Loading...
In January, NovaCorp's CTO Kai greenlights a customer portal rebuild. The engineering lead Mira estimates 12 weeks — she's done similar projects before. The VP of Product, Leo, adds a 4-week buffer 'just in case,' setting the deadline for mid-May. By March, the team has burned through 8 weeks but completed only 30% of the work. Mira notices engineers spending days perfecting login animations and debating database schemas that won't matter until Phase 2. With 16 weeks of runway, nobody feels urgency — meetings multiply, code reviews become philosophical debates, and a 2-day API integration stretches to three weeks of 'doing it right.' In April, Leo calls an all-hands: they're behind. He revises the estimate to August, confident this time because he's 'accounting for the delays we've seen...
Popular framing: The project failed because people made bad estimates and managers lacked the courage to intervene early. Better planning tools, more experienced leads, and harder deadlines would have prevented it.
Structural analysis: The project was structurally guaranteed to expand. Parkinson's Law ensured work filled the generous runway; Hofstadter's Law ensured every correction underestimated remaining complexity; the planning fallacy ensured each revised estimate anchored on prior estimates rather than empirical velocity. No individual decision was irrational given local incentives — the system produced the outcome without anyone intending it. The 'planning fallacy' is correct but misses the 'Goodhart's Law' aspect — the buffer itself is what killed the estimation.
Blaming estimation or leadership preserves the belief that better judgment next time will fix it, which protects the organization from confronting the incentive structure that rewards confident deadlines and punishes uncertainty. The competitor's success was not a talent advantage — it was a constraint advantage: no runway meant no Parkinson's expansion. Until NovaCorp changes the structural conditions, the same pattern will recur on the next project.
