Call me unimaginative, but I think fingers are good at manipulation more so than hand signs. So I can imagine either stuff like 3d modelling tasks, or when you'

Call me unimaginative, but I think fingers are good at manipulation more so than hand signs. So I can imagine either stuff like 3d modelling tasks, or when you're interacting with a 3d world (think games). Smush-deforming play doh with your hand is so much quicker and more precise than trying to do the same with a mouse or shudder keyboard. Or imagine doing knitting. I think hands are great when there is a massive multitude of possible next actions, like the knitting example, or artistic applications. Otherwise, if we can logically eliminate options of what the user could be doing (this object can only go in so many slots, etc), it's easy to just assist the user with that (think snap-to-grid) and then a precision input device is less necessary.