If and why is randomness a bad thing in competitive games?

Question

AFAIK it is a consensus within esports that randomness should be reduced? League of Legends has little RNG, Dota has more but both games employ rng smoothing algorithms (details for Dota, LoL).

The rationale is that if there is randomness then random chance can easily determine the outcome of a match (instead of player skill) which is unacceptable for pro play and tournaments. Quoting the articles I linked to above:

In general Randomness is rejected from competitive gaming since it can decide over loss and defeat without reflecting the actual skill level.

 

In League, a single action can determine the outcome of a lane, or even the game. For this reason, randomness in the game is a tricky topic. While landing a huge critical strike (crit) on your opponent can be satisfying, being the victim of a “lucky” crit can be frustrating, or worse: a random event could cause a skilled player to lose to a less-skilled player, even though the skilled player made a better play. In a casual game, the decision to include randomness might be a no-brainer, if it makes the game more fun. However, the decision to include randomness in Esports has the potential to make or break a pro player’s career or determine the outcome of a tournament. As a result, many game developers take measures to reduce the influence of randomness in their games.

Seems reasonable, but... I dont understand, there just seem to be too many counterexamples! Pokemon is insanely random, yet Pokemon is being played competitively as far as Im aware. More traditionally, Poker and Bridge seem to be ones of the most obvious counterexamples. These games would seem to demonstrate that it is not the case that pro play demands little to no randomness?

If and why does it make sense to reduce randomness in games intended to be played competitively?

For the sake of completeness, I feel I need to include two more opinions on the subject I heard/read.

Firstly: On Discord, someone told me that Poker is not really comparable to LoL as it is less popular (I was astonished to learn that LoL tournaments have tens of million watchers while Poker only has hundreds of thousands) and Poker is a gambling game, which is why randomness works for Poker - randomness in games makes them gambling games which presents a problem on its own.

Secondly: Extra Credits made a video on the subject, where they said randomness can work for competitve games (and IIUC implied that the claim that competitive games must have little to no randomness is incorrect) but tournament format needs to be adjusted - instead of single elimination they should make the tournaments in round robin. But doesn't this introduce new problems, since - correct me if I'm wrong - don't most people, both players and spectators, find round robin excessively boring?

Answer 1

I feel thinking about randomness as a single, monolithic thing isn't super useful in a game design context. A distinction should be made between input and output randomness.

Output randomness

1. Player(s) make decision(s)
2. Random event determines the outcome.

Explanation

This is the type of randomness we're most familiar with. In the board game Risk, players place their pieces, decide who to attack/defend, and THEN a dice roll determines the outcome of their decisions.

Input randomness

1. Random event occurs
2. Player(s) respond.

Example

In League of Legends, there are 4 types of dragons that can spawn, each which give different buffs and have different attack characteristics. The dragon-type may determine how many resources a team wants to invest in the objective, or whether they want to contest at all. In LoL, you're given ~4m heads up telling you the next type of dragon that will spawn. This is crucial, as it means teams are given the chance respond and shift strategies depending on the random event.

Conclusion

In general, Input randomness is fine. It can lead to interesting decisions as it gets players to react dynamically to new scenarios which forces players to understand the systems of a game instead of memorizing a set of actions which have been solved by a meta. On the other hand, in a competitive context, Output randomness is bad. It adds chaos, meaning that more games will have to be played to determine the skill of each player/team. If the purpose of a competitive game is to determine skill (knowledge of a games' systems, knowledge checks, and mechanical proficiency), then output randomness is always undesirable as it increases the sample size you need to do so.

In Practice

Now of course in reality things are not as clear. A blurry spectrum exists between platonic input and output randomness. Critical strikes in LoL in the context of a single auto attack are purely output randomness. The summoner decides to auto, and then a virtual dice roll determines the output damage. However in the context of an entire prolonged fight, crit-chance skews towards input randomness. Players get to respond to these unexpected variations in crits and shift their play accordingly by retreating, chasing, or holding abilities in response to shifting luck.

Here's the article which founded these terms if you're interested in a much more eloquent and comprehensive explanation

Answer 2

First things first, to be BIG in competitive games, you need spectators. If nobody is watching, then nobody is to there to be made money of. So obviously whatever randomness is in a game is more or less important to the viewer count.

The viewers like to watch a game, if it is interesting to watch. Its fascinating, intriguing and not foreseeable who will win. And in most cases, the viewer know more than the actual players. E.g. in Poker, the viewers often can see, what the players hands look like, even the percentage of winning with their hands compared to the other hands etc.

Now imagine a completely deterministic game with the difference from game to game is the decision a player makes. There also is something like the optimal way. Like chess, but the starting player has a way of always winning. Like a series of certain moves the opposing player cant stop. Noone would watch it. And pro players in any game will certainly reach this optimum.

Now lets just add a bit of randomness and it is open who will win. Does the Pokemon attack hit and one-shot the enemy Pokemon or will it miss? Even the selection of Pokemon is 'random' as the enemy does not know beforehand, what Pokemon his opponent will have. That is where Meta will kick in. And it just became a lot better to watch.

In Pokemon, one unlucky move can cost you alot. One paralyze, missed moved etc. It can determine a game in one round of around 10. In opposition to that, in League just one lucky crit does nothing (most of the time). In this kind of game, the players skill is more important than one crit. If a game like this would be influenced more by a lucky roll than skill, then spectators would get disappointed that the better team would still loose.

And then there is Poker. Poker is not purely luck-based. Its more what a player does with his dealt cards. Does he fold, raise, bluff or roll high on his insight on his opponents bluff? This again is skill-based with a lot of randomness. Good players know how to reduce the amount of luck they need and just go the way of higher success in relation to their assumption of the oppnents cards.

All in all, it seems that there is a sweet spot that determines if a game is enjoyable to watch or not. No randomness will make the game to deterministic and obvious who will win. To much randomness and the team you are rooting for looses every game, although they are clearly the best team (seriously).

Answer 3

You have to take into consideration the frequency and significance of randomness.

Let's take a MOBA as an example. Say that we are auto-attacking 20 times per minute, and each attack has a flat 5% chance of being a double-damage critical hit. This means we can usually expect one critical hit per minute, and our damage output is going to be about the same. Sometimes we might get very lucky and get in four crits in 20 attacks, giving us a 15% damage boost for that period (we do 120% damage instead of our typical 105%), which is noteworthy but not usually game-breaking.

With smoothing we can make the randomness even less random; for example, we can code the game to always award exactly 1 critical hit per 20 attacks, simply making it random which one of those 20 attacks is a critical hit. This will ensure our damage rate doesn't suddenly spike due to particularly good luck.

Now, consider a different scenario where there are no critical hits, but the damage of each attack varies by +-5%. On paper this doesn't seem like a big deal. If we had perfect luck, we'd be doing 5% extra damage all of the time (or -5% if we had perfectly bad luck). In practice, we might have streaks of good or bad luck, but it should average out to 0% over extended periods of time.

However, imagine that our base damage is 100, which makes our randomized damage 95 - 105, and there is an enemy unit with exactly 100 HP. This means we can expect about 50% of our attacks to kill the enemy in one hit, and 50% to leave it with a sliver of health left. That's a huge problem, and bursts of particularly good or bad luck can significantly change the outcome of the game.

Now imagine we have +- 5% randomness on an AoE ultimate attack that does 800 base damage. If the ultimate hits multiple targets, that +- 5% could add up to hundreds of points of extra or reduced damage, which again can significantly alter the outcome of the game.

Too much randomness can also lead to a surge in player complaints which can generate negative publicity and put extra stress on customer service reps.

When is randomness a good thing?

Randomness can still have an important place in "gambling"-style moves or characters. For example, you might give one hero an ultimate that does +- 50% damage, or has a 50% chance to hurt allies instead of enemies. Because this is a single, highly notable ability only possessed by one character, its randomness may be more widely accepted and players will likely develop strategies around utilizing or avoiding the move. It can also lead to hallmark moments in competitive play where a player decides to make a big gamble with a highly random ability and it pays off or fails in spectacular fashion, making the match a fan favorite that is destined to be rewatched and debated endlessly for years to come.

For gambling-style abilities to fit into the game well, their nature should be obvious to the player (everyone knows this move is heavy on random), and the player should have total control over when the ability is activated (so they can choose when to take a gamble, as opposed to having it thrust on them continuously or at random times).