An overview of Rod Beckstrom’s The Starfish and the Spider on enhancing opportunity and mitigating risk in the age of AI as he joins us to discuss The Starfish AI Model at OODAcon 2026.

Rod Beckstrom’s OODAcon 2026 session, The Starfish AI Model, offers an opportunity to revisit the decentralization framework he developed with Ori Brafman in The Starfish and the Spider. Its central strategic question is directly relevant to AI: how can organizations distribute initiative while retaining the coordination needed to manage shared risks?

Executive Summary

Who should have the authority to act when intelligence becomes widely available? For leaders assessing AI, this question reaches beyond model performance into organizational design: where decisions happen, how knowledge travels, and what holds an enterprise together when more people (and software agents) can operate independently.

Published in 2006, The Starfish and the Spider: The Unstoppable Power of Leaderless Organizations contrasts centralized organizations with networks whose capabilities and initiative are distributed. It also explores hybrids that combine these characteristics.

An AI deployment is also an authority decision to:

• Give a team access to an assistant
• Allow an agent to retrieve internal information, and
• Authorize software to modify a production system to create different organizational arrangements.

The starfish lens helps leaders examine those arrangements. A system may distribute work while retaining centralized permissions. A company may encourage local experimentation while depending on one shared provider. A network may contain many participants yet remain vulnerable to a single identity service or data source. For OODA decision-makers, the practical issue is whether distributed capability produces faster learning and better decisions – or simply more activity that the organization cannot adequately understand.

From the Spider to the Starfish

The spider represents an organization with a recognizable center of control. The starfish represents a network that can organize around a shared purpose and reproducible connections. The metaphor draws attention to the consequences of removing a leader or disabling part of a network. This is a strategic analogy, rather than a universal biological or organizational law. Centralized institutions can have succession plans and redundancy. Decentralized networks can contain hidden concentrations of power.

The useful diagnostic is therefore specific: Which functions can continue, adapt, or recover when a critical participant disappears?

The Five Legs of Decentralization

The book’s five supporting elements, summarized in Decentralisation: The Starfish and the Spider, explain how autonomous participation can become collective action.

Component	Meaning in the original framework
Circles	Autonomous groups of peers that coordinate through relationships and shared norms.
Catalyst	A person who initiates connections and enables a group to function without continued dependence on that person.
Ideology	The shared beliefs or purpose that motivate participation and sustain cohesion.
Pre-existing network	Relationships and social infrastructure through which a new effort can spread.
Champion	An energetic advocate who advances the cause and recruits participation.

The catalyst and champion perform different functions: one helps people connect and organize; the other supplies sustained advocacy. These are roles, and need not always be held by different people.

Applying the Framework to AI

The following mapping offers questions for organizational design. It does not imply that software agents possess the trust, judgment, or shared commitments of human participants.

Book concept	Possible AI application	Question for leaders
Circles	Teams develop AI workflows around their own operational knowledge.	What can each team decide and execute independently?
Catalyst	An enabling team provides access, training, and reusable methods.	Can useful adoption continue without every decision returning to the enabling team?
Ideology	A shared mission guides choices about acceptable AI use.	How does that mission become observable behavior and enforceable limits?
Pre-existing network	Professional communities and trusted working relationships spread lessons.	Can successful practices travel with evidence of their limitations?
Champion	Practitioners demonstrate value and help colleagues adopt useful approaches.	Are advocates rewarded for verified results and candid failure reporting?

Consider a hypothetical enterprise with several regional teams. Each team uses AI to analyze local customer needs. A central group supplies approved access, evaluation methods, and security requirements. Teams share effective workflows across an internal community.

This arrangement creates room for local initiative. Its effectiveness would depend on whether teams can demonstrate improved outcomes, whether lessons transfer across contexts, and whether the organization can identify and contain a harmful workflow.

Key Points

• Distinguish participation from authority. Many users or agents do not necessarily mean that control is distributed.
• Examine dependencies. Apparently independent teams may share a common point of failure.
• Treat shared purpose as something to implement. Mission statements need supporting incentives, permissions, and review practices.
• Evaluate collective learning. Local experimentation becomes more valuable when others can inspect and reuse its results.
• Assign accountability explicitly. Every deployed workflow needs an owner who can explain its scope and respond when it fails.

What’s Next? Questions for the OODAcon 2026 Session

Beckstrom’s OODAcon 2026 session offers an opportunity to clarify how his AI-specific framework extends the original organizational argument. Attendees should listen for:

1. The unit of decentralization: Is the model concerned with people, organizations, AI agents, technical infrastructure, or their interaction?
2. The mechanism of opportunity: How does the proposed structure improve discovery, coordination, or value creation?
3. The mechanism of risk reduction: Which risks become smaller, and which move elsewhere?
4. The limits of autonomy: What decisions should remain centralized, and under what conditions?
5. The evidence of success: What observable outcomes distinguish a functioning model from an appealing metaphor?