How to use RNG in Competitive Games

Overview:

RNG is used everywhere in games, it's almost impossible to find video games that don't have some element of randomness involved. There are many benefits to it, but it can also be overused, especially in games with a competitive scene.

In this post, I will be mostly comparing Hearthstone and Magic the Gathering, but all of these examples apply to a wide variety of other games.  If you're unfamiliar with these games I've included links giving a brief overview of what the games are about. The basic idea of both of the games is that you build a deck of cards containing spells and creatures which you summon to try to reduce your opponent's life to 0.

Links for those unfamiliar:

https://www.youtube.com/watch?v=QpO0r-3pEuo

https://www.youtube.com/watch?v=CVZ4qyx-c2o

https://www.youtube.com/watch?v=GadqHggATe8

https://www.youtube.com/watch?v=LbnfSG_WGXI

Purpose of RNG:

Usually the main reason to include RNG in a game is to attract new players and to keep casual players playing.  If a game is purely skill based like chess, the more experienced player will nearly always win. However if there are random elements involved the weaker player will be able to win some percentage of the games and won't get as disheartened.  There are also benefits to the competitive player as well. It both helps ensure that there is no single dominant strategy, and can add interesting strategic challenge that the player had not planned for. It also keeps the game exciting and keeps every game you play from feeling the same each time, which it beneficial for everyone.

Delta of randomness:

This is one of the most important things you'll need to understand when designing randomness into your game. It describes the distance above and below the expected result that the random element could result in. This is most easily described in card games through a term referred to as the power curve. It's an imaginary line that represents the average amount of power you should be able to expect for a given mana cost. This can still be applied to other games as well. Let's say you have a sword in an MMO that does 12-16 damage every time you hit with it, and another that does 6-10 damage each hit, but has a 25% change to score a critical strike tripling it's damage.  The average amount of damage you expect per hit is similar, but the second has a much higher delta of randomness. High deltas of randomness are what you want to avoid if you plan on giving your game an enjoyable competitive experience. Lets take these two Hearthstone cards as an example.

(Spare parts of one mana cost cards with minor effects)

These cards are both similar in stats, power, and manacost. The main difference here is that Piloted Shredder has a much higher delta of randomness. This has resulted in it being hated by the majority of the competitive community. In order to play competitively players were forced to use this card in their deck because it was on average slightly higher on the power curve than anything else at 4 mana. This resulted in many matches being almost completely decided by what creature came out of the shredder instead of the skill of either player. Mechanical Yeti on the other hand very rarely decides games on it's own. It comes down to how the spare parts are used and allows for some counter play as players try to guess which one the other player received. 

What Wizards of the Coast does differently is that when they design cards for Magic that have a high delta of randomness they try to ensure that the expected average for the card is below the power curve. This way it doesn't effect competitive play.  While it doesn't always work it does seem to be something they actively try for. It is very apparent by the next card that Blizzard has no such goals.

These cards both have an incredibly high delta of randomness and will often end the game as soon as they are played. The main difference is that 8+ mana in Magic is much harder to attain than it is in hearthstone. This leaves Scrambleverse well below the power curve even though it has a decent chance of winning the game when played. This allows it to still be fun for the casual players who enjoy that sort of card, but keeps it from effecting competitive players. Pre nerf Yogg Saron on the other hand was the bane on competitive Hearthstone for months. They could have changed his textbox to read "60% change to win the game, 30% to lose the game, 10% to do nothing" and it would have been essentially the same card. This is a great effect to have in the game as it's fun for lots of players, but you don't want to see thousands of dollars in prize money riding completely on the roll of the dice.

Giving players control:

Large amounts of RNG effects can still have a place in competitive games provided you give the players some control over the outcome.

Both of these cards have a similar effect in that they give you a random card from outside the game and both have a very high delta of randomness. The main difference here is that Jeweled Scarab gives the player some choice in the matter.  While the number of potential cards is still very large the player very rarely feels cheated in the exchange as they will be able to avoid picking useless cards the majority of the time. Giving players a choice it something you want to make sure happens as much as possible in your game. Both of these cards can lead to exciting and unpredictable events in the game, but manages to avoid most of the feel bad moments at the same time. 

Input vs output variance:

This is on what level of the game the RNG takes place. With output variance you don't know what is going to happen when you play a card or activate and ability. Drawing cards from your deck is an example of input variance, you don't know what it will be, but you've drawn it you know what it is and what it does before you decide to play it. Two examples from Hearthstone that illustrate this well are Imp-losion and Blackwing Corrupter.

These two cards serve the same role. They both kill your opponents creatures and leave you with creatures yourself. The difference here is where the randomness takes place. With Imp-losion the random element occurs when you cast the spell. You don't know what the result will be when you cast it making it an example of output variance. This is a large problem with the card. Often when playing this card you would need it to deal 3 or 4 damage in order to kill a minion. If it didn't you were left disappointed with the result. On the other hand if you did roll high and kill your opponent's minion they were left unhappy with the exchange.  Nearly every time this card was cast one player would feel cheated by the game. This was compounded more by how wide a range of power this card be. Blackwing Corrupter on the other hand is a prime example of input randomness. The random element here takes place before you play the card. When you're building your deck you don't know if you will have a dragon in your hand on turn 5 when you want to play this card, but by the time that turns rolls around you will have the information you need. This card doesn't always do three damage the same way Imp-losion doesn't always do 3 damage, but in Blackwing Corrupters' case you know that before you play it and you have the option to simply play a different card that turn and save it for later.

Magic the Gathering is a game that works almost entirely on input variance. Nearly all of the cards have a set effect when you play them and most of the random effects that happen are when they cause you to draw cards. The small amount of time that they do use output randomness it is often with very small deltas of randomness and with the player partially in control of it. Cursed Scroll and Ghoulraiser are good examples of this.

The lack of splashy random elements is mostly because of the limits of it being a physical card game. It is much more difficult to work in random mechanics into a physical game like Magic, while it is incredibly easy in a digital game like Hearthstone. While this may be a limiting factor it does often result in a much more enjoyable experience for the player as when they win it was more often because they played well and not because of some random splashy effect. 

Input and output variance can be applied outside of card games as well. Let's look at an MMO example again. Say you are setting up a vendor that sells weapons and armor. You want to give the player a chance at getting something more, like a random enchantment for example. There are several ways you could go about this. One would be to use output variance, make it so that after buying the weapon there's a small chance it upgrades. This way the player doesn't know what's going to happen when buying the weapon but get's excited when the upgrade happens. A better way though would be to use input variance. In this example the vendor randomized what he is selling, sometimes he sells weapons with fire enchantments, sometimes he sells armor with frost resistance. In this example the players still feels the same excitement when rolling into an enchantment that they need, but they don't need to spend money on weaker weapons hoping for certain upgrades.  

Conclusion:
While adding randomness to games has many benefits you'll want to be careful not to over use them, especially if you are trying to make your game competitive. Competitive players typically hate random effects that they don't have control over because then the result of the game can be decided by that instead of their skill. Even casual players will dislike randomness if taken too far. No one is going to play a game for very long if they feel like they have no control over it. This is why competitive chess exists, while competitive snakes and ladders does not.

Additional Resources:

Extra Credits has an excellent series on randomness that covers some of the same topics mentioned in this blog. You should check it out if you'd like to explore this further: 

https://www.youtube.com/watch?v=-9ZI9kMsvRQ

https://www.youtube.com/watch?v=0V5eq4IQ6Go

https://www.youtube.com/watch?v=GiOA_CS25Kw