If you thought the speed of writing code was your problem, you have bigger problems - Andrew Murphy

About this video

• Video Title: If you thought the speed of writing code was your problem, you have bigger problems
• Channel: NDC Conferences
• Speakers: Andrew Murphy
• Duration: 00:48:45

Overview

Andrew Murphy's talk argues that the speed of writing code is rarely the primary bottleneck in software development. Drawing on the theory of constraints, he explains that focusing solely on coding speed with tools like AI, without addressing other constraints in the development process, can actually worsen outcomes. He encourages a shift in focus from lines of code to value delivery, cycle time, and identifying and optimizing the true bottlenecks within the software development lifecycle.

Key takeaways

• Coding is often not the bottleneck: While AI can significantly increase code writing speed, it's typically not the limiting factor in delivering value. The majority of an engineer's time (84%) is spent outside of actively coding.
• Theory of Constraints (TOC): Every system has one bottleneck that determines its overall throughput. Optimizing non-bottleneck areas does not improve the system and can even make it worse by increasing work-in-progress and confusion.
• AI's Impact on Development: AI increases code output and churn, but can lead to reduced delivery throughput and stability if not managed. This is because the bottleneck shifts to other areas like discovery, review, or deployment.
• Shifting Focus to Value: The true value of code lies in its utilization in production. The focus should be on delivering value to customers, not just on the volume of code produced.
• Optimizing the True Bottleneck: Instead of solely optimizing coding speed, engineers should identify the actual bottleneck in their workflow (e.g., discovery, review, deployment processes) and use their spare capacity (gained from AI or process improvements) to address it. This includes improving tooling, processes, and potentially re-evaluating team structures and responsibilities.
• Measuring What Matters: Productivity should be measured by metrics like cycle time, wait time, and deployment frequency, rather than lines of code or the number of PRs. Ultimately, delivering value to customers is the key indicator of success.