The Role and Recognition of Glue Work in Tech Companies

Goedecke specifically references projects such as the 'cargo-nextest' initiative, which is described as an example of effective glue work, particularly in the context of test automation. He also mentions his own contribution at Oxide, called 'flake-review,' which was developed over a few hours and has since been used to resolve multiple test flake issues. These examples are drawn from real-world projects within large tech firms, where glue work is often done in isolation without recognition or reward. This kind of work is typically not captured in performance metrics or OKR systems, which are more focused on feature delivery and measurable outcomes. [1]

The post also touches on the broader organizational challenge of legibility, which is the need for large companies to have clear, measurable structures in place to function effectively. This has led to the proliferation of systems like performance reviews, OKRs, and KPIs, which, while useful for tracking progress, often fail to account for the value of glue work. Goedecke notes that some companies, like Meta, have attempted to make glue work more legible by investing in internal tools and infrastructure teams, but he argues that many forms of glue work will remain difficult to quantify and therefore continue to be undervalued. [1]

The author of the post also critiques the inefficiencies inherent in large tech companies, estimating that these organizations often operate at 20-60% efficiency, with a significant portion of employee time spent on solvable problems that are not recognized as valuable by leadership. This inefficiency is not just a byproduct of scale but is also reinforced by the lack of investment in glue work. The post highlights the frustration of professionals who are expected to deliver features while the infrastructure that supports those features is neglected, leading to a culture of underinvestment in the very systems that make feature delivery possible. [1]

Sean Goedecke argues that glue work is critical for the smooth operation of projects but is often undervalued by companies. He explains that companies do not reward glue work because they want employees to prioritize feature delivery over improving general efficiency (Lobsters, 2024). Goedecke suggests that while glue work is essential, companies may not explicitly recognize it due to a focus on measurable outcomes and project success (Lobsters, 2024). [1]

Goedecke highlights that glue work is not inherently harmful but is frequently misunderstood by organizations. He notes that companies often leave value on the table by not properly recognizing the importance of glue work (Lobsters, 2024). While some companies have found ways to make glue work more legible, such as through internal tools and developer infrastructure teams, there will always be some glue work that resists measurement (Lobsters, 2024). [1]

Goedecke also emphasizes that glue work can have long-lasting benefits by permanently removing friction from processes. He provides examples of glue work done effectively, such as running tests and developing tools that improve team efficiency (Lobsters, 2024). However, he acknowledges that companies may not always value these contributions, leading to misaligned incentives and inefficiencies (Lobsters, 2024). [1]

Sources

Lobsters

Prepared with AI assistance from the cited sources.