Sure there is a lot of logical thinking involved but some of the real insights where not based on the logical progression of thought in fact they where based on breaking them. Exactly to deliver insights outside the realm of the logical axioms.
Randomness is not as valuable as logic when working towards a specific end. That is, if you already know the beginning and the end, using randomness to find the path between them is probably the slowest and least likely to bring results.
I am not sure why you think logic is so central here. Of course you need logic, but it will only work if your premise is correct and IMHO if you are trying to solve a problem then most of the time it's your premise that's wrong not the logical reasoning that springs from it.
For a fiction writer, the difference may be negligible. For a rocket scientist, it's a matter of life and death. Literally.