Teamfight Tactics and randomness in competitive game design

Riot Games recently released Teamfight Tactics, its entry in the auto battler genre, and the game appears to be extremely popular, pulling huge numbers of streaming views and lots of YouTube attention for creators making Teamfight Tactics content.

However, a lot of the discussion around Teamfight Tactics has focused on the prevalence of randomly determined events and their impact on the outcome of any given match. While Riot has defended the importance of random events and luck in Teamfight Tactics, it also recently released a patch that will smooth out some of the game’s variance.

Random events, which players call “RNG” — short for random number generation — have deep roots in gaming. When you shuffle a deck of cards and deal them to players, the outcome of the deal is a random event. When you roll a die, that is a random event. But random events become controversial in competitive gaming when the impact of randomness seems to outweigh the impact of player decisions and the expression of skill. People like the idea of “luck,” but they want the proper amounts of luck that can be overcome by skillful playing, if necessary.

Players’ objections to RNG may stem from narrow thoughts about what gaming can or should be, but in some cases, including this one, they’re right. The impact of luck can be too pervasive, and it was definitely an issue in Teamfight Tactics before the patch.

RNG in popular games

If you’ve played an MMO like World of Warcraft, you should be familiar with randomized loot drops. Sometimes you’ll get a quest to collect some items dropped by enemies, and not all the enemies will drop the correct thing. You may have to kill 40 enemies to collect the 15 items. Routine enemies may also drop rare or high-quality items, as a function of random chance.

And these conditions may change with time as developers find the sweet spot between stretching out player engagement and losing people to frustration. Here, “luck” can be viewed as a factor of how much time you’re spending trying to earn certain items, and it can also be used to keep you interested in filler tasks by dangling the prospect of rare goodies in front of you.

Loot is distributed randomly around the map in Fortnite and other battle royale games as well. You might find a very powerful gun at the beginning of the game, or sometimes you might get killed before you even find a gun at all. Here, “luck” becomes a randomized variance in when and how much power is distributed among players.

Good early loot can sometimes keep a weaker player in the game longer than they might last otherwise. But a skilled player can often beat a player who has better loot by using superior game knowledge to set up an ambush, or by using superior mechanical skill. And players in battle royale games drop their stuff when they die, so once they’re defeated, the victor will get their lucky drops.

Teamfight Tactics is an auto battler, which means — brace for a shock — that the battles are automatic. Players are responsible for making strategic decisions surrounding those battles and for managing various resources, including their items.

The problem was those resources were distributed extremely unevenly, in a way that good planning and strong play may not have been able to overcome.

Some of the game’s rounds match you against NPC creeps from League of Legends, such as Minions, the Krugs, the Wolves or the Raptors. These enemies may drop an item when defeated. Basic items can increase a stat like attack speed or health, but when you put two basic items on the same champion, they’ll combine into a more powerful item with a special effect. The right combination of items on your strongest characters can win games.

The problem is, you have no way of controlling what items you get or even how many you get. Sometimes, a player will get five drops from the first few waves of Minions while other players get as few as one, and this will allow the player with the item advantage to snowball the early and midgame. This has been extremely controversial.

Riot has implemented “bad luck protection” with the latest patch, which increases the odds of getting items in later rounds if you don’t get them early. However, in many cases, a good player will parlay an early item advantage into an insurmountable gold advantage by midgame, while players who are item-starved may not last long enough to stabilize when they get several items in the Raptor round.

RNG is a widespread element of game design, and it’s usually fine

When Magic: The Gathering came out in the early 1990s, the random elements of the game were controversial. In a constructed-format game, each player began with a deck of 60 cards, with a starting hand of seven, and would draw an additional card each turn. Over the course of the game, players would only get access to a fraction of their deck. Sometimes they’d draw the perfect hand, and sometimes they’d fail to draw enough resource-producing land cards and would be unable to do anything.

The challenge of deck construction is to minimize the likelihood of a bad draw shutting down your game. But sometimes, you get the nuts and win the game on luck, and sometimes your opponent does.

Magic’s creator, Richard Garfield, argued in a 1995 column for The Duelist magazine that the fact that luck could sometimes determine the game’s outcome wasn’t really a fatal flaw, because, beyond the single game, there is a “metagame” — the game beyond the game. This larger context included not just the individual rounds, but the series of all games. While luck might play an outsize role in a single match, skill and a strong deck should win out across a series of games.

A good analogy is Texas Hold ’em poker. If you look at a single hand as a discrete game, it usually looks like a game of chance. Each player is dealt two cards at random, and then the community cards are dealt randomly. The player who gets the best cards is likely to win the hand, while players who don’t perceive themselves to have good odds will likely fold.

But poker isn’t just a single hand. It is a series of many hands, and the game generally tilts in favor of the more skilled players. Or, to quote the movie Rounders: “Why do you think the same five guys make it to the final table at the World Series of Poker every single year? What are they, the luckiest guys in Las Vegas?”

Elaborating on this concept in a 2009 column, Mark Rosewater, Magic’s lead designer since 2003, explained that random events actually favor the more skilled players in general.

“The more unpredictable, unknown variables that get added to a game, the more opportunity there is for the better player to take advantage of them,” Rosewater wrote. “This is one of the major reasons, for example, that experienced Magic players have such an advantage in Draft formats. It turns out that the ability to react requires a lot of skill.”

But sometimes RNG is not fine at all

While Rosewater defended RNG as a useful game design tool, he acknowledged that its poor reputation among players wasn’t entirely the result of misunderstanding. He explained that it is “devastating” for randomized factors to make a game seem frustrating and arbitrary, and noted that it is important to avoid making Magic feel like it hinges on mechanics that are beyond the control of players.

For example, players are very comfortable with the idea that you shuffle a deck of Magic cards before a game, and that the cards that come out of the deck are randomized, because they get to decide what goes into the deck beforehand to try to compensate for the variation in card options caused by random drawing from game to game. The player who is better at building decks around the conditions of random draws is a better player. That’s part of the game.

On the other hand, suppose there was a card in Magic that had an effect like: “Flip a coin. Heads, you win. Tails, you lose.” Players wouldn’t like that very much.

But Blizzard Entertainment came close to doing just that in the 2016 Hearthstone expansion Whispers of the Old Gods, when the studio released a card called Yogg-Saron, Hope’s End.

Yogg was an endgame card that cost 10 mana, and his battlecry cast a random spell with random targets for each spell you had cast up to that point in the game. This was an effect that could backfire spectacularly; most top players dismissed the card as a joke upon its initial release.

But Yogg ended up seeing a lot of play in professional tournaments, which was something his designers had not expected, and he shifted the outcome of many of the games in which he was used. Blizzard nerfed the card by stopping Yogg from casting more spells if he destroyed himself, silenced himself, or returned himself to his user’s hand. This sharply reduced the expected value of playing the card, and was enough to banish it from most high-level play.

But Blizzard still seems to think the effect is fun; a new card announced for the upcoming Saviors of Uldum expansion is called Puzzle Box of Yogg-Saron, and it is a 10 mana mage spell that casts 10 random spells with random targets. We’ll see how players react, and how much use it gets in competitive play.

Riot, similarly, is pretty attached to the randomness of the item drops in Teamfight Tactics, but its latest changes will at least narrow the range of variability. Players who don’t have many items in the early game know that items of some kind will be coming their way in the later game, and they can plan accordingly.

Riot designer Andrei “Meddler” van Roon posted on the League of Legends message board that the new changes will ensure that a player should never get six items from the first three Minion rounds, and that a player should also never get zero items. Also, players will receive extra gold from creep rounds where they don’t get items. This means everyone should get more items, but some players will still get more items than others, with the numbers not evening out until bad luck protection kicks in, possibly as late as the Raptor round.

I’ve personally seen games since the most recent Teamfight Tactics patch where I got only one item from the first three Minion rounds and no item from the Krugs, and I’ve also had games where I got four items from the Minion rounds. In the games where I got poor outcomes on items, I got a total of nine extra gold.

But you also receive one gold for each round you win, which is extra gold players get when their early items tilt rounds in their favor. And you also get gold for win streaks, which allows players to snowball an early advantage into an often insurmountable lead. I’ve found I have to spend gold more aggressively to try to keep up in games where I had few items, so it’s a lot easier to build an interest-producing economy when you get early items, despite the new gold drops.

Teamfight Tactics product lead Richard “MapleNectar” Henkel notes that this variability creates fun game experiences where a player gets everything they want, and that Riot wants to encourage those kinds of stories. But it also creates situations where players face significant disadvantage arbitrarily. Not being able to control which items will drop is good RNG; players can respond to that randomness by building their team compositions around champions who can best exploit the items they randomly got. But variance in the number of items get from player to player and from game to game feels like the kind of bad RNG that Magic’s Rosewater warned about in 2009.

It’s like a game of Hold ’em where some people randomly get dealt a third card, and others only get one card. It is possible that very good players will be able to salvage games like this, or at least manage to hold it together well enough to finish in the middle of the pack (TFT considers the top four players out of the field of eight to be winners). But a lot of people would lose games in that situation, and would feel the losses weren’t really their fault because of the way these items are distributed.

Can a game sustain RNG and still be competitive?

Teamfight Tactics’ item variance may allow for scenarios such as one in which players try to assemble a very specific item build requiring an unlikely five Needlessly Large Rods, which allows Akali to one-shot multiple enemy champions. Streamer and retired LCS pro William “Scarra” Li, who has emerged as a top TFT YouTuber with his informative guides to the game’s systems, managed to put this build together, but was down to his last sliver of health before he got all the pieces lined up.

In the first match using his fully armed and operational battle station, the Akali gibbed his opponent’s entire team, but in the next round, Akali got deleted by a critical strike from an opponent’s Kha’zix, and Scarra finished the game in fourth place — a win, according to Riot’s definition of the term, but just barely. The situation created a lot of drama, and was fun to follow.

This is the kind of TFT story that MapleNectar hopes will emerge thanks to TFT’s random qualities, and these kinds of builds do make for great streams and YouTube content. But Riot is launching a ranked mode for TFT, and people who play to climb won’t be playing for the memes. A more common scenario will see players using item advantages to augment more conservative builds and winning relatively unremarkable victories against opponents who arbitrarily got less stuff. That could be bad news for the health, and fun, of spectated games. The latest changes will hopefully split the difference between those two dangers.

Riot will be patching TFT weekly, an even faster cadence than its famous every-other-week updates to League of Legends, so there will be plenty of opportunities to further adjust item distribution in future patches.

However, if Riot is too cavalier about how the random drops are calibrated, and the variance in power that is arbitrarily distributed to different competitors, TFT may continue to feel like an unfair game. And if it does, the stories that come out of ranked play won’t be as fun as the ones in the popular videos with big swings and surprisingly outcomes. We’ll have to wait and see, and the final balance may take weeks to find.

That’s because luck isn’t always luck in game design, nor are most things in gaming truly random. The odds of each drop and “random” occurrence are set by the designers behind the game, even if the players don’t understand that what may seem random is actually a tightly controlled circumstance designed to maintain players’ interest in the long haul.

Whether or not luck is being managed “well” for each game may come down to a simple question: Are people still playing?