Flaky tests — tests that pass or fail without code changes — are one of the most insidious productivity killers in modern software engineering. Lucas and Luna dig into a 2025 study from Google that found flaky tests cost their teams roughly 16 developer-years per quarter in reruns and debugging. They unpack the three main causes: order-dependent tests, race conditions, and brittle assertions. Then they walk through practical fixes: quarantine systems, rerun strategies with exponential backoff, and deterministic test ordering. Luna brings a counterpoint from her own team's experience with a particularly nasty flaky test tied to a third-party API mock. The episode closes with a forward-looking question about whether AI tools can automatically quarantine flaky tests before they poison a CI pipeline. Concrete, actionable, data-backed — no fluff, just signal.

#TestFlakiness #CI #SoftwareTesting #QA #GoogleStudy #DeveloperProductivity #RaceConditions #TestAutomation #ContinuousIntegration #BrittleTests #TestOrdering #ExponentialBackoff #FlakyTestQuarantine #Technology #FexingoBusiness #BusinessPodcast #SoftwareEngineering #Reliability

Keep every episode free: buymeacoffee.com/fexingo