Iwould argue that the issue here is that the term 'realism' is more like 'internal consistwncy' or 'verisimilitude'. When we have an event that happens and breaks verisimilitude it breaks us out of the game. Whereas accepting the initial premise of warp drive means that a ship warping into space next tk you isn't braking the expectation. Whereas dissassociated mechanics are breaking those expectations because there's no reason why the character ahould only make a one handed catch per game.

I like the idea of discovering something random that gives you a boon.

On the other hand I also like the idea of a caravan saving you energy but making an encounter more likely. I suppose the most logical would be that the scale shifts and your're unlikely to ckme accross a smaller encounter bht quite likely to come accross a larger oneI wouldn't want to up the likelihood too much more the result if you do get an encounter.

This is an excellent method for any kind of challenge. Not just a boss fight.

Very good enounter tables to use.

.

While I agree that rolling for things you want them to succeed at can unecessarily stunt the narrative, when combined with the idea in this article of rolling to see how well or poorly you do it could be used to give you narrative fodder. Like this example of the door. You don'tneed to roll but rolling might give you a nice connecting moment so you'renot skipping through the adventure to the 'good' stuff not having any fibre in amongst the sugar.

.

.

You could also make the roll a part of the solution. i.e. if they'retryong to do something hard it might tilt the balance ratger than it being the focus.

One weakness is trying to fit everything into specific boxes. I woukd argue this isn't persuasion.