It works well for 99% of the time, but when it doesn't, this design decisions make the experience frustrating. I'm-going-to-write-a-blog-post-and-boycott-this-thing frustrating.
And if the description is not clear (as in the Rdio example), it'll surprise the user. And surprise is not good. Let's say you are mindlessly managing your collection, doing some syncs along the way. When you finish, you notice that there are musics in the collection that you didn't put there. Even if that was exactly what you wanted, you don't know how you did it and now the software is not predictable anymore. You'll flinch before using it again. I know I would.
Multi-action options are a bless when aligned with your flow, but if not used carefully they'll bring down the whole user experience.
Totally. That's the point of a good product person -- they need to understand who is using the product, and what the point of the whole thing is. If it's something that they know users will understand, then it can be done in the subtle way that Rdio has. If they get feedback from users that are confused, they can change the product to make the process make more sense.
Surprise isn't always bad. I bet a lot of people are surprised the first time they turn an iPad sideways and it rotates. That doesn't make it bad. A good product person surprises users with good things.
I think we have a different definition of "unpredictable".
I don't mind (within reason) when an option has unadvertised consequences, such as in the Rdio example. That doesn't strike me as "unpredictable" as much as it's "complex".
The "unpredictability" that I dislike is when I take an action and cannot predict what happens. Sometimes X happens and sometimes Y happens. When I go to a page, I want to see this thing. When I sync my iPhone, I know I want that thing to occur.
"I'll listen to these podcasts and that audiobook on my trip, but I don't want them in my collection once I've listened to them..." for an obvious scenario.
I think engineers know about use-case analysis. Don't bother to present the user with options for which there is no use-case. This isn't some sort of voodoo that only "product people" understand, it's software engineering 101.
And if the description is not clear (as in the Rdio example), it'll surprise the user. And surprise is not good. Let's say you are mindlessly managing your collection, doing some syncs along the way. When you finish, you notice that there are musics in the collection that you didn't put there. Even if that was exactly what you wanted, you don't know how you did it and now the software is not predictable anymore. You'll flinch before using it again. I know I would.
Multi-action options are a bless when aligned with your flow, but if not used carefully they'll bring down the whole user experience.