How network clusters work: community detection and force simulation

When you look at a spreadsheet of 1,000 professional contacts, every name looks equal. In reality, networks are not flat lists; they are dense, overlapping communities held together by shared history, past companies, and mutual circles.

Sixgree groups your connections into clusters visually and mathematically. This guide explains how our clustering engine works, why force simulations reveal your natural groups, and how to read the clusters across your map.

What is a network cluster?

A cluster (or community) is a group of people who are connected more densely to one another than to the rest of the network.

In a typical career graph, you rarely have one homogeneous network. Instead, you have distinct islands:

• Alumni clusters: classmates, professors, and lab partners from university.
• Past company hubs: former coworkers, engineers, and managers from a previous employer.
• Industry circles: investors, advisors, founders, or specialists you met through conferences or sector meetups.

Sixgree detects these groups automatically, without requiring you to manually assign tags or organize folders.

Force simulation: turning relationships into physics

Rather than rendering your connections in rigid rows or arbitrary charts, Sixgree treats your network as a physical particle system:

1. Repulsion (The Coulomb Force): Every person acts like a charged particle pushing other dots away. This prevents clutter and spreads unconnected people apart.
2. Attraction (Spring Links): Mutual ties and shared circles act like springs pulling connected people toward each other.
3. Centering & Rings: Key connectors and high-power catalysts are drawn toward the core, while distant acquaintances naturally settle along outer orbits.

When the physics simulation runs, tightly interlinked groups pull inward toward a common center of gravity, naturally forming distinct visual clusters separated by empty space.

Community detection without cloud processing

Most social graph tools upload your address book to a server to run heavy graph algorithms. Sixgree runs all community detection and physics calculations locally on your machine inside your browser or native desktop app:

• Local SQLite adjacency: Mutual links are queried directly from your local SQLite database.
• Real-time relaxation: The force engine stabilizes in seconds on consumer hardware.
• Privacy preservation: Your clusters never leave your device. Nobody else sees how your network partitions into communities.

What clusters tell you about your reach

Reading your clusters gives you strategic clarity that a standard search bar cannot:

• Bridging connectors: Notice the individual dots suspended between two distinct clusters? Those are your rare bridge contacts—people who have feet in both worlds (e.g., an ex-colleague who joined a venture firm). They are often your highest-value warm intro paths.
• Isolated silos: Clusters with very few cross-links to the rest of your network represent untapped reach into new industries or regions.
• Network health: A balanced multi-cluster graph indicates a resilient career network, whereas a single giant clump suggests vulnerability to shifts in a single company or market.

You can tune link strength, repulsion, and cluster presets directly in the Galaxy's Physics Lab under the Filters panel.