A veteran staff engineer argues that accurately estimating software projects is fundamentally impossible. Most engineering work involves unknown research and discovery, making precise time predictions futile. Teams often resort to t-shirt sizing or arbitrary heuristics, which management then converts into concrete deadlines for business planning.

This practice persists because estimation serves a political function rather than a technical one. Estimates help non-engineers decide which projects get funded or cancelled. They define the scope of work itself—teams will design solutions to fit a given timeline, not the other way around, fundamentally altering the technical approach.

The author's method involves gathering political context before writing code. Instead of asking "how long will this take," they determine which approaches fit within the expected timeline. This acknowledges estimation's true purpose: a tool for organizational decision-making, not an engineering discipline. The tension between this reality and the industry's polite fiction drives much dysfunction.