We operated that way even when steaming on the surface in the shipping lanes shared with the Port of New York.
So it isn't always a matter of "actually knowing how to pilot a boat". Sometimes its just plain recklessness.
Edit: this one
https://features.propublica.org/navy-uss-mccain-crash/navy-i...
But submarines don't spend a heck of a lot of time on the surface where piloting skill matters. They stay submerged through the Straits of Gibraltar, the Straits of Hormuz, a lot of the Red Sea, etc. Submarines are slow on the surface and appear much smaller than they really are, so it's generally best for everyone if they go underneath or around chokepoints.
Theoretically yeah, but radar is pretty great for resolving ambiguity.
https://upload.wikimedia.org/wikipedia/commons/9/98/USS_Los_...
To answer your first question, a significant amount of RCS for a capable commercial radar, not least of all because it is rounded for hydrodynamics, which is a terrible shape for minimizing RCS.
Second, I should have been clearer: the submarine stands to gain by operating their own radar rather than working from sonar alone.
Finally, submerged transits are mostly advantageous for security concerns.
In the debates surrounding some of the most recent incidents (see, e.g., https://www.propublica.org/series/navy-accidents-pacific-7th...) I remember reading claims to the effect that submarine captains are the last ones in the Navy who actually know how to pilot a boat. I know that claim is hyperbolic rhetoric but those Pro Publica pieces really make me think it's not far from the truth, and that while submariner training may still be exceptionally rigorous and responsive to newly identified risks, that's not necessarily true in the case of surface ships.