Urgent.News

What's breaking now, across thousands of outlets.

Tech

Design for Understandability

Designing code for understandability is a crucial aspect of modern software development. In a world where large language models (LLMs) are generating much of our code, it has become increasingly important to ensure that the code we write is not only functional but also comprehensible. Google's Code Health Summit in November 2025 provided an opportunity to explore this topic further and share insights on designing code for understandability.

The fundamental question posed at the start of the talk was: if LLMs can write code, what role do human software engineers (SWEs) play? While LLMs can generate code that appears to follow established patterns, the ultimate responsibility for judging the maintainability and health of the code still lies with the human SWE. LLMs are not foolproof and can produce code that is off-track or requires correction. Therefore, SWEs must review and validate the code generated by LLMs before incorporating it into their systems.

The speaker emphasized the importance of understanding code, not just writing it. As the industry moves towards a future where humans spend more time reading and understanding code rather than writing it, it is essential to improve our ability to predict the behavior of every line of code without running tests. This goal may seem ambitious, especially considering the complexity and legacy of existing codebases. However, by designing code to be more understandable, we can make this goal more achievable.

The talk introduced a "starter kit" of practices to help improve code understandability. These practices are not new but are often not taught or applied as effectively as they should be. The speaker aimed to sharpen the understanding of these practices and make them more practical, thereby kick-starting a process of making code more understandable over time.

One key practice is to consider what behavior to rely on when calling a method. By focusing on the expected behavior of a method, developers can better understand the code and its purpose. This practice encourages developers to think more critically about the design and functionality of the code they write.

Another practice involves designing code with testability in mind. A decade or two ago, writing tests for every public method may have seemed infeasible. However, as the industry evolved, practices for designing code to be testable became ingrained. The speaker encouraged adopting similar practices to design for understandability, ensuring that code is easily comprehensible and modifiable as needed.

The speaker also highlighted the importance of documentation, clear error messages, and simple, straightforward code. These elements contribute to the overall understandability of the codebase and make it easier for developers to work with the codebase, make changes, and maintain it over time.

In conclusion, designing code for understandability is an essential skill for modern software engineers. As LLMs generate more code, the responsibility of ensuring code quality and maintainability falls on human SWEs. By focusing on understanding code, setting aspirational goals, and applying learnable practices, developers can work towards a future where they have more confidence in their code and can increase productivity simultaneously.

Written by urgent.news from Lobsters's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at jschuster.org →

More in Tech

Your SVG optimizer silently broke your icon. Here's the one-line check.

Your SVG optimizer silently broke your icon. Here's the one-line check. I spent an afternoon comparing SVG optimizers on a set of 28 real icons from Simple Icons. The sizes looked fine.

  • Optimizer replaced path coordinates with "NaN" string
  • Issue due to handling of arc commands and spacing
  • One-line check using "grep" to detect "NaN"

More from Thursday 19 November →