Watts-Strogatz Model
Rewiring 0.1-10% of connections creates small-world properties—high local clustering (team effectiveness) plus collapsed path lengths (fast cross-org communication). Milgram's 6 degrees explains why.
In 1967, Stanley Milgram asked people in Nebraska and Kansas to get a letter to a stockbroker in Boston by passing it through friends. Of the 64 chains completed, the median was 5 hops, the mean about 6—the origin of 'six degrees of separation.' In 1998, Duncan Watts and Steven Strogatz published a mathematical model in Nature explaining how this works. They showed that the neural network of the worm C. elegans (282 neurons), the western US power grid (4,941 nodes), and the collaboration graph of film actors (225,226 actors) all share the same structure: high local clustering (your friends know each other) plus a few random long-range connections (shortcuts). Start with a regular network where everyone connects to their neighbors. Rewire just 0.1-10% of connections randomly. The result: path lengths collapse dramatically while local clustering remains high. Google's 20% time, rotation programs between offices, and TGIF all-hands function as organizational shortcuts. For organizations, this means you don't need to connect everyone to everyone. Keep teams locally cohesive, add a small number of bridging roles, and information flow transforms—without the chaos of full mesh connectivity.
When to Use Watts-Strogatz Model
Use when organizational communication feels slow despite team-level effectiveness. Apply when silos prevent information flow across departments. Deploy when designing cross-functional roles, liaison positions, or community of practice structures. The model provides mathematical precision: you need only 1-10% bridging connections to achieve small-world properties.
How to Apply
Start with Regular Structure
Begin with organized local connections—teams, departments, hierarchies where people primarily connect to neighbors. This creates high clustering (teammates know each other) but long path lengths (getting information across the org takes many hops). Regular structure is the default in most organizations.
Questions to Ask
- Do team members primarily communicate within their team?
- How many hops does information take to cross the organization?
- Are there natural clusters (teams, functions, locations)?
- Is local cohesion strong but cross-org communication weak?
Outputs
- High clustering coefficient
- Long average path lengths
- Strong local relationships
Add Random Shortcuts
Create a small number of long-range connections that bridge otherwise distant clusters. Only 1-10% of connections need to be shortcuts—this is the key insight. Cross-functional roles, skip-level meetings, company-wide Slack channels, rotation programs, and communities of practice all function as shortcuts in the organizational network.
Questions to Ask
- What bridging roles exist (or could exist) across silos?
- Are there skip-level communication channels?
- Do rotation programs move people across clusters?
- What company-wide forums enable direct cross-org connection?
Outputs
- Dramatically reduced path lengths
- Preserved local clustering
- Bridging connections documented
Achieve Small-World Properties
The result balances segregation (specialized teams function independently, preserving expertise depth) with integration (information flows quickly across the organization, enabling coordination). Milgram's experiment showed 6.2 steps in human social networks; well-designed organizations can achieve similar properties through deliberate network architecture.
Questions to Ask
- Can any employee reach any other in 3-4 hops?
- Do teams retain autonomy and expertise despite global connectivity?
- Is information flow fast without requiring everyone to talk to everyone?
- Are shortcuts being actively maintained or have they atrophied?
Outputs
- Small-world network topology
- Balance of local autonomy and global coordination