What I'm suggesting is that the problem isn't so much the mechanic itself as the AI's failure to account for it in their own positioning. I've found that it's generally not that difficult to avoid having your own characters knocked off cliffs. So turning it into an action? Sure, fine. Reducing the distance? That's fine, too. But if it's too easy or too common, I'm asserting that's simply because the NPCs stand near cliffs too often. (For my part, I don't often use it against bosses, if only because I generally want to loot them afterward, which bottomless pits or lava usually prevent. Nere isn't that hard to kill conventionally, and you'll break at *least* one quest by pushing him into lava.)
(BTW, jfutral and Wormerine, there's a range of views expressed in the thread, but both the original poster and mrfuji3 both seem to me to argue that the availability of a 'one-shot kill' action like pushing someone off a cliff derogates from the D&D combat system which is premised, in mrfuji3's words, on "dealing damage over multiple turns to reduce enemy HP to zero".)
I agree that "enemies standing near cliffs too often" is at least partially a function of AI programming. But it also definitely depends on shove distance - the large shove distance in BG3 makes it much harder to
not stand close to a cliff. Especially as you can have multiple of your party members push an enemy before it gets back to their turn - now they have to stay 2-4x the distance away from any cliff, which is impossible in many areas.
And yes, I am arguing that there shouldn't be a
common/easy mechanic that one shot kills enemies. The Hag, Spider Matriarch, Absolute leaders, Nere, Githyanki patrol, boat Duergar - these are all encounters where an instant death pit is a central feature of the arena, and it's not great that so many of these are boss fights. D&D is a party-based game; a OHKO (bonus) action button devalues the rest of the game mechanics, party synergy, and any tactical combat.
If shove was reduced to 5 feet AND enemies avoided being near cliffs, then the risk-to-reward would be more fair. You'd likely have to succeed on ~2-4 shoves to push an enemy off a cliff, which would be a large investment of turns&actions: high risk, but high reward.
