Joints have been a huge pain point while I am learning the rest of Fusion 360 extremely quickly and your video was the first of 5 to get me on the right track.
(I think that skipping over the demonstration of creating a joint between two faces could make this video easier to absorb. Maybe.)
I think I have captured the source of the cognitive disconnect:
1) The term "joint" can be thought of as a replacement to "mate" and "mate" can be a verb instead of a noun. So I click on "joint" and think that I'm about to create a connection, not create a connection point.
2) The dialog box for "join" includes the option to select two things by default. This reinforces the false assumption that I'm about to perform the act of "joining" something.
3) The motion type is defined in the creation of the joint dialog box. This reinforces the false assumption that
Humans figure things out by observing patterns and applying existing known patterns and intuition. In this case, the "joint" workflow and terminology creates a trap that confuses users of other systems.
Also, even if I experiment and my joints WILL work, the OK button stays grey so I never know it. I found out the hard way that if I simply click in an empty spot on the canvas the joint does what I expected. This caused a huge amount of confusion.
It doesn't help when you want to do a surface-surface joint (I think a very basic operation) and the official help page involves complex spinning gears but no example for surface-surface.