Is it possible that some directors do it intentionally? Like, if the character goes into the scene with the layout and geography of everything in their head, they give the audience that information too, but if they don't, then the info is kept from the viewer?
I dunno, but it seems like a mistake to just have things popping in and out of an action scene nonsensically. The surprise should probably come from unpredictable actions instead of previously unseen objects just appearing in frame to do things. Editing is a hell of a thing.
I always think of The Matrix Reloaded's car chase and being a gold standard, because the Wachowskis loved to use wide shots in all their action, so you have shots like this:
where you see Morpheus on the truck and Trinity on the bike in the distance. This is just before
he grabs the Keymaker off her back and hurls him up onto the truck, so the action makes perfect sense because you just saw the locations of everyone involved all at once.
This is not something Michael Bay does well. He can frame a pretty looking image, but he often cuts too incoherently between places, people, and actions without establishing where everything is in relation to everything else. As a comparison,
here's a similar scene where Bumblebee rescues Sam and another guy from falling off building. It's a slick shot and it's framed well, but where the hell was Bumblebee before that? The rescue sorta makes for a nice surprise, but it lacks impact and doesn't make structural sense because the scene never took a moment to establish where Bumblebee was in relation to Sam & Starscream before the save. Not knowing the save was gonna happen isn't the problem. It's not knowing Bumblebee was even going to be involved in the moment.