Yet I remember choosing a data structure too hastily and that killing a project a couple months later. By the time the error was realized, there wasn't time to reboot, project dead. You want a minimal viable product and both words create very difficult judgement calls.