I see your point here — you’re trying to say that over time, users will get it. But that doesn’t make a user efficient over time. A good system might require training, but will still feel obvious (helping build muscle memory) over extended use. That is how iOS used to be, especially upto iOS 6. To your point about iOS 9, the interaction with the lock screen camera didn’t introduce gesture conflicts. In iOS 10 — Yes, people will learn this interface over time. But is there a cognitive load in day to day use? Yes, again. — And that is not a great thing.

I am totally with you on the convenience of 3D touch (I am a huge fan of it) but that still doesn’t alleviate the issues mentioned in my post. If anything, it only adds to the number of ways in which one can interact with a notification.

To your point about the third task, users won’t necessarily remember where they came from to easily go back to what they were doing. I sometimes use widgets to just have a quick glance at information. Like I mention in my post, redundancy of widgets combined with different ways to exit the way becomes an issue. Users might figure it out, but it will make them think — again, the pause is caused by a cognitive load which cannot necessarily be measured by task completion.

You rightly mentioned it yourself that this lockscreen is a complicated mess and this post tries to dig into why that is the case. :-)