The moment a second person has to act on what the first one knows, context stops being personal. It becomes a team asset. Treating it as one person’s notes was always the workaround, not the design.
This is not a motivational claim about collaboration. It is a structural one, and it has an uncomfortable corollary: shared understanding is not a property of proximity. Two people in the same room, on the same team, inside the same company, holding the same document, do not automatically see the same thing.
Sharing is achieved, and it fails quietly
Grounding — the turn-by-turn process through which people establish common ground — has a specific and well-documented failure mode. Both parties can believe they have converged while their interpretations have quietly diverged. It is documented most sharply in asymmetric dialogue, where a third party who hears every word still arrives at a systematically weaker interpretation, because they were not party to the acts of grounding that produced the understanding.
Which is the honest description of what a meeting recording gives someone who wasn’t there. Every word, none of the grounding.
Heath and Staudenmayer named the organizational version of this coordination neglect: managers hold a working theory of organizing that is very good at dividing labor into specialized parts and reliably bad at anticipating how those parts have to be reconnected. It is made worse by a curse-of-knowledge effect — a specialist who has fully internalized a frame loses the ability to model what a non-specialist does not know. The person best placed to hand off context is the person least able to see what needs handing off.
Custody is the expensive default
Argote and Ingram put the stakes in competitive terms. A firm’s advantage comes from transferring knowledge internally faster than it leaks outward — and knowledge held only in people’s heads is the easiest kind to move quickly and the easiest kind to lose outright, to turnover, to attrition, to a team simply forgetting what it once knew. Knowledge embedded jointly in tools, routines and shared systems is stickier, harder to lose, and considerably harder for anyone else to copy.
That is an architectural argument, not a cultural one. Building context into infrastructure rather than leaving it dependent on a particular person staying in the room is not a nice-to-have. It is the difference between an advantage that compounds and one that walks out the door.
The field evidence resists the usual objection. Inside a Fortune 500 telecommunications firm, external knowledge sharing predicted group performance — rated independently by senior executives, not self-reported — and the relationship strengthened the more structurally diverse the group was. Ordinary demographic diversity showed no comparable effect: it is structural and informational difference, not identity, that makes deliberate sharing valuable.
And yet, inside that same organization, where sharing demonstrably paid, most members still didn’t do it by default. The incentive existing is not the same as the behavior existing.
Which makes the object of the work clear
If context is shared by definition, then a product that gives one person a better private memory has not addressed the problem. It has optimized the failure mode. The unit of value is not what one person can recall; it is what the next person can act on without a conversation.
Why individual adoption is still the way in
None of this argues that a single person should get no value until their whole team arrives. That would be a product nobody could ever start using, and the history of tools that spread inside organizations runs the other way — one person adopts something for their own benefit, and the shared value arrives as the second and third person follow.
The distinction is what the first person is actually holding. If the tool gives them a private memory, each additional user gets an additional private memory and the team ends up with parallel custody. If it gives them context that is already structured to be handed on, the second user inherits something rather than starting over. Same individual entry point; different asymptote.
That distinction sets a hard test. Not whether the record is complete, and not whether an individual finds the tool useful — but whether the second person, walking into the next conversation cold, arrives with what the first person had.
Everything else is custody wearing better clothes.
Sources
Clark & Brennan (1991) on grounding and its silent-divergence failure mode.
Heath, C., & Staudenmayer, N. (2000). Coordination neglect. Research in Organizational Behavior.
Argote, L., & Ingram, P. (2000). Knowledge transfer: a basis for competitive advantage in firms.
Cummings, J. N. (2004). Work groups, structural diversity, and knowledge sharing in a global organization. Management Science, 50(3) — 182 real work groups in a Fortune 500 telecommunications firm; 39% of members shared knowledge externally only rarely or less.