I think this is the difference between research and industry. Industry should try to grind out obvious improvements through brute force iteration. I really wish the culture of academia was more of an aim towards moonshots (high risk, high reward).