Are there some types of RNG you’re ok with?
Looking at my notes, yes, there is.
Enemy attacks in a TON of action games are at least partially random, but they’re designed to always be longer than 15 frames of startup, meaning you have time to react. There’s always a tell of some kind more than 15 frames away. Even if the attack startup is shorter than 15 frames, the tell can be the enemy approaching you, and knowing they’ll attack when in range. DMC3 has audio cues for every single enemy made so the game can be played practically blind with royal guard.
Action puzzle games can afford to be random, because they give you a lot of space to mitigate random bad luck before it piles up, because set patterns would become optimization/memorization challenges, and because they use very fair randomization algorithms, the most common being going through all the pieces once before starting over, or otherwise scheduling piece frequency to prevent droughts and floods. Plus in multiplayer, the same random seed is used for both players.
Game and Watch’s hammer and Peach’s turnip pulls are random, however G&W has the same hitbox every time, so it never changes the value of the move as a neutral option, and peach gives you a chance to see what item it is beforehand (though to be honest, this needs a better regulating algorithm). In Project M, G&W cannot use the same number as the last two he used, and there is a light telling you whether the next number is even or odd, so you can get a further idea what you are likely to get, and stack the odds.
I’m not fond of Roguelikes, but those are largely personal reasons. The genre would not exist without randomness, and it’s a genre that needs to exist (even if it gets a bit degenerate by nature of resetting until you win, presumably because you got an easy seed).
Randomness creates the opportunity for adaptation challenges. The ability to adapt to unknown circumstances is a skill like any other, however testing it fairly is tricky. Most random algorithms used by games are pretty much just a pure probability table with no regulation on what comes up next. Good use of RNG is a matter of recognizing what you actually want to test in players and how to keep the level of challenge and the possibility of success consistent. Bad use of RNG is random bullet spread, it pretty much epitomizes everything bad about RNG. Random chance to hit, random chance to avoid damage by pacifying enemies, random amount of damage. Just don’t make the spread random, or make the particle effect random or something, but the underlying mechanism deterministic. Perhaps you could even rig it so 1 in X bullets misses depending on the range, and when they get really far out so 1 in X bullets hits. Want it spread over targets in the cone of fire? Move the bullet that missed to another target within the cone automatically. You don’t need to do this stuff randomly.
Not all randomness is bad, some is desirable, it just needs to be treated really carefully. It’s currently applied without thought.
What are the worst forms of RNG you’ve experienced in videogames? For me, it’s tripping in Brawl.
I think I’d have to agree. I’ve had a pretty bad time with inaccuracies on spray weapons in all sorts of multiplayer online games.
Oh, and mario party.
I want to make a Mario Party Betting game, for observers (players would be a bit biased, tendency to bet on yourself because you like yourself, or others and underperform). Something to make the hopelessness of mario party more horrible. I only have a loose framework so far, like every X turns, you are allowed to place down a bet on one of the players winning the whole thing, so you have to make big commitments on something far off in the future that can get upset at any time. Then probably smaller bets on each minigame that let you somewhat shift your bigger bets around based on whether you win or not.
The drinking game would just be drink every turn. Just drink.