When perfection is table stakes

When rewriting or replacing existing software, the minimum acceptable outcome is matching every capability of the previous system for all users at all times — perfection is table stakes. Using Linux audio (PulseAudio vs ALSA, then Pipewire) and Wayland vs X11 as examples, the author illustrates how users will complain about any regression regardless of new features or architectural improvements. The key lesson: proceed with rewrites but set expectations correctly — anticipate complaints until feature parity is reached, don't dismiss user frustrations, and understand that even UI changes force relearning costs on users. Success looks like users not noticing the switch at all.