Thoughts on delivery, product teams, and building software that ships on time.
Data provenance, workflow reconstruction, statistical methods, corrections, and limitations behind the finish-date analysis.
Almost nobody found Spotto's travel-time search. Almost everyone who typed in a place finished it. Why that meant building more of it, not cutting it.
A software estimate is not a finish date. Public data from three projects shows why deadlines need probabilities, not promises.
Being the bottleneck feels exactly like being indispensable. A 5-minute audit to find out which one you are.