next »

[randomThrowAway]

My (fairly straightforward, turn based) jrpg battle system currently uses no randomness at all, which I decided on for several reasons. Fighting is going to be the central part of the game, every little fight vs 3 rats is meaningful.

AI: Much easier to find a "correct" move when the results are deterministic, I'm just looping through all 15-20 actions available in a given turn once and weigh the outcomes by the AI "personality". I can look into the future with certainty, which has many benefits.

No bad beats: I come from an online gambling background and I know for a fact people hate losing a 50 percenter much more than they like winning a 50 percenter. Tangentially, that's what's wrong with PvP in general imo.

Accountability: No blaming the rng (and I'm even thinking about "balancing" encounters to be winnable which again is much easier to check for, using no random), this means every player move counts and winning or losing is exclusively about the player's decisions. Fights are more like puzzles than crapshoots, especially if they are hard but you know for certain if you make the right decisions, you'll win.

General neatness: It's hard to explain, but using no random gives me a warm, fuzzy feeling about my combat, as if it were more pure and elegant, that's probably bullshit but oh well.

I'm still not sure though, there are definitely pros to randomness:

Convention: Every gamer knows about and basically expects crits and crit chance, evasion, damage ranges and so on in a rpg. It's a simple and easy way to give people what they (think they?) want and a template to produce content (equipment, skills that are based on crit etc).

Status magic / debuffs: I want those spells to be actually useful and getting used vs stronger enemies as well. Without random, a sleeping spell will always work or never work on a given enemy during an encounter, so I have no granularity, compared to setting "sleep vulnerability" on the boss to 20 pct chance or whatever.
If "sleep" regularly has a 0 pct chance to succeed, nobody in their right mind will try to use it. I do want some concept of (sliding) resistance towards status magic, but the current system of (if resistance > spell power fail, else succeed) is lame. 

Grambling isn't all bad: A bit contradictory to the "bad beat" point, people do like to get lucky; gambling is fun, after all. Turning a "hopeless" situation around is always a good story.

Express yourself: Random allows the player to make decisions that are risky or safe (haymaker vs jab), not only at a given moment but also long term (glass cannon vs balanced hero), although for this one, I think I can emulate it without using random.

I could go on but that's it for now, looking forward to hear your thoughts re: randomness in jrpg battles.

[lolpatrol]

As long as the randomness is not a crutch to add artificial "depth", and it actually enhances the combat.

I tend to prefer randomness you can react to before a 'move', and not something that happens after you did the, on paper at least, best move - like a random "MISS". That can be pretty frustrating.

[s0]

tbh most jrpg battle systems are already pretty deterministic. in your average final fantasy the chance to actually miss an attack is tiny and the random damage roll is nearly irrelevant (it doesn't really matter if you do 37438 or 37987 dmg).

[randomThrowAway]

True on both accounts. The thing is, it's not really a jrpg, I'm only using jrpg to quickly communicate the way combat works.

I'm not bound by tropes and I'm designing the combat from ground up (no rpg maker like stuff used), so the question isn't a binary "no random" or "ff random", but also if I used random, how much. As an aside, I use a lot of randomness for the game outside of actual combat, every enemy has random stats and so on.

So yeah, both points you mentioned don't make much of a difference to the experience ultimately, so if pressed between no random and "slight dmg variations, once in a blue moon you miss and get angry", I'm comfortable with going with no random.

I guess I'd be more interested in something pnp rpg ish then; having tight and wide damage ranges on different skills like between 2d6 and 1d6+4 dmg for example. Also (player) evasion is something that would be cool to simulate as an alternative way to deal with damage. One idea I have for evasion in a non random world could be 10 pct evasion lowers dmg taken by 10 pct, while 10 armor lowers it by 10 points.

I'd never want to go full (you have to be German from your name, DSA experience I call it) "try to attack - miss", "try to attack - success. Enemy tries to block - success. Outcome: nothing happened." though, I want every turn to be meaningful.

Things I'd like to use random for:
- Meaningfully different damage ranges
- Player evasion
- control / status magic

I don't like random:
- Player hit chance
- AI targeting

I'm neutral about:
- crits

I guess, writing this out, I actually want to introduce randomness again, but I have to be careful about it and not make dice rolls the core of combat.

[ThemsAllTook]

I'm normally against randomness in a game like this, but a particular case came to my mind that feels pretty good. In Paper Mario, there's a badge you can equip (Pretty Lucky, I think it's called) that makes enemies occasionally miss their attacks. It's unreliable, but when it happens it's a really nice bonus. The rest of the game uses very little randomness (all attacks do a fixed amount of damage, depending on stats and player actions when executing them), so it's very predictable. The Pretty Lucky badge just adds a happy accident occasionally that works in the player's favor.

[randomThrowAway]

Thanks, yeah I'm stealing that one. For now I will use random only for damage ranges, evasion will not exist except for when a hero or enemy has a special evasion attribute, then it will play a big role, like 25% evasion chance.

Heroes will have to sacrifice other stuff if they want to "equip" evasion or need to cast it as buff and only a handful enemies / bosses will use it where it's fitting.

[JWK5]

With gambling, the greater the stakes the greater the payoff. High stakes with low payoff is not worth it.

A sword that reduces an enemy's health by 1/3 (kills in 3 hits) but has a 25% chance to miss can create tension because the gamble is pretty high, especially when you are just one hit away. The tension increases if the enemy can also take the player out in just a few hits, since now the gamble is even greater (if the player misses consecutively the enemy could wipe the floor with them).

If the sword reduced the enemy's health by 1/10th (kills in 10 hits) but had that 25% chance to miss it'd get pretty frustrating because it each hit already has such a poor payoff that every miss is going to feel like thie player is on a slippery slope (with the enemy working the player down far too fast).

When adding random to your game, don't think of it as "a random chance to do X". Think of it as gambling. What are the stakes? What is the payoff? Always make sure the gambles in your game have payoffs that match the stakes.

[BeeZee]

I remember thinking Chrono Trigger was going to totally revolutionize the genre, and that everything that followed would copy it and put the battles on the primary map. I'm still blown away that didn't happen.

[Jason S. Longia]

As long as the randomness is not a crutch to add artificial "depth", and it actually enhances the combat.

I tend to prefer randomness you can react to before a 'move', and not something that happens after you did the, on paper at least, best move - like a random "MISS". That can be pretty frustrating.

Random in that sense would be cool, like for example have a "Random Bar" that shows up on the enemy and your side. And as you level up or something it raises. You could face the enemy several times and have completely different battles. Like Dungeons and Dragons.

[voidSkipper]

I think you're creating a false dilemma with your granularity statement, there.

Yes, a sleep spell traditionally has a success rate and a failure rate based on your aptitude and your enemy's susceptibility.

However, that still results in a binary output - success or failure.

There are different ways to implement granularity that don't involve randomness.

Going with the 'status effects' example, rather than "asleep" being a binary value, it could be a scale from 0 to 100. A single sleep spell's effect on "asleepness" is deterministic based on caster aptitude/victim susceptibility, so the strategy becomes how many times do I need to cast this spell to get a meaningful effect?

Sleepiness could lower attack speed gradually, eventually at some point causing entire turns to be missed.

This allows determinalistic battles with a very large amount of permutations.

[BeeZee]

I really like what voidSkipper is saying here. One of my biggest issues with, say, the traditional FF status effects, is that they rely more on RNG than on skill or strategy. Not only is the success of a status effect random, but many enemies and bosses are immune, meaning that you could waste a huge number of turns trying to get one to land with no meaningful feedback on whether or not it's even possible. Simply removing the RNG from the equation makes the mechanic infinitely more viable.

[randomThrowAway]

I agree completely on the status effects, what I'm doing now is actually quite similar to what voidSkipper said: Hero has a spell "Sleep 1", enemy has sleep resistance 1.9.
Hero has to cast sleep 1 twice to make the enemy fall asleep, or two heroes cast it once etc.

I think binarity (is that a word?) works better than granularity for some standard status effects like sleep or petrified since they are so ingrained as binary already, while "becoming more sleepy" would be more of a new (pretty cool) status effect (progressive stat debuff that is) alltogether.

[dao cowboy]

In discussions like this, I used to always come down on the "random is good" side of the aisle, since I thought dealing with an rng might develop a healthy sense of risk tolerance.  But I realized as a gamer I didn't like it.

My take on this is that it might help to make status effects very meaningful.  A lot of times in JRPG's, all the player is thinking about is how to deal damage.  Fight command, powerful spell, whatever.  What you would have to do to make a non-random JRPG compelling IMO is to reward players heavily for *not* always choosing damage.  You could:

--make certain attacks only effective against enemies who have a status condition.  Or have attacks that can only be accessed in a buffed state.

--have multi-stage status/buffs.  for instance, sleep spells only work on dazed enemies. 

or etc.

edit:  of course you are going to want to have situations where dealing high damage quickly is a priority, eg enemy glass cannons.

[s0]

--make certain attacks only effective against enemies who have a status condition.  Or have attacks that can only be accessed in a buffed state.

--have multi-stage status/buffs.  for instance, sleep spells only work on dazed enemies.

that's what xenoblade does and it works very well.

[valrus]

Another way to sequester the effects of randomness would be to have randomness be a field condition (either local or for the entire battlefield) that itself can be manipulated.  (Like the elements in Chrono Chross or the alignment in Chaos Reborn.)

Some things you can do/take/equip in battle increase Chaos and increase the effect of the RNG on play, others bring the battlefield back to Order where there's no randomness.  So the player can pull the game in the direction they prefer to play in, or pull it back and forth between Chaos and Order depending on what gives them the most advantage at the moment.

[King Hadas]

I hate Random Player Hit Chance. In tabletop rpgs I can see their purpose because you're not just trying to "Win", you're building a narrative. Responding creatively to an incredible hit or an embarrassing miss is a lot of the fun. The further you move from actual "role playing" though the more this is unacceptable. So when your focus shifts entirely towards "Winning", these results you can't consistently predict become unbearable.