Darkest Dungeon - “Git Gud” as a thought-terminating cliche

by Wholesome Rage | 5 September 2018

Like many my age, my first real career interest was in video game design. Unlike many it was because I loved reading game dev magazines and studying the walkthroughs that got passed around in the early 2000s. The ones where you knew it was going to be a good guide because they’d put the effort in to make an opening that looked


Full body nostalgia.

It’s what got me interested early in the questions of what made things fun, and why other aspects were grindy, and why you kept playing games you weren’t enjoying. What made things feel fair or unfair, and what were the game’s intentions.

I actually got my first creative industry apprenticeship at a game studio, which is how I learned I wanted to be a writer more than anything else. None of my code got used for that project, but a lot of my jokes for achievements and loading screens did. So I found those questions more interesting to ask about movies and novels.

But still. It’s a topic I still have a lot of latent interest in, trading in my magazine subscriptions for online blogs like Fortress of DoorsDesigner Notes and Design of Dragons.

It’s an interest I’m usually pretty quiet about, mostly because it’s too much on the Wholesome side and not particularly heavy on the Rage.

25 hours of playtime sunk into Darkest Dungeon changed that though.

Let’s frame this as a problem. After about 2014 there became this trend in gaming that all criticisms of a ‘hardcore’ game was to be met with ‘get good’. That a bad experience was only logical because of bad play. This is infuriating when you look at a lot of the cases of what constituted ‘optimal play’; Narrow playstyle options, usually involving a specific kind of ‘cheesiness’. That is to say, playstyles that seemed to break the spirit of the game but were entirely within the rules of it. 

As a result, I have a special rage-button in my heart for any game that excuses bad game design as an intentional ‘difficulty’ feature. It’s like seeing spiciness options on the menu of an Indian restaurant: If you have the option, odds are high that they haven’t made the dish ‘properly’ in a way that got that spice. It means they’ve made a not-spicy variant of that dish, and then add chili powder to taste when they serve it. That chili adds nothing to the flavor of the dish, it’s just to disguise the fact that it was improperly made.

Darkest Dungeon uses a ton of chili flakes to try to stop you noticing it lacks a lot of flavour. And if you try to explain why it’s bad, the response is always; “I guess you just can’t handle spicy food.”

So let’s step into the kitchen and see how this curry was made.

First, I’m going to explain the difference between ‘difficulty’ and ‘complexity’, as they’re what get mistaken for each other in these conversations. A more complex game is always more difficult, but a more difficult game isn’t always more complex. The best games often have the highest complexity they can manage without forcing difficulty.

There are plenty of ways, though, to increase the difficulty without increasing the complexity. When this happens, and the audience conflates these two things, is what you’ll find in a lot of the ‘hardcore’ genre games that aren’t actually fun to play. An example of a mechanic to increase difficulty without complexity would be ‘infinite spawner’ enemies, enemies that endlessly produce full-health minions. Infinite spawner enemies in games like this are almost always unfun ways of making a difficult enemy that isn’t particularly complex. Darkest Dungeon has several, like the Miller, the Collector and the Necromancer boss enemies, and slimes as regular enemies, but another example that comes to mind is the first-floor boss in the Binding of Isaac that is Ragman.

They’re a lazy way to add difficulty, similar to shooter games that only make your guns do less damage instead of improving the enemy’s intelligence and playstyle. It’s more difficult, but it’s rarely more interesting, like adding speedbumps to NASCAR.

Put another way, a good curry is always spicy, but a spicy curry is not always good.

One of the most important features of a roguelike, which Darkest Dungeon is, is how it handles random elements and randomization. Darkest Dungeon uses ‘true’ random number generation and that’s a bad thing, because true randomness doesn’t feel random. In fact, it feels awful.

Consider XCOM for comparison, usually because it’s the example everyone brings up when I make these complaints. XCOM remains one of my favourite games ever made, and it’s because they actually solve these problems in invisible ways. XCOM, on most difficulties, has a hidden ‘buff’ to it: If you miss a shot that was over 50% likely to hit, you get an invisible stacking +15% bonus, added each time you miss until it caps at 95%, and then resetting once you do hit.

As a result, you’ll miss two 70% shots in a row far less than the 1/9 times you should actually expect. Because the human brain doesn’t actually think of the second shot as an independent 70% chance, it thinks of the two events as being linked – if the odds are missing both is only 1/9, then the second shot should feel like it misses only 1/9 of the time, not one in three.

The reason for this is because true RNG feels bad. It upsets our monkey-brains that think in patterns and heuristics, not actual numbers.

So when I say ‘Darkest Dungeon uses true RNG’, understand that what I’m saying is that Darkest Dungeon’s random elements already feel awful on their own. It then exacerbates this by committing the most cardinal sin of game design:

It stacks random elements on top of random elements.

The purpose of random number generation in a game is to prevent a game from being ‘solvable’. By having variations in every outcome, no one solution is going to work every time. No matter how complex the game might otherwise be, high level play will always break down to rote memorization otherwise; Just look at high-level chess. The reason grandmasters are capable of playing games in their mind, or multiple games at once, is because they have all those board states memorized.

However, randomness acts entirely outside player agency. No amount of skill or talent or knowledge of the system can affect it directly. This is a fine and good thing on its own, because it mitigates those solvable states which is the difference between playing a game and filling in a spreadsheet — EVE online players are scratching their heads in confusion at that one. It also means that when truly random elements are capable of severely punishing you, it feels completely arbitrary. 

When I complain about Darkest Dungeon, it’s also something like Magic or Hearthstone for this reason. In deck building card games you can be punished by drawing the wrong cards in the wrong order. Sometimes that is just bad luck, and it sucks. Usually, though, it’s a sign that you need to plan your deck better or an indication that you’re using a ‘greedy’ strategy that relies on luck too much to cook off. Your bad luck is actually the result of bad decision making.

Thus the random element can be mitigated by better preparation and knowledge of the game, and really if you feel unlucky it could be a sign you’re playing badly. This, they explain, is the reason behind my dislike of Darkest Dungeon; an assumption that luck mechanics can inherently be manipulated by skillful play. This is unfortunately a conflation between difficulty and complexity. 

The truth is that, yes, when a game is designed well that will be the case. But we’re so used to seeing that done well in games that it’s forgotten that it’s not an inherent truth. 

One of the ways this is mitigated is by having only a single layer of randomness at a time.

In Magic or Hearthstone it’s the fact that cards are drawn randomly from your deck, but aren’t randomly played from your hand and your deck itself is player crafted. The playing cost of the cards isn’t random. This means that, while what you get to work with turn-by-turn is random, the deck construction and the cards played are entirely skill based and in your control. This means that you get a much better sense of what you did right and what you did wrong in a game.

XCOM is similar in that, while your actual shot percentages and damages are rolled, you at least get complete control over your whole team’s actions. You can choose the order they shoot, move, break their turns up however you want. So, even when something goes wrong because of a bad shot, you can plan the rest of your turn around that and mitigate the damage, or structure the turn around the possibility of those failures.

Darkest Dungeon differs in that it stacks random chance on top of random chance. Spin a wheel to roll a dice to flip a coin. It’s obviously intentional in that it’s designed to prevent players from being able to make informed decisions, and in that it succeeds.

The problem is that every extra layer of randomness feels more and more arbitrary, less and less like bad occurrences are the results of poor play and more like the results of bad luck. When it doesn’t feel like it’s your decisions that got you smacked down, then you’re playing slots and not poker. 

On one run through of the Weald level my party, who is steamrolling, runs into a giant. I have an extremely dodge-specialist Houndmaster in my second rank, and after a few rounds dancing the giant’s blows, he gets critically hit by a swing, gets knocked down from max health to no health in one strike.

This happened in part because Darkest Dungeon is an incredibly ‘swingy’ game by design, massive bursts of health gain and loss as a proportion of your characters hit points, so the game can switch from safety to absolute panic in a moment. To facilitate this there’s a mechanic called ‘Death’s Door’ which means that you have to be hit again at zero health to be permanently killed. Every time there’s a chance for your character to resist. This allows the game to consistently drop you to the danger zone without feeling overly cruel or frustrating. 

The Houndmaster fails his move resist roll and gets knocked to the backline by the attack, and the next enemy’s turn starts, an enemy that gives negative status effects to your backline troops. The Houndmaster failed his dodge roll, again, and took a 1 damage status effect that killed him permanently, destroying my team composition and forcing me to lose the level, completely undermining the purpose of the Death’s Door mechanic in the first place.

This feels awful. I had a well equipped team, suitable for the level, that was doing extremely well up until that point. My team was almost at full health, and was healing faster than the enemy could damage them. I lost because of a series of stacking bad rolls – The crit, the failed move resist, the enemy turn order, the second hit, and the failed death save roll were all cascading random elements that happened subsequently.

I think it’s also useful to break down why each of those rolls happened. As in, what was the intended effect of all those mechanics that they had to be piled on top of each other to result in that outcome?

Death’s Door I’ve already covered. It’s the games way of keeping big swingy combat without killing your people off too often. But why did the status-effect enemy kill my Houndmaster? To make Death’s Door riskier, to make non-damaging attacks capable of killing someone. This is done to make it more difficult to ‘farm’ healing in combat.

It’s an odd problem to have, but Darkest Dungeon doesn’t let you heal between combat events, only inside of it. As a result, most combat is spent whittling the enemy down to a final weak enemy, and then spending as long as you can ‘grinding’ healing so you don’t get floored at the start of the next combat. 

This is what I mean when I say it’s adding spice without flavour. That’s not interesting gameplay, it’s tedious. It exists solely to make the game more difficult, but it also creates a situation where you will prolong every combat past the point there is any real sense of risk to it just to heal your team, because being able to heal your team outside of combat would be ‘too easy’, or get rid of the intentionally crafted aesthetic of being worn and ground down. It creates the perverse incentive, though, of having to stay in combat to heal.

I understand that, and the game designers apparently did too, because their solution was to fill dungeons with parties of significantly weaker than usual monsters between the beefy big hitters to give you that opportunity to farm heals, and maybe get a little bit more treasure. While this is a working solution, it prolongs the dungeon experience by providing combat that feels rote and mindless, which also severely undermines the point of preventing heals-outside-of-combat making you feel hopeless when there are groups of enemies that exist solely to be virtually risk-free.

“Overconfidence is a slow and insidious killer”, the game warns you after these encounters, so again the developers are at least aware of the power-trip.

So that’s the final result of why my Houndmaster had to die. A mechanic that’s designed to allow the game to arbitrarily crush you down to zero hitpoints without feeling too cheesy about it, having a bad synergy with a mechanic that’s designed to prevent you from farming combat and abusing the death save, to justify the fact that the game doesn’t let you heal outside of combat even though that makes no sense, in or out of game. This also feels terrible because if the two attacks had come in the opposite order — the 1 damage attack followed by the crit — the chance of that character dying would have been 0. Another bad luck dice roll.

So what would the alternative be?

Well here’s my pitch. By breaking down what the game’s going for, we can reverse engineer something that keeps the same desired goals — being worn down over the length of a dungeon, the constant feeling that you could be iced any moment, etc. — while making more fights fun and challenging, preventing this stack of randomness teetering on top of itself, forcing heal-farming in-fights, and forcing every four-person team composition to bring one of the game’s two only viable healers.

First, healing outside of combat. This is the frayed end of this whole knot. There are plenty of characters that have very minor healing abilities, or abilities that heal a bunch when the characters set up a campfire — some combats give you rest breaks — like the Plague Doctor and Arbalest and Antiquarian. Others, like the Leper and the Houndmaster, can heal themselves. 

They’re completely impractical most of the time. You need those characters doing other things, because they’re going to be healing for 2-3 health points at a maximum when your enemies are usually hitting for 5-6 points at a minimum, and you can only pick four of their eight abilities to bring into combat. The healing ability is usually a wasted slot. If you had out-of-combat healing with those characters, then they’re much more reasonable, and it has the added benefit of making no-dedicated-healer team compositions practical and viable. I’d love to have a proper attempt at a glass cannon run.

How do we make this an interesting choice though? I’d add an ambush mechanic. Better out-of-combat healers will be able to do it faster, quieter, and have a lower ambush attempt.  They already have this function in-game for character’s being able to disarm traps.

Thus the out-of-combat heal isn’t always free, and is in and of itself a risk/reward gamble. Giving players more meaningful decisions to make is the true end goal of game design. Two choices that aren’t always right for all situations, and making players decide which is better moment to moment.

So what does this fix? Well, now you have a wider array of viable team choices, you remove the need to ‘farm’ healing and prolong combat when it’s past the point of any risk, and you no longer need to have quite so many ‘padded’ fights intentionally designed for healing grinding. All without making the game ‘easier’. Brilliant

Hell, there are a lot of in-combat abilities that take a turn to use that really shouldn’t, because an out-of-combat toggle isn’t an option. The Jester’s Inspiring Tune — one of the game’s only consistent ‘stress’ heal —  comes to mind. Being able to run these from the corridors and choose them in advance would be a far more interesting decision in a part of the game where there are currently very few meaningful choices to make. Inspiring Tune passively reducing your stress, but increasing your Wandering Monster chances? That’s a genuinely difficult decision.

This means we lose an element of being ground down over a dungeon, surely? Something that the designers of Darkest Dungeon find important to their experience.

Sort of. There’s now a void in the developer’s intentions that this was meant to fill: The feeling of being ground down over a long dungeon. How can we better fill it?

There are two other mechanics that can fill the void already implemented: Disease and Stress.

Actually, let’s talk about how stress is also currently an unfun mechanic, even though I think it would be the best candidate to fill the intended role.

Stress is like a second health bar, except it’s much harder to heal. Only the Jester character can reliably treat it at lower levels — and right now only in combat — and he’s not really optimal for most teams so you’re hardly ever going to bring him. As a result, it’s a serious problem to deal with in most dungeons. Usually the bar will go up slowly over time from traps, the encroaching darkness, from taking critical hits from enemies, and creepy items. This means making taking rest stops to cool off, though you have a very limited amount of those in each level so you need to time them well. This already fills the purpose of something that progressively grinds you down over the course of the dungeon, there’s no need to double down on it with the _actual _health bar as well.

So far, so good. A second bar that goes up slower over time. The problem is that, currently, the penalty for a full stress bar is disproportionate to our desired effect.

At 100 stress your character functionally ‘snaps’. They mess up their move order, they can stab themselves or their teammates for no reason, they can simply refuse their turn. They’ll do this for the rest of the level they’re on, and every turn they’ll add around 5-10 stress to each of their companions for having ‘snapped’, which quickly pushes everyone else over the line too… then they all add about 15 cumulative stress a round to each other, until they have heart attacks and die. This stress effect is permanent for the level.

Let’s compare that to XCOM’s handling of a panic mechanic for a moment. When a squad member in XCOM panics you lose control of them for a round, at which point they’ll make a reasonable action. Take the best shot they can from their position, hunker down or take a reaction shot. It’s not ideal or optimal, but it doesn’t destroy the entire level for you.

There are some enemies that attack your stress rather than your health. They are the worst enemies in the game and I hate them. I don’t even have anything clever to say about that, but the amount of times I’ve been punished by a random combat that opens with a character taking 40 stress damage and immediately ‘snapping’ because his stress previously hadn’t been worth risking the rest…? Not fun.

The second mechanic would be diseases, which are status effects that can only be cured at the end of the level, in town. They can do some pretty significant, nasty effects to a characters statistics which means they perform worse the longer the level goes on. Making diseases more punishing, easier to get, but easier to treat, would be a fantastic and thematic way to fill the void of wearing you down as a dungeon progresses.

Finally, there’s mitigation, which is the real reason the Stress mechanic is currently too all-or-nothing. That is, the player’s ability to make strategic choices to deal with the results of bad luck. If you take a few bad runs of stress or health damage in Darkest Dungeon, there’s very little you can do outside of combat to treat it. Backtracking is almost exclusively never worth it, and you might simply not roll the right random events to backtrack towards anyway.

This isn’t attributable to it being a Roguelike, because all good roguelikes have their way of dealing with this. The roguelikes Slay the Spire, Faster than Light and Enter the Gungeon all do this by having an in-level shop mechanic. 

They give you a chance of a random drop from enemies, and usually a certain amount of a static currency. Scrap in Faster than Light, coins in Slay the Spire and Enter the Gungeon. That currency can be spent mid-run on shops. Shops are great because they give you exactly what you want, but you have to make a decision of what to spend your very limited currency on.

It mitigates the random element — random weapon drops in Faster than Light, getting random keys to get the guaranteed chest items in Enter the Gungeon, and the semi-random card selection for Slay the Spire — with an element that allows you to make a priority choice. What is the best value for that currency.

Darkest Dungeon doesn’t have an in-level shop equivalent. There is no strategic choice to be made, no decision besides ‘fail the level’, when the doom spirals occur because of a bad roll, or a bad situation. There’s no high-risk high-reward strategy to be made outside of combat that would right a sinking ship.

Fans of the game will say of course. That’s intentional because you’re meant to feel the creeping darkness, the doom, the pull of the unrelenting death spiral. Of course you’re meant to arbitrarily lose characters sometimes, because your people are supposed to feel expendable. Of course it shouldn’t happen too often though, or else you need to “get good”. 

That’s ignoring that these problems have been solved in other games, Firaxis have put a great deal of literature out on the subject, and they make for an experience that is still this difficult and challenging, but also more robust, enjoyable and complex as well. 

The problem with Darkest Dungeon, then, is that it can’t add these more interesting and complex mechanics because it’s already too difficult, and its difficulty is arbitrary and artificial. There is no room for substantial growth, because it’s already been taken up by bandaid fixes, lazy implementation and bad ethos.

I love spicy food. I can handle the heat. But you can’t just add hot sauce to mall food court butter chicken.

Darkest Dungeon is just lucky that I have a particular craving for Indian food right now.

« Previous Post

Next Post »