Maybe just different styles and approaches and typing speed is a easy, obviously, and measurable metric to focus on but this sounds like brute force problem solving. I tend to prefer to build a mental model/plan such that once I start writing the first draft is probably within 90% of the final, some iteration taking place on that last 10%. But I don’t write and delete large swaths of code. This sounds like real time refactoring/lack of planning and completely erases any gain I’d get from being able to type twice as fast (I’m probably middle/average speed).

To use your mechanic analogy, the difference is the mechanic is confident in his abilities to tinker. The confidence is built on having a plan and some background knowledge. When he disassembles a carburetor he already knows the performance tweaks he’s going to make and how he’ll reassemble it. He doesn’t just start taking things apart with no plan, or start playing swapping on and off a handful of carburetors to see which performs best. He probably did his research and chose the carburetor/mod that was appropriate. It’s actually even more important with the tangible world because he has to plan and acquire the parts and tools before getting started or risk having a bricked vehicle for some period of time.

Non-mechanics know better than to tinker at all because they’ll likely break something they can’t figure out how to fix. So they might do low risk things like put high performance fluids in, or air filters, but they’ll never match up to the guys that have upgraded the mechanical systems.