A computationally-literate idea that's well developed, on the other hand, can be very valuable, and can account for probably 90% of the interestingness of idea+implementation. Not always: sometimes you find really major things in the implementation that cause you to rethink the idea. But there are many times that I've implemented a theoretical idea myself and not really learned anything in doing so. You read a paper, or even a blog post, which explains an idea in detail, motivates why the author developed it, gives a broad sketch of how you'd implement it, etc., etc., but the author hasn't actually implemented it. Then I implement it myself. Have I provided the majority of the value, because I'm the first person with working code? Not really; in many cases the implementation was a pretty straightforward translation of the idea into code.

A computationally literate and well-developed idea is arguably something close to "execution", but not quite the nuts-and-bolts variety. To use a physics analogy, my ideas on space travel are not very well developed or valuable, but Freeman Dyson's are valuable, even though he's implemented his ideas to the same extent I have: neither one of us has ever attempted to build spaceships. He's a pure idea-person, but his ideas are developed quite fully, so readers can understand what he proposed and why, what its pros and cons might be, what possible pitfalls await, what the broad outlines of possible fixes for those pitfalls might be (even if they depend on materials or other things not currently available), etc.

I'd say the same of people even further into idea-land, like Isaac Asimov, who provided valuable ideas with nothing close to an implementation. The trick imo is that most ideas either just aren't novel enough to be interesting, or aren't sufficiently well developed and explained to provide value to a reader.