Predictable and unpredictable randomness in games

RNG or random number generation is a controversial topic in gaming with equal numbers of players cursing and blessing the mystical rn Jesus who dishes out crits seemingly on a whim. Such a mechanic has the potential to vastly change the game experience and must be utilised carefully in order to ensure games still feel fair whilst garnering the many benefits afforded by randomness.

Randomness versus skill – what exactly is ‘luck’

Randomness is the idea that there is uncertainty in the outcome of an action whereas skill is the degree to which an outcome can be assured by a player according to their ability. In a game format randomness is typically manifested via random number generation whilst luck is modulated by player experience and innate talent. Typically, skill and randomness are viewed as opposing elements within a game, however, randomness can be broken down further into predictable and unpredictable randomness which each relate somewhat to skill.

Predictable randomness refers to irregular elements that, to some degree, a player can account for and foresee. A common example would be hip-fire in FPS games where the bounds of accuracy are displayed but exact bullet location randomly varies within. Contrastingly, unpredictable randomness is a set of circumstances that are impossible or virtually impossible to meaningfully predict. For example, world generation in Minecraft is for all intents and purposes, random – you may be able to predict that sand will be next to water as a general rule, but nothing specific enough to be meaningful.

Often when players die in online lobbies they’ll immediately jump into chat and accuse their opponent of getting ‘lucky’. Randomness is one way someone may feel justified in claiming luck, but to what degree? Unpredictable randomness, by definition, cannot be predicted meaning that any action reliant on such randomness can be attributed to luck. However, an action relying on predictable randomness can be hard to attribute directly to luck – Spraying a shotgun from the hip will result in a random pellet spread, but it can be reasonably assumed that in the vast majority of cases enough will hit at close ranges to secure a kill. This calculated action based on experience and game knowledge is a manifestation of skill despite requiring an element of luck.

A Mixed Case study – Luck in competitive games, League of Legends

League of Legends is a massively popular game centered around online PvP requiring tightly honed mechanics, expansive game knowledge and fast reaction times. It contains two examples of randomness that exemplify the issues and boons of randomness in competitive games.

Somewhat recently LoL added an update where the PvE enemy, the dragon can vary in type. Each type offers a different buff to the team that kills it and these stack and persist over the course of the game. The type of dragons spawning is randomised each game and also varies during the game (there are some additional rules to what type spawns but they are inconsequential to the points being made here). The dragon types also affect the map potentially adding more healing resources or removing sections of walls for example. This is an excellent use of randomness and has been integrated intelligently enough to mitigate the potential negatives. Firstly, this increases the skill cap for game mastery. By knowing which dragons will benefit your team the most you can choose which to risk fighting for and which ones to ignore. Additionally, players better at fighting will likely win more fights near the dragon and thus secure more dragon buffs, rewarding better players. The integration of the dragons is well done with clear UI telegraphing which dragon types will spawn and when allowing players to make mid-term strategies whilst also rewarding those who can adapt to new information the best. Finally, it encourages new types of gameplay – teams oriented around early-game fights will win more dragon buffs enabling them to potentially crush the enemy quickly whilst their team is strongest, but equally if a late-game team composition is able to sneak a few dragons they will power ahead to the late game victory.

Is that dragon soul worth it? Assessing the significance of neutral  objectives in League of Legends via Gradient Boosting Classifier | by Mary  Zhou | Medium
The dragons bring a unique flavour to each game encouraging adaptation and improvisation whilst being telegraphed enough to feel fair

Contrastingly, LoL also features critical strike chance. This is a mixed problem as Greg Street, Lead Game Designer at Riot Games agrees, calling it an ‘unsolved problem’. In this situation critical chance is the percentage that an attack will do 175% of its normal damage (this can be modified but again, the point still stands). This can as Street puts it ‘feel like bullshit’ when someone with 5% critical chance lands 2-3 critical hits in a row when someone else with 75% lands none in 5 attacks. For a game with such a large eSports scene it is surprising that such randomness exists, especially as such a core mechanic with 1/5 players typically focusing their game around critical chance. Due to UI displays. this is technically predictable randomness and over 100 fights the higher critical chance character will on average land more critical hits, but this can feel incredibly unfair in one of situations and mitigate a potential skill gap. In LoL PvE and PvP kills generate gold which can be spent on critical chance, meaning to some degree skill correlates to the ability to rapidly acquire critical chance. The possibility of a player with less critical chance landing more than a player with more is therefore directly opposed to the concept of rewarding skill which seems counterintuitive to LoL’s otherwise competitive nature. The reason for this is output variation versus input variation.

Output vs Input variation and balancing fairness

Simply put, input variation is randomness before action whereas outcome variation comes after action. Random world generation would be input variation whereas randomly generating critical chance after landing a hit would be outcome variation. Greg Street agrees that critical chance is a typical case of outcome variation feeling unfair as equal actions can result in unequal outcomes between players.

Output variation is typically construed as unfair but this can be modulated by its predictability. Firing from the hip is predictable to some degree and the player knows that there will be randomness in bullet spread. This combination of knowing there will be randomness and the potential for failure combined with the ability to predict the degree of likeliness of this reduces the feeling of unfairness except at the extremes. For example, firing from the hip at close range will typically hit as predicted but missing a single slug at near melee range will feel unfair if it misses even if the player knew there was a small chance of such occurrence.

Designers can balance the odds in the player’s favour to reduce the occurrence of such events. LoL for example has some ‘bad luck protection’ on what dragons spawn next and the real spread of a gun’s hip fire could be somewhat constrained versus the on screen reticle’s indication. This has the added benefit of making players feel more skilful even when they are not.

Player deceit Case Study – XCOM

XCOM and XCOM 2 see humanities’ force resist against an alien threat via a semi-RTS strategy and turn-based combat format. Combat is its bread an butter though, particularly with the XCOM 2 expansion, War of the Chosen. The core experience requires the player to accumulate soldiers and technology before deploying to a range of randomly generated and pre-generated missions. The catch – when a character dies it is gone forever, along with its kit if you don’t bring back the body. The game is all about weighing up decisions – who will you risk, how will you set up sightlines and how will you push through before the turn counter ends and you’re overrun.

XCOM: Enemy Unknown – review | Games | The Guardian
XCOM’s UI provides all the information required to balance risks and rewards yet players fail to prey to statistical fallacies requiring behind-the-scenes help

During combat the player is informed of their hit chances versus enemies and before deployment you can customise your character’s kit to enhance their base stats too. This allows the player to accurately predict their chance of success, but ultimately it’s all only probabilities. Or is it? It turns out that the numbers on the UI are meddled with in order to enhance the experience. For example, your chance to hit is actually multiplied by 1.2x, after a miss you gain  flat +10 aim bonus on all shots over 50% until you land a hit for each miss and when hit you can a stacking -10 aim reduction as long as you have less than 5 squad members. This is all because humans suck at calculating percentages. This means that despite all the informative UI allowing the precise chances of success to be calculated, players will still get annoyed when 3x 33% chances to hit fail. So XCOM lends a helping hand. This additionally generates more ‘hero moments’ when only one squad member is left and dodges 4x 90% chance to hit shots in a row before evacuating safely. War of the Chosen capitalises on this generating personalities for each character that evolves with their experiences allowing the player to become invested in what are initially, just randomly generated templates.

XCOM is a perfect example of limiting randomness at the extremes and tailoring an inherently random game to the player experience. It is still possible to have unfair feeling moments but the chances are reduced allowing unskilled players to get into dangerous situations and still succeed with a story to tell. The behind-the-scenes trickery even benefits the enemy at harder difficulties making it all the more rewarding when skilled players pull off a difficult mission – XCOM thrives when players are pushed to their limits and the game facilitates that regardless of skill level.

How to use luck and randomness

Typically, luck and skill are thought of as being a sliding scale; either a game is luck based or is skill based. However, Richard Garfield dismisses this idea, instead offering the idea of a 2-way independent interaction.

For example, a randomised chess involves a dice roll after a chess game. On a 1, the loser wins instead. This game still requires a lot of skill as the consistent chess winner will average out wins, but luck is enough to swing the result in a close series of matches.
  • Increases variety in a game – force players to adapt and improvise to new situations, just make sure to calibrate the planning time to the game’s tone.
  • New and unusual situation force a player to try new things and get more experiences from a game. Procedural games often have a wide variety of locations and AI agents, just make sure to not sacrifice depth at the expense of making more.
  • Randomness leads to innovation. Chess has been honed so much that the opening moves are all from a known pool, but randomness would keep things fresh at the expense of learned experience to some degree.
  • Randomness protects egos. Weaker players can blame randomness preventing them becoming too annoyed to keep playing, whereas skilled players may account for and acknowledge it.
  • Watching learned, repetitive, skilful gameplay excites some people but seeing the range available within a game and the skill of adaptation excited others.
  • High randomness games makes matchmaking easier and means experienced players can play with friends who may still win via luck.
  • Blur optimal strategies – randomness makes it harder to calculate the optimal strategy encouraging innovation and use of high-risk high-reward versus low-risk low-reward strategies.
  • Randomness keeps a game feeling new and fresh without every match feeling the same.

Ultimately a game needs to balance skill and luck carefully. Games erring towards rewarding skill can still incorporate randomness but it should be intuitive, telegraphed and predictable with carefully calibrated caps to prevent unfair randomness. Casual games can utilise randomness in more ways to balance the playing field between players, broaden the range of events and actions possible and increase the rate of unusual, interesting circumstances that could not be predicted and would seem unfair in a competitive game.

Leave a Comment

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s