A quick warning : This is a messy post because my thoughts on the topic arent very clear. But it's been sitting in drafts so long without coming together into anything snappy that I no longer care. So if it's a confusing mess, don't blame me, blame yourself for reading it!
I fuckin love context sensitivity. Whenever a game has context sensitive melee attacks, throws, proximity normals or something similar, attacks which are heavily tied to player movement, abilities tied to the presence/absence of certain level elements, etc. I get excited. I've always had trouble explaining exactly why I get excited about this shit, so I wanna try & make sense of it.
There are basically 2 types of mechanics in games - universal & context sensitive. Universal mechanics are more-or-less always accessible to players from a neutral state, and are based on the player's input. Your ability to move from a neutral state is mostly up to your whims as a player. Context-sensitive mechanics aren't like this, they are locked until the player and/or enemies get in a specific situation, you have no access to any of this shit from a neutral state.
A layer of universal mechanics is necessary (good luck making movement context sensitive without fucking up a million different things!) but once the universal mechanics move away from a way to set up context-sensitive interactions and themselves become the driving force of those interactions you start getting a lot of problems. These mechanics exist in a void. A vacuum that's separate from the current state of the playfield. The competition between them is "internal" in a sense - 2 similar attacks will always butt heads directly, increasing the likelyhood that one of them will always win, leading to degen strats.
Context sensitive tools are the panacea to this. They prevent similar but slightly different attacks from competing with each other directly because your access to them is situational - often even if one of them is always better on paper, the situation will not allow you to get easy access to it. This is the sorta shit that creates tactical variety through the player's situational awareness and ability to capitalize on "good context". Or better yet, the player's ability to create said good context.
This lack of direct competition between tools & limited access also means that the tools themselves can be incredibly strong, even overpowered. After all, an insanely powerful but context sensitive option is usually called a goal, or the final stage of a game plan. The more robust and interesting the "set up" phase is, the more true this is.
Context sensitive mechanics are integrationist design, which I absolutely love. And the more you integrate things the better it is. If the context sensitivity is based on things like proximity to enemies, it creates proactive play because the main ways to gain an advantage are the very same things that allow enemies to engage with you most actively - it's a good way to fight off stagnation in game state, making sure actions always advance it. It makes the minor differences in enemy behavior matter a lot too. If where an enemy stands directly affects your ability to attack a huge group (via throws or contextual AOEs) then their positioning TRULY matters in a way thats not a binary "are they in attack range? True/false" check. Context sensitive actions being tied to scenery makes the terrain itself matter, that's a wonderful aspect of cover shooters which needs to be reevaluated. Tie both together and you get a very tightly interconnected well oiled machine of a game where everything affects everything else.
I think it's worth distinguishing between "setup" and "active interaction" when talking about context sensitivity, as vague as that line is. I think that setup tools benefit from being universal (movement being the best example) while interactions benefit from being context sensitive. Mostly cause if you don't have a robust setup element, then the depth/variety you can get out of context sensitive interactions is limited. The good thing about universal mechanics is their flexibility, and ideally you want to capture as much of that as possible while still keeping things contextual. So if you could only initiate the RE4/5 kick from a single direction, it'd be contextual but very rigid in how its used, shallow. Context sensitive actions are better when they are orbit points, rather than discrete prebaked interactions, if you get what I mean. You want multipurpose moves.
Whats more is that context sensitive interactions, like all interactions, should feed back into the rest of the game. So a RE4 kick shouldn't just be a way to kick that enemy, it should let you meaningfully affect the other enemies via big hitboxes, collision damage, etc. Same with proximity normals - you probably want there to be some kinda broader tactical reason to use 1 normal over another, rather than it being contained to that one specific enemy youre interacting with. This is why iframe finishers or beat em up/RE5 pickup iframes tend to be cool - they give you an overall advantage.
