The thing is, Iran doesn’t need to. US maybe can defend their ships, but they can’t defend commercial ships well enough for them to resume regular operations. Even unsuccessful attacks would cause insurance to make it not possible.

Houthis closed their straight some years ago and US wasn’t able to do anything about that neither. And Houthis are nowhere near as capable as Iran.

US gambled on decapitation strike and failed.

Yes, that is a fair point. However, the cost of drone versus latest generation ballistic missile that has a chance to reach us naval ship is very different. And in that sense, iterating on a drone is closer to iterating on a line of code because one drone would cost you a thousand bucks and your iteration is a small tweak like adding a different grenade triggering mechanism. Rockets require custom design, custom manufacturing lines, and generally much more difficult to modify and make more effective.

You also have a lot more tries with cheap drones since the target is lower value, so you have hundreds of data points on how each iteration performs vs hitting a naval ship which is an extremely rare event, so it's hard to see whether your iteration on a rocket actually succeeded.