← Amplification
Amplification · career

Next Role

Staff and product engineering craft, one real source at a time, with an exercise and the interview question it turns into.

1 issue · 1 piece · Nov 15
November 2026
Sunday Full issue →

Planning around open decisions, not dates

Staff track · Will Larson, lethain.com

The skill is seeing what actually limits a project and scheduling that instead. Larson's claim is that most projects are not constrained by execution time but by missing decisions, approvals and handoffs, so reshuffling the roadmap to free calendar time is a distraction; the thing to accelerate is "making high-quality decisions." You still set one slightly conservative external date, because "dates are the external interface for coordinating across companies," and then manage the work internally as a list of open decisions to close. His example: a credit card application project whose real work was about two weeks was planned for two months, because it carried "a dozen ambiguous decisions" and the plan assumed the team's existing decision pace. Most planning advice tries to estimate effort better; this says effort was never the bottleneck. His other example is passkeys, which never got onto the roadmap: he built it behind a disabled feature flag, used that for review, and it shipped.

PRACTICE IT Take Sunny or the Artifact Uprising rebuild and write a one-page decision log: every decision that blocked shipping, who owned it, how many days it stayed open, and what closed it (a prototype, an escalation, a default someone picked). Circle the one that cost the most calendar time and write two sentences on how you would force it in week one now. Keep the page; it is your answer to the question below.
SAY IT IN AN INTERVIEW "Tell me about a project that slipped. What would you do differently?" A strong answer names the open decisions rather than a bad estimate, shows how you surfaced and closed them (a prototype behind a flag, an escalation with a recommendation attached), and gives the before-and-after timeline in numbers.