In March 2020, NovaBuild — a 90-person construction-tech startup in Austin — went fully remote overnight. For six months, everything worked. Engineers shipped features, sales hit targets, and CEO Mira celebrated their 'remote-first culture' in a viral LinkedIn post. Then the company tripled. By January 2021, NovaBuild had 280 employees across 14 time zones. Mira noticed the first cracks in Q1. The average Slack workspace had 340 channels. Engineers in Lisbon duplicated a payments feature that the Toronto team had shipped three weeks earlier — nobody knew. A customer-facing bug sat in a queue for nine days because three teams each assumed another owned it. The real crisis was invisible work. Engineering manager Leo ran a survey: 62% of individual contributors felt their contributions wen...
Popular framing: Big remote teams fail because remote work doesn't work.
Structural analysis: Past roughly 150 people the informal trust network that lets work flow without managers exceeds human cognitive capacity for relationships. Coordination cost scales with the square of connections; without hallway visibility, managers fall back on weak proxies (Jira, Slack) that the principal-agent gap makes worse. The breakdown is a Dunbar-number ceiling, not a remote-work problem.
Attributing the failure to remote work misdiagnoses the cause and produces the wrong remedy — return-to-office mandates. The actual remedy is coordination architecture: explicit ownership registries, public work attribution, async documentation norms, and performance systems that measure value over activity. Without the structural diagnosis, organizations repeat the failure in hybrid or in-office settings at the same scale.