I agree with both. I tend to have Red Days and Blue Days. Red Days are coding days, working from my plan, no creativity allowed - that's the enemy of lean code.
Blue days are creative days, very little coding - maybe a bug fix if creative me says that bug is really impacting experience. These days I plan big picture, I talk to potential customers and testers. I draw big picture plans on the whiteboard that Red me will implement.
Can break up a day too, but you can never ever be creative and engineer well at the same time. At least I can't, nor can most people I've observed.
Also, in my experience it's the subset of successful products that involved more creative thinking that I am most proud of and look back on my time working on more positively. Hell, this is even true for some of the failures. The raw hypothesis driven successes were high-five worthy, but not pride-worthy (other than perhaps a pat on the back for demonstrating my faith in the Church of Science), and the failures tended to be fairly painful since the work itself was unrewarding intrinsically and the outcome was negative. Nothing is really worse than hacking out code in misery because, counter to your intuition and your gut, signals point to it potentially driving a metric up, and seeing it fail anyway. You have to resign yourself to the process and tell yourself that, despite being right, your intuition is meaningless -- not a fun exercise as a human.
It turns out, in my experience, that the most successful creators are talented at both: trying lots of stuff and learning from their failures, but also having that sense of empathy, intuition, and taste to simply "not be wrong" more times than your average person. You can't A/B test your way to having an opinion. And you can't let an A/B test let you change your opinion on the things that matter.