5  Module 3: Governance and Leadership Models (40 minutes)

Facilitator: Jake Chen, Functional Genomics (CM4AI)

5.1 Learning Objectives

By the end of this module, participants will be able to:

  1. Distinguish between traditional hierarchical, distributed, and network governance models and describe when each is appropriate.
  2. Map the decision-making layers in a large research consortium (working group → steering committee → program officials → funder).
  3. Identify where within a given governance structure they have meaningful agency to influence outcomes.
  4. Articulate the tradeoff between control (protecting project integrity) and freedom (enabling scientific creativity) in consortium governance.

5.2 Module Overview

Governance sounds like an administrative concern — the province of directors and program officers, not scientists. But every researcher in a large consortium is already operating inside a governance structure, whether or not they know it. That structure determines who can approve a data release, who needs to sign off before a paper is submitted, and who is empowered to resolve a conflict between two teams. Scientists who don’t understand the governance structure around them waste enormous time routing decisions to the wrong people, waiting on approvals that aren’t needed, or — worse — making unilateral choices that create downstream conflicts.

This module provides a practical map of how governance works in large research consortia, using Bridge2AI as a concrete example. It introduces three governance models across a spectrum from centralized to distributed, then grounds the theory in the real experience of the people who actually navigate these structures every day. The fishbowl format is chosen deliberately: it allows participants to observe authentic conversation about governance challenges, including the tensions and ambiguities that polished presentations would smooth over.

5.3 Participant Background Reading

Participants are encouraged to review the following before the session. Each takes 10–20 minutes.

  • A primer on how NIH Common Fund consortia are structured. Before the session, review how large NIH-funded consortia typically organize themselves — particularly the roles of steering committees, working groups, program officials, and External Scientific Panels. The NIH Common Fund website (commonfund.nih.gov) and published consortium papers (which typically include a governance section) are good sources. Participants who already work in Bridge2AI should review their specific consortium’s governance charter if available.

  • An accessible overview of distributed or shared leadership. The module references distributed leadership as an alternative to traditional hierarchy. Before arriving, read a short accessible piece — a blog post, magazine article, or book excerpt — that describes what distributed leadership looks like in practice. This could be from a science management context (e.g., an essay on running an open-source scientific project) or a general organizational context.

5.4 Instructor Notes

5.4.1 Conceptual Background

Three governance models and their tradeoffs. - Traditional hierarchy concentrates decision-making authority at the top. It is efficient when decisions are routine and well-defined, and when speed matters more than buy-in. It fails when the people at the top lack the specialized knowledge to make good decisions, or when the chain of command creates bottlenecks in fast-moving science. - Distributed leadership allocates authority by expertise rather than position. It requires high trust among team members and clear norms about who has authority over what. It is well-suited to interdisciplinary teams where no single person can have deep expertise in all areas. The risk is ambiguity: if authority boundaries aren’t explicit, decisions fall into gaps or get made twice by different people. - Network governance — hub-and-spoke or web-like structures — is common in large data consortia like Bridge2AI, where semi-autonomous subgroups (centers, cores, working groups) need to coordinate without a single unified command. Provan & Kenis (2008) identify three forms of network governance: participant-governed (all members share governance), lead-organization-governed (one member coordinates on behalf of others), and network administrative organization (a separate entity governs the whole). Many NIH consortia use hybrid forms.

Agency vs. awareness: a key pedagogical move. This module makes an important distinction between macro-structure (the consortium’s overall governance, which participants typically cannot change) and micro-structure (the governance of their specific working group or subteam, which they often can shape). This reframe is essential to prevent learned helplessness: participants should leave feeling empowered to act on what they can influence, not overwhelmed by what they cannot.

Facilitating the fishbowl. The fishbowl format is pedagogically powerful because it models authentic conversation rather than polished presentation. Key facilitation moves: - The empty chair is an explicit invitation for audience members to enter the conversation. Facilitators should name this at the start and gently encourage participation if no one volunteers. - Keep the conversation focused on how decisions get made, not just what decisions are made — the process, not the output, is what participants need to observe. - Draw out disagreements and tensions gently. “It sounds like you and [other panelist] have different views on this — can you both say more?” is a useful prompt. - Close by asking each panelist: “What do you wish you had known about this consortium’s governance when you first joined?”

5.4.2 Key Concepts

  • Governance: The systems, processes, and norms by which decisions are made and authority is allocated within an organization or collaboration.
  • Traditional hierarchy: A governance model with a clear, top-down chain of command and centralized decision-making authority.
  • Distributed leadership: A governance model in which authority is allocated by expertise rather than position, with multiple people holding decision-making power in their domains.
  • Network governance: A governance model suited to consortia and partnerships, in which semi-autonomous units coordinate through shared norms, liaison roles, or a coordinating body rather than a unified chain of command.
  • Working group: A subunit of a consortium responsible for a specific area of work (e.g., data standards, ethics, a specific use case). Typically has its own internal governance.
  • Steering committee: A consortium-level body with broad oversight and decision-making authority across working groups.
  • Agency vs. awareness: The distinction between understanding a governance structure (awareness) and having the ability to change it (agency). Participants may have limited agency over macro-structure but meaningful agency over micro-structure.

5.5 Content Block (10 minutes)

(Narrative Arc: A Maze, a Map, and the Guides)

5.5.1 Context: Navigating the Layers of Management (3 minutes)

  • Hook: In a single-lab environment, you usually have one boss (the PI) and a straight line to decision-making. In a massive consortium like Bridge2AI, that straight line becomes a maze.

  • Reality: We need to visualize the vertical depth of these projects. Decisions pass through multiple layers: from the working group to the steering committee, to the program officials, and finally to the funding agency (NIH).

Goal: Not everyone has to become an administrator. We are sharing this knowledge so scientists don’t waste weeks waiting for a decision from the wrong person. At all times, we need to know who holds the keys.

5.5.2 Theory: Overview of Models (5 minutes)

Models: Introduce the spectrum of governance, ranging from traditional to experimental.

  • Traditional Hierarchy: Clear chain of command (efficient, but can bottleneck).

  • Distributed Leadership: Authority spread by expertise (agile, but requires high trust).

  • Network Governance: Hub-and-spoke models, common in large data consortia.

Governance Models

Resource: For those who want to dig deeper into the sociology of these structures, we have included key references in your workshop booklet (e.g., Pearce & Conger on Shared Leadership).

5.5.3 Reality Check: Agency vs. Awareness (2 minutes)

A word of caution: We don’t always get to choose one of these models. When we join a massive project like Bridge2AI, the governance structure is likely already set in stone (maybe!).

Lesson: We cannot always negotiate the macro-structure (the Consortium), but we can often influence the micro-structure (our specific working group). Your goal is to understand the rules of the game so you can play it effectively, not necessarily to rewrite the rulebook.

5.6 Activity 3: Fishbowl Fireside Chat (30 minutes)

  • Format: To illustrate what this looks like in the real world, we are moving into a ‘Fishbowl’ format.

    • Setup: Chairs arranged in the center. One chair is left intentionally empty to encourage audience rotation/participation.

  • Moderator: Introduce Dr. Jake Chen, who will guide the conversation, ensuring we probe the tension between structure and innovation.

  • Keynote Lead: Introduce Dr. Casey Greene.

    • Why him? Casey represents a modern approach to scientific leadership—open, decentralized, and highly effective. We want to hear how he leads without stifling creativity.

  • Institutional Context: Introduce the Bridge2AI representatives (Pamela Foster, Colleen Cuddy, Yulia Levites Strekalova, and Jamie Toghranegar).

    • Role: They represent the connective tissue of the consortium—the people who ensure the disparate parts actually talk to one another.

Objective: We will watch them discuss how decisions actually get made. Pay attention to how they balance the need for control (to keep the project safe) with the need for freedom (to let the science happen).

Module Materials