Why Do Teams Duplicate Work That Sharing Could Prevent?

A short summary of Hattori, K. (2026) “The Sharing Trap: Duplicate Work and Coordination in Teams”.

Background

Anyone who works in a team has had the experience of discovering that a teammate has been doing the same work, without either of them knowing. The remedy is simple: say which task you intend to take on. Shared workspaces, project-management systems, and Slack are all available, and a single message reaches everyone. Yet sharing is still neglected, and duplicate work does not go away.

What this paper shows is that this is a problem of incentives rather than of communication technology, and, moreover, that the incentive to share can become weaker precisely in the kind of work where duplication is most likely and most damaging.

What the paper does

Consider a team of two members. Each member first decides whether to share her work plan, the task assignment she intends to carry out, with her teammate. Sharing is costly. Then, after observing what has been shared, each member chooses her productive effort.

If you do not disclose your work plan, part of your teammate’s effort is directed toward tasks that you also intend to perform, and that part produces no distinct contribution. The same holds in the other direction if your teammate withholds her plan. The fraction of effort lost this way is called duplication risk. Team output is the sum of the two effective contributions plus a complementary surplus generated by combining them. Each member receives the return from her own effective contribution and one half of the complementary surplus.

The key point is where the benefit of sharing goes. Sharing your own work plan does not raise your own efficiency at all. It raises only your teammate’s. What comes back to the sharer is only the indirect return that flows through the complementary term once the teammate can work without overlap.

What it finds

First, information sharing is a strategic complement. You are more willing to share once your teammate has already shared, because having received her plan you have a healthy productive base of your own, and the return from improving her contribution is correspondingly large. As a result, under the same technology and the same preferences, the model sustains both an equilibrium in which everyone shares and everyone works hard and one in which no one shares and effort is low everywhere. No special circumstance is needed for an organization to end up in the bad one.

Second, and this is the core of the paper, comes the duplication paradox. The incentive to initiate sharing is single-peaked in duplication risk. When duplication is limited, sharing prevents little wasted work. When duplication becomes severe, a member who has received no plan from her teammate converts effort into very little effective output of her own. Even if she discloses her plan and her teammate responds by working harder, little flows back to her through the complementary term. This is the productive-base effect. Meanwhile, the value of coordinated sharing to the team rises monotonically with duplication risk, which is the duplication-loss effect. In other words, in the organizations where duplication is worst, sharing is most valuable and yet the first step is hardest to take. Observed duplicate work is not necessarily evidence that people place little value on information. It may instead be evidence that the organization is trapped in a low-sharing equilibrium.

Third, the problem can be solved by leadership. The leader here has no private information and no formal authority. She is defined only by the fact that she shares first. Sharing first changes the follower’s problem from whether to initiate sharing to whether to reciprocate an observable disclosure. Since reciprocating is the better option, the high-sharing equilibrium is uniquely selected in the region where both equilibria exist. The leader herself also shares more readily, because she understands that disclosing first makes disclosure worthwhile for her teammate as well.

Fourth, the story changes as the team grows. A larger team has more to gain from complete sharing, because each additional nonsharer creates another potential source of duplicate work. But one first mover may no longer be enough. There is a critical mass: the smallest initial group of sharers that induces every remaining member to join. It depends on duplication risk, the strength of output complementarity, the sharing cost, and team size. Larger teams thus have more to gain from sharing, yet face a more demanding coordination problem in getting there.

Why it matters

First, installing a platform such as Slack and having it used are separate problems. Even with a channel through which a single message reaches everyone, the low-sharing equilibrium survives as long as the person who bears the cost is not the person who receives the direct benefit. Health information exchange offers an illustration: using it substantially reduces repeated diagnostic work, yet actual use is low, and what predicts use is less the availability of the technology than the readiness of partner organizations and support from senior leadership.

Second, the intervention should target order and numbers rather than technology. A project leader can disclose the first work plan, or make the first task allocation visible before asking anyone else to do the same. That alone makes sharing the privately attractive choice for everyone who follows. In a large team, preparing a coordinated group of initial sharers is more reliable than designating a single flag-bearer. Keeping operational units small therefore matters for more than reducing the number of communication lines: it also reduces the size of the coalition needed to escape the low-sharing equilibrium.

Third, lowering the sharing cost works directly. Templates that make writing out a work plan routine, or standing meetings at which plans are disclosed as a matter of course, are examples. The asymmetry of incentives cannot be removed, but the threshold can be lowered.

In closing

This theoretical exercise shows that “whether information can be shared” and “whether information is shared” are entirely different questions. The available technology settles only the first. The second is settled by strategic complementarity: as duplication becomes more severe, the value of sharing grows while the return to whoever moves first shrinks. Information sharing becomes self-sustaining only after enough others have made it worthwhile.

This research was supported by JSPS KAKENHI Grant Number 24K04912.