First, a question: what is difficulty?
In the context of a game, you could probably describe difficulty as consisting of two components: the leeway which the player has to make deviations from optimal play, and the degree of consequences impacting the player’s ability to achieve a favourable end-state when they make one of those deviations. In the bluntest terms: how easy it is to fuck up, and how badly each fuckup hurts you.
These may seem like very similar concepts, but they are in fact, very different, and most games will utilise them in combination one way or another in a sort of double-spectrum: the actual mechanical difficulty of a game comes from the result of the combination of these two factors, while the perceived difficulty of a game may depend on only one of these factors, in a way which means that a famously “difficult” game may actually be deceptively easy in the long-run.
Take, for example, Dark Souls and the subgenre which it has popularised. Dark Souls – as everyone who has played or heard of it is well aware – is a very difficult game – except it’s not, not really. It’s absolutely true that one wrong move in combat in a Soulslike game will heavily punish the player in the moment, with missed parries or attacks from basic enemies costing players a large portion of their health – and with hits from bosses possibly killing a player character in a single hit if it lands. However, this just means these games allow for few deviations from optimal play before punishing the player in some way. While that is one of the integral pillars of difficulty, it is also mitigated by the way the other pillar is implemented. While the player will often hit a lose state (seeing the famous ‘YOU DIED’ splash screen) that lose state has little meaningful long-term effect. In most Soulslikes, it means they lose a few minutes or even a few seconds worth of unbanked currency, and whatever progress they made form their last save point.
That might seem harsh as a death penalty – but consider the fact that this is also the death penalty for hitting a lose state in most games: you have to reload from the last point you saved at, and lose any resources you gathered and any progress you made since – except Soulslikes will also give you the opportunity to recover at least part of that progress by retrieving your lost currency. In fact, this means that while the first pillar of difficulty does in fact make Soulslikes harder, the second makes it easier.
This is actually an incredibly clever design for a simple reason: players like overcoming difficult challenges in the moment, but they don’t like the tedium of the process of repeatedly working up to those challenges. By making it easy to hit a lose state, Soulslikes establish their combat as a difficult challenge – especially in regards to boss fights. However, in making it easy to recover what’s lost when hitting that lose state, it minimises the amount of tedium which a player might endure in recovering from their loss. Players get the rush of beating a difficult challenge, without necessarily suffering the tedium of having to repeatedly prepare for that challenge over and over again. I think it’s rather telling that when players want to increase the difficulty of Soulslikes, they often use self-imposed challenges which require them to avoid hitting a lose-state or levelling up (which both banks currency and makes a character stronger) – taking the relative ease of “recovering” from a mistake and making it more difficult.
As a result, while Soulslikes are usually perceived as extremely unforgiving action RPGs, they could just as easily be defined another way: as extremely forgiving Roguelikes.
Roguelikes – or games based on the very early 1980 dungeon crawler Rogue – are usually turn-based top down RPGs with a similar degree of in-the-moment mechanical difficulty as Soulslikes. One of the most infamous, NetHack (1987) not only made it very easy to die in combat, but also very easy to die out of combat – from food poisoning, disease, or something as simple as trying and failing to mount a riding animal before slipping and cracking your head on the floor (don’t ask me how I know this). The difference between games like NetHack and Dark Souls is that in the former, hitting a lose state meant the loss of your character entirely. You had to start again from the first floor of the dungeon with a fresh player character at level 1.
Naturally, this combination of low tolerance and high consequence meant that traditional Roguelikes have an extremely punishing difficulty level, to the degree that most players find them alienating. There’s a reason that Roguelikes never achieved the same level of mainstream commercial success as Soulslikes – and why most modern conscious iterations on the Roguelike model – so-called “Roguelites” – incorporate a high degree of progression in between “runs” which mitigate a lot of the consequence of hitting that lose state.
So, what about the other difficulty models? High tolerance and high consequence, or high tolerance and low consequence? Well, those have some very distinctive exemplars too.
Anyone who’s even vaguely familiar with my gaming habits know that I play a lot of Grand Strategy mapgames: games where you control a dynasty or a state through periods of history or pseudo-history, interacting with allies, rivals, and generally affecting politics on a grand scale. By representing the course of a body of people or a political form and requiring so many decisions to be made, this genre inherently provides a lot of tolerance when it comes to sub-optimal play. You can make a lot of mistakes (and I do) and still manage to get somewhere close to a desired outcome, assuming that desired outcome isn’t something ridiculously unrealistic (like conquering the world as Albania or something. At the same time, because these games require so many decisions, a single “run” can take dozens of hours to complete – meaning that if you do screw up enough to reach a loss state, you may have lost hours or even days worth of progress. This is an example of a high tolerance and high consequence system.
High tolerance, low consequence systems, on the other hand, are often embodied by what we now often refer to as “cozy games”, experiences where a loss state isn’t just difficult to achieve, but nearly impossible, meaning that every instance of sub-optimal decision-making can be easily recovered from – and because there’s no loss state, almost every mistake can be recovered from eventually. As a result, success can be measured in the degree to which a player succeeds, rather than a binary state of success or failure: a mistake is a setback in the most literal sense, not a loss. This allows the player to continue to set goals and strive to achieve them, while relieving them of the stress that comes with the possibility of failure, leading to the reason why cozy games are so cozy in the first place.
The ability to set goals also plays into a third pillar of difficulty, one which only applies to narrative-focused games with branching storylines. Cozy games in particular tend to provide players with certain mutually exclusive success states. This usually manifests as a choice of different romantic partners (I haven’t seen any cozy games which let you form a polycule – but I’m sure there’s probably one out there). This means that one player’s “win” condition can be different from another’s, and an additional element of “difficulty” comes from choosing the particular win state that particular player wants.
This is something I explore a lot in my own work. By presenting multiple, mutually exclusive win states with differing narrative outcomes, I can separate the binary state of win and loss into a graduated scale of greater and lesser wins. Thus, the player has the ability to set the difficulty themselves in a way, by consciously choosing which ending to aim for, with the most positive ending (often referred to as the “Golden Ending”) requiring the most optimal decision-making, while a suboptimal – but still positive – ending makes allowances for more mistakes. As a result, the player can determine how many mistakes they are “allowed” to have, by essentially choosing which win states are actually acceptable to them. A player who wants a challenge can decide that only the Golden Ending qualifies as a successful run, while a less fastidious player – or one who simply wants a more relaxed time – could decide that any end state not an overt loss is good enough, resulting in an experience far more permissive of sub-optimal play.
The result is an “elastic” system of difficulty which, if done right, will effectively allow the player to set their own level of challenge in an organic and diegetic way. A player who aims high or chooses a particularly difficult goal (as contextualised by the setting and narrative) could be playing at a higher difficulty than one who chooses more humble or conventional goals as their desired end state – even if they are both playing the same game. The potential result is a work which is both accessible to those who want it, and challenging to those who want to tough it out.
What I’m trying to say here is that difficulty is a tool in your toolbox, one that lets you finetune the experience that you want your players to have. Obviously, I don’t know precisely what that experience is, but now that I’ve given you some idea on how the concept of difficulty comes together, you might have some ideas as to how to work it out for yourself.
Paul Wang 