Earlier I wrote about the opportunity cost of skips in racers/platformers, and how they can diminish the extent to which games can test skill. The same logic applies to action games, most notably with regards to quick kills. It does have some interesting quirks that aren't present in movement games, though.
Quick killing enemies usually means killing them before they get to act, sometimes even at their spawn location. All action games have elements of this, and touch of death combos can be considered a different type of the same phenomenon as well.
This invalidates a lot of the more interesting enemy dynamics that can happen when their behaviors interact. The point of randomizing enemy moves and creating distinctions is to shuffle around the possible situations games can put players in and test their decision making skills. Without that, players have no reason to learn a game more holistically because it won't make them a better player.
There is a bit of a complication here however. Unlike movement platformers, action games (especially beat 'em up inspired games) rely on a high degree of randomization for their challenge. The more optimized they get in terms of score or time, the more of a factor the randomness becomes.
Usually you have a kind of averaging out effect where even if you lose time due to RNG, consistently good performance will outweigh it, especially in lengthier run-based games. But because the internal logic of the game is RNG dependent, this won't scale up very well - the better players get, the more the outcome of any given run will be dictated by luck.
This is where quick kills and touch of death combos can actually help. They may fundamentallty change which skills the games test, but they offer a more reliable and predictable layer for players to engage with which scales up very well as they get better. A more stable meta-game.
If developers go that route though, they have to make damn sure that there is a lot of skill-based variance in how quickly enemies can get killed. A beginner's quick kill should be much slower & easier than the expert's quick kill even if both ultimately kill an enemy before they get to act.
There's nothing worse than ostensibly "deep" games which are basically over once you learn all the quick kill strats, because the act of quick killing is easy and lacks depth.
