Dave Rupert makes a point that should make every engineering manager uncomfortable: when organizations obsess over speed, the first casualty is talking to each other in When teams move fast, talking breaks first. Consensus takes time. Getting input from subject matter experts takes time. So under pressure, teams just stop doing it. Rupert argues this creates a future "messy merge conflict" between teams traveling in different directions. And he sees AI making things worse. He calls LLMs "the ultimate tool in the 'Don't talk to my coworkers' toolchain." Why consult a human expert who might say no when an AI will always say yes? But avoiding that friction has costs. Rupert writes that higher sunk costs and deeper entrenched opinions make future conversations harder. Speed also kills shared systems. Under tight deadlines, nobody improves common codebases or design systems. Developers detach Figma components, create duplicate project folders, and build slightly incompatible versions of the same thing. Each duplication is a time bomb of technical debt. Hacker News commenters largely agreed, invoking the "slow is smooth, smooth is fast" philosophy. Sure, some corporate conversations are political posturing, not genuine collaboration. Fair enough. But Rupert's argument holds. When junior developers watch seniors bypass colleagues to query AI instead, they miss the messy process of building architectural context. They're left maintaining duplicative systems they don't understand. The fix isn't complicated. Rupert says engineering management needs to stop optimizing for ticket velocity and start helping organizations "row in the same direction together."
Speed kills team communication, and AI makes it worse
Dave Rupert argues that prioritizing speed leads teams to stop talking, building consensus, and maintaining shared systems. He sees AI/LLMs making this worse by letting developers bypass experts and colleagues, creating technical debt and duplicate systems that make future conversations harder.