Tuesday, October 13, 2015

Why is Healing So Hard To Get Right?

Healing. It's a part of nearly every RPG out there, and surprisingly, it's a difficult thing to get right in a way that makes sense for the players and makes it fun.

If you look at your health as a resource, and enemy attacks generally spend that resource, healing is just a method to extend that resource. i.e.: if you have 100 health, and the enemy does about 10 damage per turn, your time-to-live is about 10 rounds. If you can heal for 30 points during those 10 rounds, your time-to-live (TTL) is actually 13 rounds or so.

Of course, if healing available is unlimited in nature, then your TTL is also infinite, short of the players just not noticing someone needs to be healed. You can limit healing by limiting the amount of throughput per action, or limit the resources healing itself uses (via cooldowns, mana cost, consumables, etc.), or some combination of the two.

So in our example, perhaps you can actually output 100 points of healing, but then you run out of mana. So the TTL might be 20 rounds, but perhaps you need that mana for other encounters down the road. Or maybe you know that the enemy damage pattern is 10 damage per turn, but maybe it's actually 5 damage per turn for 4 turns, then 25 damage per turn for 4 turns, then 0 damage per turn for 4 turns, but your healing spell has a cooldown of 3 turns. You know you'll have to output some healing during that burst to survive it, so you want to make sure it's off cooldown sometime in the middle.

Health itself is often rolled up into a more abstract metric, Effective Health (EH). An example might be you have 100 health, and you reduce all damage taken by 25%, so your EH is actually 125 (and the enemy is still doing "10" damage per round, but you see it reported in-game as 7.5ish). Resistances, absorption, passive self-healing, all tend to get rolled up into EH. EH basically makes it easier to calculate TTL over any given period of time.

Let's Make This Concrete...

When you look at a game like Final Fantasy Record Keeper, most heals are capable of taking a character from nearly dead to full health in a single cast, but you have very few AoE heals, and none on demand. This makes single target damage easily healed through, but AoE damage incredibly powerful.

To threaten the party, the designers need to either have attacks that burst a character down from near full to dead, preventing healing entirely, or blast the party with constant AoEs such that the healer can't heal targets fast enough. In theory, with 1/5 characters healing, you'd only need to do 1/5th of each character's health as an AoE each round and you'd kill one character after 5 rounds (but the other 4 would be in some in-between state).

What also seems to happen is the designers present you with groups of enemies, all capable of taking out about half of a character's health, and they target random characters. Once in a while two (or even three if you're unlucky) will target the same character, bursting them down to 0 in a single round. Or over two rounds bring two characters to half health on the first round, and you heal one character because that's all you can do, and the other character gets randomly targeted and killed. You lost a character on a coin flip.

So the only way to combat that randomness as a player is to look into mitigation: Shellga, Protectga both increase resistance/defense; and other abilities that reduce the damage your party takes by 40% (!!!) for a good 20 seconds or so. The designers designed themselves into a corner with this, because to threaten a party that's capable of increasing their EH by 40% for most of a boss fight means they have to up the damage severely, so anybody not using that mitigation (or a party in the first round of combat) will just get wiped by the first AoE attack.

World of Warcraft had/has this problem, as well. When you look at a 20-25 player raid, with 4 healers, and said healers have AoE mitigation cooldowns they can chain back-to-back, the designers have to account for that in their raid design. The result? Thok in Siege of Orgimmar, where quite literally you were expected to chain together these cooldowns to extend Phase 1 of the fight as long as possible. In a 10-player raid, if your composition didn't have decent cooldowns to stack, you were probably in for a very difficult time.

Compare that to the other direction WoW has gone in the past, where you feel like you're bailing a sinking rowboat with a coffee mug. Limiting the throughput such that without mitigation, your party's damage taken will definitely eventually overcome your healing throughput. Toss in avoidable damage to reduce the party's TTL if they're bad at getting out of the fire, and you actually have a recipe that's mechanically close to what I like to think was interesting and fun, but I know the general populace's reaction to it in Cata (and during the Warlords beta) was pretty negative.

Now, at first blush FFRK and what I view as WoW's "ideal" model are actually rather close. The difference, however, is WoW incoming damage is significantly more controllable. Between having a tank, and the existence of avoidable damage, player skill can make damage far more predictable than in FFRK, which means you're no longer leaving character deaths to a roll of the dice. That unpredictability means that once in a while, RNG truly just hoses you. Despite what people like to claim in WoW, it's actually pretty rare that a fight has RNG elements that actually hose you.

Another interesting model was Dragon Age: Inquisition, where they basically (mostly) removed healing mid-combat entirely. They had shields, and armour--which was just a tank variant of shields--so you could anticipate incoming damage and mitigate it (and shields eventually dissipated over time), but once you lost health in combat, it was gone. Well, almost. You also had like 8 party potions that healed for half health or some such, so you had very limited, consumable healing as well. 

I thought removing most in-combat healing worked out relatively well. It limited encounter length, and unlike in WoW where eventually you could just have nigh-infinite mana and crazy throughput, it meant that your tactics mattered even more. Health actually felt like a resource you could spend in lieu of tactics.

What's Ideal?

Too much healing, and to threaten the players, you need to spike them into oblivion. Too much mitigation, and you need to ratchet up overall damage in general and spike them into oblivion. Not enough healing, and players might feel like healing doesn't much matter, unless you give them other tools to compensate.

Ignoring the folks who rather like just spamming a couple heals and playing whack-a-mole for a moment, there might be a sweet spot in the middle. For me personally, anything that gives the players more agency to determine their own fate--the ability to actively avoid damage, cooldowns to reduce incoming damage for a very short period of time so they can't be chained/spammed, some active self-healing--versus just...chain-casting your biggest heals/mitigations seems like a winning proposition from a "interesting mechanic" perspective. It also means that players can feel good about using their abilities to avoid death rather than designers having to just brutally overpower their own mechanics to actually threaten the players.

But even then, finding the right balance is still pretty tricky, and I'm still not sure quite where it lies.
#GameDesign, #Healing


  1. "so your EH is actually 125 (and the enemy is still doing "10" damage per round, but you see it reported in-game as 7.5ish)."

    Actually 133.3 (repeating, of course). If you have 150 actual HP, then it's the difference between taking 15 hits and 20 hits.

    "Despite what people like to claim in WoW, it's actually pretty rare that a fight has RNG elements that actually hose you."

    Indeed. Like, I can think of maybe 2-3 fights, ever (or at least since BC) where that would be true.

    "Another interesting model was Dragon Age: Inquisition, where they basically (mostly) removed healing mid-combat entirely."

    I wish they had actually stuck to their "guns" on that -- but then the prevalence of Heal on Kill, Heal on Hit, Guard on Hit, etc gear/mods drastically altered the model.

    "It limited encounter length"

    It's more than just that, though -- you'll note you don't heal OUT of combat either. The point being that if you can easily heal out of combat then if you survive one fight with everyone at 1% HP then you can heal back up before the next fight without spending a potion. Health was supposed to matter as a *long* term resource too.

    The equivalent in WoW would be sitting around after a pull while the healer topped people off and then drank to recover mana. Unless there's a timer forcing you forward, it didn't matter whether the healer was at 100% mana or 0% mana at the end of the trash pull (or what the relative health levels were of people in the party).

    1. Yeah, I'm a bit sad that the long-game doesn't exist very often in games these days. It's all about keeping everyone/everything fresh for each encounter. For a game like WoW, where it's mostly about those boss encounters, it makes sense. But for a game like DA:I? Yeah, sticking to their guns would've been nice from the little I played.

    2. You should try the Swordflight series for NWN, it's much more in line with the long-game.

      I don't even really care which way they "go" on DA:I...I just wish they had been consistent throughout the whole game. But Bioware is known for being terrible at balance/mechanics anyway.

      I haven't even bothered playing the DA:I campaign, just MP, with that being a decent chunk of the reason. Everything I've heard says that once you get past the first like, 20-30% of the game stuff becomes stupidly easy on Nightmare, precisely because you get things like specializations and new crafting materials that Bioware doesn't account for (avoiding spoilers here).