Later, as he was working on a new feature, he realized he could achieve a particular effect without any images for a slight improvement in performance (fewer requests and all that). That achievement came because of the experimentation he did a week earlier.
I think there can be a lot of value in experimenting for the sake of experimenting. It's part of the fun of coding. Aside from the potential morale benefits, sometimes the experimentation leads to useful inspirations that make it to production.