How I estimate work (seangoedecke.com)

Article summary

The article discusses the challenges of estimating software project timelines and argues that accurate estimation is impossible due to the dominance of unknown work. The author suggests that estimates are often used as a political tool by managers to negotiate with each other, rather than as a genuine attempt to predict project duration. The author proposes an alternative approach to estimation, where the focus is on finding a set of software approaches that match a given estimate, rather than trying to accurately predict the time required for a specific project. This approach acknowledges the uncertainty and complexity of software development and seeks to manage expectations and risks.

Main themes

• Software estimation challenges
• Politics of estimation
• Uncertainty in software development
• Manager-engineer dynamics
• Estimation as a negotiation tool

What commenters say

• Estimates are often top-down and used to fill a predetermined timeline, rather than being based on a genuine assessment of the work required.
• Sales teams often pressure engineers to provide estimates in order to close deals, even if the estimates are unrealistic or uncertain.
• Some argue that estimates can be improved by following up and tracking actual project timelines, while others believe that this approach is still limited by the inherent uncertainty of software development.
• Providing a promised date for a feature can be necessary to keep a sales opportunity alive, even if the estimate is uncertain or potentially unrealistic.
• Others argue that it's better to be honest about the uncertainty of estimates and risk losing a sale, rather than providing a potentially misleading estimate.
• The ability to estimate project timelines can vary greatly between teams and engineers, and estimates should not be assumed to be fungible.
• Some commenters suggest that using a 'tracer-bullet' approach, where a small proof-of-concept is built to explore the tricky bits of a project, can help improve the accuracy of estimates.
• Others believe that estimation is an art rather than a science, and that it's more important to narrow down uncertainties over time rather than providing a single, precise estimate.