Estimating software projects is notoriously difficult, often leading to frustration within tech teams. Many engineers find traditional time estimates inadequate, resorting instead to methods like t-shirt sizing, which inevitably get translated back into time by management. The author argues that this practice stems from an underlying tension between the need for estimates and the reality that most software work involves significant unknowns. While small, well-defined tasks can be estimated accurately, most projects are complex and require exploration, making precise predictions nearly impossible. Estimates often serve political purposes rather than practical ones. Engineering teams may feel pressure to adjust their estimates based on management demands, which skews the process. The author points out that estimates can define the work rather than the reverse, meaning teams often start with a time frame and then adjust their project scope to fit it. For instance, if given a short deadline, engineers might simplify their approach to meet expectations. To navigate this environment, the author emphasizes the importance of understanding the political context surrounding a project before diving into the code. By gathering insights on pressures and expectations, engineers can formulate more practical estimates. Instead of asking how long a task might take, they should consider which approaches can be completed within the desired timeframe. This pragmatic approach helps align engineering efforts with organizational goals while acknowledging the inherent uncertainties in software development.