has a similar shape: a general factor of group performance was reported and has not reproduced cleanly across attempts.6
What survives is narrow and useful: groups attending to coordination outperform groups attending to individual contribution on tasks that require coordination. Which is nearly a tautology, and it is the part you can rely on.
The objection is that this essay is describing a mechanism with a well-documented dark side.
The case against
High cohesion, strong shared identity, attention on the relationship rather than on the task — that is the profile of the merged group and it is also the profile of the group that stops being able to hear a dissenting member.
The cost is borne by whoever notices the problem first, and the mechanism that makes the group feel good is the same one that makes their objection feel like a betrayal.
The guard is the Confucian correction: harmony is not sameness. A group where merging has eliminated disagreement has not achieved this; it has lost something and gained a pleasant feeling. The test is whether the least senior person can still say the inconvenient thing.
Attend to the gap rather than to your part. The relationship between what you are doing and what the person next to you is doing. That is the whole mechanical change.
Make one thing genuinely shared and specific — a timing, a rhythm, a handover point. Vague togetherness produces nothing; a shared object produces coordination.
Protect the dissent explicitly. Name in advance who is expected to raise problems, and make it somebody junior enough that the naming matters.
And check the test regularly: can the least senior person still say the inconvenient thing? If not, whatever the group has, it is not this.
The first band is where most groups discover something uncomfortable.
The reports are consistent and the mechanical core is small: attention moves from your own part to the relationship between the parts.
The research is real and considerably more modest than its popular version, and none of the neurochemical story should be leaned on.
And the same conditions that produce the good version produce groupthink. Harmony is not sameness, and the test of which one you have is whether the least senior person can still say the awkward thing.
Attend to the gap between your part and the next person’s, rather than to your own part. Make one shared object specific — a timing, a handover, a rhythm.
Then run the test: can the least senior person still say the inconvenient thing? If not, what the group has is not this.
Terms are given in the language of the tradition they come from and are not treated as synonyms. Modern research is cited by author and publication.