Is it, though? How many video games have you played where you can sneak right up behind an enemy without them detecting you because their backs are turned? It's not much different, really.
True, but I think that for stealth to work that way it would need to be simplified even further - so it could be readable without vision cone.
My concern is that stealth in BG3 is a binary state - you are either spotter or you are not. As such there is no way to indicate player if they enter enemy vision range - that applies to both distance, and elevation. While some (especially older) stealth games have no vision cones, they do have awerness indicators (through UI, or verbal like in Theif1&2) which inform player and allow them to react. Of course, with how clunky and unresponsive BG3 controls are that wouldn't really be very helpful either.
As a side note, there is a game that uses similar system that works very very well - Invisible Inc. Being a pure stealth titles, there is of course far more going on to support that design.
Yup. D&D uses very abstracted roll system to represent character's ability to remain stealthed. Larian decided to make this system more granual, but implemented only one aspect of what could be considered a more comprehensive stealth system. Adding one aspect of traditional stealth system doesn't result in a better stealth system - just an unfinished and exploitable one.
That an issue I have with quite a few Larian additions - who our character is should decide what the character can do. That is how D&D is built. Many of Larian additions dont work that way. I think that is what makes them so OP most of the time - you can just reliably do them, and that doesn't play well with RNG-based core of the game. That's the very reason why things like stocks, granades and drones in XCOM2 break the game.
