Loading...
Mira leads product development at a growing startup. When the team was just 4 engineers, their Monday planning meetings took 20 minutes. Everyone spoke, decisions happened fast, and they shipped features weekly. Then the company raised a Series A and hired aggressively. By month three, the engineering team hit 8 people. Mira noticed meetings now ran 55 minutes — not double, but nearly triple. Side conversations broke out. People repeated points others had already made. She calculated: 4 people meant 6 possible conversation pairs. 8 people meant 28. The communication channels hadn't doubled — they'd nearly quintupled. By month six, the team reached 12. Meetings ballooned to 90 minutes. With 66 possible communication pairs, nobody could track who knew what. Mira added a 13th engineer, Kai...
Popular framing: Big teams move slow because people get lazy or don't communicate well.
Structural analysis: Communication pairs scale as n(n-1)/2, so doubling headcount roughly quintuples coordination load. Past a threshold, the coordination tax exceeds each new hire's output and a single coordinator becomes the bottleneck — the same people reorganized at the right scale stop being gridlocked because the channel count drops back into the productive range.
Treating the symptom (long meetings, slow output) as a cultural or process failure leads to interventions — better facilitation, more documentation, targeted hires — that don't address the underlying combinatorial explosion. The real intervention is architectural: decompose the team into smaller autonomous units with minimal synchronization dependencies, reducing the communication graph from O(n²) to O(n).
