Tuesday, October 11, 2011

Understanding Difficulty

Although we frequently have discussions about difficulty in games - is it too hard?  which parts did you have trouble with? was it too easy and therefore boring? - we rarely direct our attention to the different fundamental types of difficulty which make up our experiences and colour our perceptions of the challenge a game provides.  In this article, I'd like to go over a few of those most basic types of difficulty as well as the problems associated with implementing them, as well as bring out that it's often not just the sheer challenge of a game that matters, but the nature of those challenges that matter.

Trial and Error

The first, and most obviously identifiable type of difficulty that we find in games, and by far the most common, is trial and error.  Put simply, trial and error revolves around getting the player to perform a task, either through experimentation (i.e. "I don't have anywhere to go, maybe I'll try this") or outward suggestion (i.e. "these are your orders, soldier, now move out!").  At least theoretically, the main difficulty this presents to the player is that the degree of challenge (types and numbers of enemies, for instance) will always be slightly higher than what the player is comfortable with, meaning that he or she will have to rise to the occasion in order to come out on top, either by trying out new tactics, by taking greater risks, or through sheer force of will and dumb luck.

As many of us can attest, trial and error difficulty treads a very fine line.  Typically, too many failures, and players will become frustrated, while too many successes and players will feel as if the game isn't going hard enough on them.  The main issue with this, aside from basic balancing, is that different players have different thresholds for difficulty.  Whereas a more casual player who's just enjoying a game for its story will find that more than the occasional death is a turn-off, the hardcore player who plays on the "insane" setting will want to be challenged at every turn and made to work for every single victory.  Ultimately a developer might run into a situation where they're balancing not just one, but three or four versions of the same game, due to the different needs of different players.

Of course, pacing is also a chief concern by and large governed by the ebb and flow of difficulty, usually of the trial and error nature.  The player needs to have portions of the game which fly by quickly and without too much issue, breaks in combat to absorb the world and feel unchallenged, and nail-biting experiences that are tense and have a feeling of urgency to these.  Building these into a game when taking different gameplay preferences into consideration is a difficult process; after all, while it can be easy to balance a single encounter out to give the player the desired experience, doing so within the context of a full game is another thing entirely.

Adaptive difficulty settings are one way to get around this problem.  On the most basic level, this will typically change the amount of resources (health, ammo, etc.) provided to the player, as well as the proportion of powerful versus weak items based on the player's performance (i.e. more "full heal" pickups if the player is struggling).  This feature is actually extremely common in games, either because developers want to avoid providing separate difficulty levels (a poor decision in my mind), or because players have a curious habit of selecting difficulty levels that aren't appropriate for them (everyone has a different understanding of what "normal" should be).

Screw Alyx, these crates were my best friends in Half-Life 2.  Turns out the reason was a bit more calculated than my platonic love of all things boxy.
 Adaptive difficulty can be both explicit and hidden from plain sight.  Prey, for example, has adaptive difficulty as a toggle option in the game's options screen, and so it can be disabled based on the player's preferences.  Half-Life 2, on the other hand, while providing three difficulty settings (easy, normal and hard) also has a layer of code dedicated to analyzing the player's progress in the game, level of resources, the ease at which certain encounters are completed, and so on; the game will then adjust the items enemies drop, the amount of resources available in breakable crates, and so on in order to make sure the player is always kept on edge by having "just enough" health and ammo to get through an encounter, but never quite enough to feel completely safe or fully-loaded.  Other games will implement it in still subtler ways, like allowing the player to finish off a tough boss monster just a little bit more quickly than normal if the player's death is imminent, creating a dynamic feeling of getting through by the skin of his or her teeth.

The biggest issue for me with adaptive difficulty is that, when left as a built-in feature that can't be disabled, it removes control from the player's hands.  Although I'll usually take an entertaining and engaging experience over one that's simply difficult for the sake of difficult, I also fully understand that some players don't want hand-holding provided that they explicitly ask for it.  Furthermore, adaptive difficulty can also lead to a feeling of predictability and sterility, without a hand-made feel to encounters (which was a major source of criticism for The Elder Scrolls IV: Oblivion).  To this end, I feel that adaptive difficulty is best left as it is in Prey - a toggle switch in the options menu - or specific to a difficulty level, with the hardest mode taking off all assists, which mitigates the problem of too much challenge by allowing the player to rationalize it as his or her own choice (i.e. "well, I picked hardest, I should have known it would be too much for me").

Endurance & Attrition

Another way to test the player focuses on the long term rather than the short term.  All forms of endurance, at their most base level, revolve around resource management, with the player given a limited quantity of a valuable or vital item, its distribution carefully controlled.  Resources are controlled in three main ways in just about every game:

  1. "Random" drops.  It's quite common for enemies to part with valuables when defeated, or for the player to uncover supplies in crates, chests and so on.  By tinkering with the tables that control those supplies, based on difficulty, the player's progress, the amount of resources the player already has, and the player's level of ability, character level, number of party members/companions, and so on, difficulty can be precisely controlled and monitored in order to provide a degree of challenge.
  2. Attrition rate.  Depending on the game, the rate at which a player burns through supplies can be highly variable.  For instance, in a shooter, going up against a tough boss monster might not consume too much ammunition, but may consume a huge amount of health.  Conversely, going up against many smaller hordes of enemies will end up with a player ill-equipped to proceed, but chances are, a healthy one.  Learning to anticipate what the player needs in order to continue in the game is important.  If a game uses an adaptive difficulty system, this might already be handled, but even so, careful consideration of how quickly the player goes through certain resources will lead to better encounter design and a game that feels more alive and responsive to the player's needs.  Strategically denying certain resources can be just as important as strategically providing them, too, in building tension and pacing the player's progress.
  3. Player ingenuity.  Most common to role-playing games, smart players will often stock up on useful items like potions and ammunition before heading out into a difficult encounter; the duration the player can stay out in the wild before returning to stock up on supplies again is by and large controlled by the player's prior action, as well as whatever the player might uncover during his or her outing.  This is one thing that is hard to control in a game, and frankly, shouldn't be.  Keeping aware of what players can and can't do, and building challenges around that is a good thing, as are systems, such as encumbrance and fatigue, which can provide a soft limit on how much the player can carry.  However, imposing unreasonable hard limits (i.e. "you can only hold three health potions at once") rarely feels like a fair way of managing this.
You don't need to be Arcanum to have compelling attrition and resource management (but it helps).
 Long-term attrition may not be suitable for many games, but looking at attrition in different ways can actually reveal interesting opportunities for mechanics that may go unnoticed with a casual glance.  For example, a puzzle game like Tetris has a strong element of attrition in the sense that the available space on the game board is continually shrinking based on the player's performance, the difficulty level, and which puzzle pieces the player is provided with.  On top of that, game speed is another gradually-depleting resource the player must carefully manage as things move quicker and quicker over the course of the game.  There is a veritable economy of space and time in Tetris, even though there is no health bar, ammunition counter, etc. to speak of.  Recognizing that attrition and endurance can exist as more than just basic physical resources will help flesh out and provide depth to existing mechanics.

"Fake" Difficulty

A subset of trial and error difficulty, what I'll term "fake difficulty" here is something which is actually quite common in the games industry, but depends a good deal on the genre in question.  Fake difficulty is a fairly broad spectrum of difficulty, but in common with all of the various permutations is the fact that they typically revolve around tricking the player or bending the rules of the game in order to provide their challenge - often causing significant frustration and annoyance for players, whether they're keen to those tricks or not.

One of the most common forms of fake difficulty actually fits within the category of adaptive difficulty - namely, it revolves around manipulating the rules of a situation in order to provide the player with increased challenge, usually referred to as "rubber-banding".  The key difference is that while adaptive difficulty works in favour of the player (for example, you'll find 50% more health kits if you're low on health), fake difficulty tends to work in favour of the enemies or opponents.  However, since enemies rarely compete on fair terms with the player, and in fact tend to use an entirely different set of rules, this usually means that the bonuses given to the player's opposition fall into the realm of super-human - increased speed beyond normal limits, temporary damage boosts, the ability to negate the player's own abilities when normally they can't, and so on.

A great (and persistent) example of this type of difficulty can be found in Mario Kart - in fact, the series is somewhat infamous for it.  While the goal of the game's rubber-banding is to provide a tense and exciting experience for the player, making sure that each race is as close a finish as possible, and that enemies are able to always keep players on their toes, in the long run, or for more experienced players, this form of difficulty tends to only breed contempt.  While the illusion created is often enough to fool players who are of a lower skill level, as the effects are much more subtle and can often work in the player's favour, when that same system is put up against players who are able to make a mockery of even the high difficulty levels, the computer is forced to go to incredible levels to try and keep up with the player, to the point of blatant cheating, gaining items and abilities far in excess of the player, and even defying the laws of physics (or whatever analogue exists in the Mushroom Kingdom).

Another form of fake difficulty that rears its head is that of the false challenge (which I admit, sounds a little redundant).  In the false challenge, the player is typically asked to perform a standard feat - defeat some enemies, race to the finish in the allotted time, etc.  However, what starts out as a relatively routine task quickly turns out to be an extreme test of reflexes and ability, as the player is beset with all manner of unpredictable obstacles, traps and powerful enemies.  The key thing is that in all of these situations, the player is caught off guard, and unable to sufficiently prepare.  Usually, this results in a quick and frustrating death, as the player likely felt he or she was successful up until that point.  Worst, usually, the only way to surmount this type of challenge is to try it again, often from the very beginning of the sequence, armed with the foreknowledge of the hidden challenge ahead.  When these are compounded one after the other, it can lead to rage-inducing moments for the player.

You can't see it here, but in 0.25 seconds, the driver in the blue car is going to develop a sudden case of sociopathy and swerve straight into the player's bike.
One game series which is notorious for this is Grand Theft Auto.  While the game's mission-based structure suggests that the challenges faced are relatively self-contained and straightforward, it's very common for the games to prey on the player's expectations in the worst way possible.  One example from Grand Theft Auto: Vice City I frequently cite is a race sequence where the player has to reach a number of checkpoints in a set time.  No big deal, right?  That would be the case, if it wasn't for the fact that other cars, trucks etc. are scripted to pull out around difficult corners and immediately as the player passes by at full speed - the player is almost guaranteed to hit these cars and ruin his or her attempt outright, unless he or she is able to slow down and let them pass instead.  This just doesn't happen once, but close to five or six times throughout the race, meaning that even if the player does everything right, there's still a huge statistical probability that he or she will fail anyway, solely due to the designers pulling a fast one.  A similar occurrence can be found in Max Payne, where enemies are scripted to throw grenades at the player at certain triggers, and these are literally impossible to avoid without prior knowledge.
Suffice is to say, fake difficulty, no matter the variety, isn't fun for players, even if it's built into the game with the best of intentions.  Although often the goal is to provide an unpredictable or challenging experience regardless of the player's skill level, more often than not it just comes across as mean-spirited, and at worst, can completely turn a player away from the game by rendering attempts at competition null and void.  Unlike most forms of difficulty, this type is actually best avoided altogether, unless your goal is to make players hate your guts.

Random Number Gods

Although this is typically a type of difficulty reserved for strategy and role-playing games, random mechanics do exist in a wide variety of genres, whether they manifest in terms of how enemies behave in combat, the spread and accuracy of weapons, or whether or not the player is able to sneak by a foe successfully.

I've occasionally seen mechanics based on random elements derided by people, claiming that it takes away from the skill of the player to hinge success upon unpredictable odds.  The key thing to understand about building difficulty out of a random number generator is that challenge is not substituted for "luck", as some might claim.  Rather, difficulty arises as the player is forced to respond intelligently to new developments that aren't entirely predictable - it is the culmination of actions over a period of time that are important, not the individual actions themselves.  Unlike trial and error, which typically tests reflexes and coordination, systems built on random elements test the player's ability to respond to change and to cope with new situations.

As mentioned above, it's also important to mention that random elements are often a staple in all types of games, regardless of whether or not difficulty is provided by trial and error, by manipulation of odds, or, ahem, fake difficulty.  Driving a car in a racing simulation, for instance, there's bound to be some random effect in the vehicle's handling, or on varying types of terrain, even if it's only a small piece of the overall picture.  There is absolutely nothing wrong with this, because usually player skill is able to account for random elements anyway.  More to the point, random doesn't necessarily mean unpredictable - it just means that there can be a certain degree of noise or interference in playing the game, to prevent things from playing out exactly the same way every single time.  Otherwise, when playing Tetris, we'd see the same blocks always become available in the same order, and that wouldn't be nearly as fun to play, as the game itself is based wholly around bringing a degree of order to that randomness.

Unfortunately, building systems out of random number generators, particularly in role-playing games and strategy games, it's easy to fall prey to a problem - not in the mechanics themselves, mind, but in the player's perceptions and understandings of them.  This usually manifests as what's commonly called the gambler's fallacy.  The simplest example is a coin flip.  Even though a coin only ever has a 50/50 chance of landing heads or tails (assuming it's a fair toss), we tend to assume that the 50/50 probability applies to all instances of the event in sequence, rather than the isolated event.  In other words, we form a narrative as we flip that coin over and over again, perceiving each coin toss not as a single incident, but part of a larger whole - and as such, we also tend to assume that prior events have an influence on future events, or, put simply, that the more the coin lands heads, the greater the chances we think it has of landing tails.

Frayed Knights is about as fun as a game with spells like "Exploding Kneecaps" can be, but I often got the sense that the Random Number God was out for my blood.
 In gaming terms, this can be described in the context of a turn-based role-playing game.  A skill might have a 70% chance of success when used, yet we become frustrated when, turn after turn, the skill misses and we end up wasting both our time and resources trying to rectify the problem.  What just happened?  Surely, the game is fudging the numbers!  Well, no, not really.  We assume that, because the skill has a 70% chance of working, it should (or will) succeed seven out of ten times, like clockwork.  This is, of course, not at all the case, as each individual attempt has the same odds as the last, and therefore, it's possible to chalk up a huge string of losses despite what should be good odds.

There's no easy solution for this problem, because you aren't battling the numbers, you're battling player expectations.  Many developers actually get around this problem by instituting measures to make sure that random odds are, in fact, more predictable.  For instance, if I have that 70% chance of success, I might program a clause into the game where it's impossible to miss more than one time in a row - even if ultimately the math is completely off.  That's right, often, the random odds most players feel they rely on aren't actually random at all, but instead manipulated to fulfill the expectations players have.  The irony of all this is that usually the player only ever notices that there's a "problem" if the math is correct in the first place.  Obviously this is a controversial decision, and not everyone will agree with it one way or the other, but in the end it's probably better to fulfill player expectations than it is for those same players to wind up frustrated over what they feel are unfair and incorrect odds.

Presentation is Everything

The header here might draw some flak, but I think that this is a lesson that is very much unsaid when designing games, and yet at the same time one of the most important to learn.  Difficulty, as I've outlined, comes in many flavours and is highly subjective - however, it is also important to recognize that the way difficulty is presented to the player is also just as, if not more important.  Similar to the gambler's fallacy, sometimes it's not a particular mechanic that's the problem, it's the way that players perceive it that's at fault.

Let's take a recent example in Dead Money, the Fallout: New Vegas DLC add-on.  The game came under attack from both players and press alike for what they perceived as a steep difficulty curve.  In Dead Money, the normal endless freedom of Fallout gives way to slavery, as the player is thrust into a very specific and mostly linear path through the game by way of a bomb collar, which will instantly kill the player if he or she strays too far for the beaten path.  Many of the challenges in the game rely on destroying the radio transmitters that broadcast the detonation frequency, which are often hidden underneath tables, inside closets, or are otherwise difficult to reach.  The goal in this situation is to create tension for the player as he or she desperately rushes to find the radio transmitter before his or her head is explosively removed.

It's pretty clear, from an outsider's perspective, to see why this mechanic would be frustrating to players.  The bomb collar produces a high-pitched, persistent beeping when under threat of detonation, which players quickly learn to avoid like the plague, for one.  There's also something particularly demeaning about being enslaved in such a way by the antagonist.  Other games that do this typically do so in such a way as so that the player regains his or her freedom quickly - while it's a good way to breed contempt for the villain, draw it out too long and that contempt falls onto the game developer instead.  Last, this kind of enforced limitation goes against what most players take the newer Fallout games for, namely, open-ended role-playing games with a variety of solutions for every situation; in Dead Money, frequently there is only one solution, and it's often the one players aren't happy with.

Bomb collars got you down?  Don't worry, we've got a special offer on brain surgery to keep your spirits up!
 However, the problem with Dead Money isn't the mechanic itself.  Analyzed at a basic level, all it is a simple race against time to remove an environmental threat - turn off the switch before you die.  The bomb collar mechanic, while effective in terms of the storyline, could have been replaced with any number of similar mechanics and still would have been just as effective.  More importantly, it wouldn't have been nearly as frustrating to players.  For example, radiation and toxic hazards are extremely common in the Fallout world - why, then, didn't Obsidian choose to instead implement the same threat in the form of radiation and, say, vents to clear it up?  Interestingly, this variation actually exists in Dead Money, but is used to a much lesser degree.  Had the bomb collar been replaced with a game mechanic which was functionally identical, but less at odds with Fallout's design tenets, I think there would have been far fewer complaints about the game's difficulty, because in that case, the challenge would have been perceived by players as fitting far better into Fallout's world, and less limiting overall - after all, if it's just radiation or acid blocking your way, that's a much more incidental threat than the villain's scheming, which if anything comes across as deliberate griefing.

Looking around, I think you'll find more and more examples of perception of difficulty being a bigger problem than the difficulty itself.  I can already think of a few off the top of my head - the jarring and repetitive taunts made by the bosses in Deus Ex: Human Revolution, for instance, are extremely grating on the nerves even if the boss fights themselves aren't overly challenging with a little preparation.  Usually, in fact, associating a character with a given type of difficulty (say, Boswer and his castles in Super Mario Bros.) can quickly cause players to become frustrated and annoyed in situations when that character is either already rather annoying, or when the game mechanics themselves aren't enjoyable - it gives people a face to yell at.


This analysis, while far from complete, should have given a pretty good overview not only of a few different types of difficulty, but it should also have made understanding why people get upset at different types of games, different scenarios, and different sorts of difficulty a bit clearer.  Creating and fine-tuning difficulty is always an ongoing process, and it's extremely difficult to get it right for all players.  Even so, hopefully this piece has shed some light on exactly why that is, and what steps can be taken at a more fundamental design level, in order to ensure that your game is fun to play, and challenging, without being frustrating as well.


  1. Hah! I completely agree with the crates in Half-life. They were my "life" blood, if you will.

  2. My least favorite way of increasing difficulty is to make it so I can't save (like in Persona 3).

  3. Fantastic article Eric - I've been really delighted with your pieces, it's the kind of in depth (so in depth that I sometimes have to specifically put aside time to read them all the way through - but it's worth it) commentary that's really lacking. Keep them up!

  4. Download Gta Vice city Game here---->>>