Most startups fail. That doesn't mean that starting a startup isn't an activity with high social value.

This depends on the extent to which a startup's success can be predicted in advance. My impression is that startup accelerators and venture capitalists do a reasonable job of predicting the set of startups that are likely to succeed. If a startup is in the threshold where a venture capitalist or accelerator considers them fundable, then yes, I think the activity has high social value in expectation, even if it ends up failing.

Ancient Greece had their version of the steam engine that was build to impress people. They didn't get economic value out of it, but the desire to build something to impress produced innovative thinking. A lot of innovation comes out of exploring a subject and simply trying to do something impressive.

I would be interested in a more detailed analysis of the value produced by innovative thinking in such historical contexts. At the same time, it seems the case to me that if there is a choice between something socially valuable and something that's not, then ceteris paribus, the more socially valuable thing is preferable.

In the context of people pursuing mainstream academics, they can be somewhat misguided about how well they understood material, but not too much.

That['s] not true.

There could certainly be a disconnect between their academic understanding and their ability to deal with real-world phenomena that apply that academic understanding. My comment was more about their level of understanding with the specific academic realm. That said, I do think many people don't learn even the material they are directly learning well enough. I've linked in the past to Eric Mazur's video on physics teaching as am example.

Steve is more likely to tackle problems differently then the people who already work in a field than Dave. To the extend that I want to make strategic choices that encourage innovation, I want more people with the kind of mindset that Steve has.

Do you have the impression that people who design marketing materials for nonprofits are in general more likely to think out of the box than people who can learn a complicated subject such as Japanese calligraphy?

I would be mildly inclined in Steve's favor based on what's known in the post, but I don't think the information as presented is strong enough to make a very strong case for one candidate.

The people I know anecdotally who did more of the Steve sort of stuff in high school don't seem to have accomplished notably more in adult life than the people who did more of the Dave sort of stuff. This could be due to small sample size or selection bias in my sample.