Contra "Glue work considered harmful"

The essay challenges the advice that engineers should limit “glue work” - coordination, maintenance, and improvements that help teams function - to projects they personally lead, because employers reward shipped features instead. It agrees that large tech companies often create those incentives, but argues that treating them as unquestionable is a choice to accept waste. If glue work is necessary to a company’s mission, neglecting it undermines that mission. The author also separates sensible advice to preserve energy for genuine deadlines from the harmful idea that employees should avoid making coworkers’ work easier simply because management struggles to measure the benefit.

Some glue work, the essay argues, removes recurring friction rather than delivering only a temporary efficiency boost. It cites cargo-nextest, a test-running project, and a catalog of test-flake patterns used to build an agent skill that has helped fix nearly a dozen flaky tests. The broader difficulty is that large organizations need work to be legible, encouraging performance reviews, OKRs, and KPIs, yet some valuable support work resists measurement and is consequently overlooked. The author points to Oxide, where asking “how can I be helpful?” has largely dissolved the distinction between feature and glue work. For larger firms, the proposed alternatives include avoiding excess hiring and headcount incentives, not insisting that every task be measurable, and trusting employees to identify useful work.