You're assuming that a) good-speakerness is a fixed quality, and b) that the speaker selection process is something like pulling bingo numbers. Neither is true.