State of Gameplay, Sep 2021

Our thoughts on how various aspects of gameplay have been this year.

Hey everyone! I've talked to you in a few lol pls and Season Start videos before, but this is the first time I’m writing a blog! My name is Jeremy and I make sure that everything our Gameplay teams do is in service of giving you all the most competitive, fun, and high quality game experience possible. For this State of Gameplay post I want to check in on how balance, champion, and competitive projects have gone over the course of 2021, and where we’re focusing our energy to improve League in the coming months.

Patch Balance

Let’s start with patch to patch champion balance. We've been using our champion balance framework to guide our balance work and have seen good and stable results. On any given patch, we tend to only see a small handful of champions crossing Overpowered thresholds, and for the vast majority of the year, every champion has been viable in at least one of the four skill levels. Since the current framework has led us to a stable state of balance for most games, we think we can investigate improvements around some variables without putting patches at significant risk:

  • Experienced win rates: The balance framework focuses on the average success of a champion without sufficiently accounting for how strong the champion is in the hands of dedicated mains. The goal of balance is to ensure everyone has a fair chance at winning if they play well. So, even though we want experienced mains to be rewarded for dedicating a ton of time to their champ, they shouldn't command such a higher winrate that victory becomes unrealistic for the enemy team in champ select.
  • Low playrate champions: Certain champions are seen too infrequently to raise flags in the current champion balance framework, even if they've been sitting at a higher win rate for extended periods of time. While this overlaps with the last bullet, we want to be clear: Just because a champion is low playrate doesn't mean it's just a few dedicated mains playing them. Some champions, like Aurelion Sol, see a small but steady rotating door of players trying him out for a few games and then moving on. We want to adapt our current champion balance framework to be able to account for these cases as well.
  • Elite play thresholds: Our elite player segment represents the top 0.5% of players (roughly Diamond 2 and higher). One of the challenges of balancing for this group is is that there are significantly fewer matches being played, which means we have way less data to work with. The framework already uses more than one patch to account for this, but we're investigating possible threshold changes or new strategies we can use to ensure we are making balance changes that are hitting the right targets for this segment
  • Mid-scope champion work: For some balance issues, like pro vs regular play skew or particularly longstanding problems, the most effective approach is to make larger-scale adjustments. We've shipped mid-scope updates to champs like Sona, Tahm Kench, and Amumu this year and will continue to look for additional candidates.

Pro Play Meta

Since we're gearing up for Worlds, we want to touch on the pro meta. At MSI, we accomplished our goal of no 100% pick/ban champions, though a few came very close (Udyr at 95%, Renekton at 93%, and Lee Sin at 91%). We saw 82 different champions, and four out of the five roles saw at least 10 picks with 5% or more presence (support missed the mark).

Recently, our balance changes in patches 11.18 and 11.19 focused almost exclusively on ensuring the Worlds meta has as much potential for diversity as possible. Based on how summer playoffs went, we can't wait for what should be an exciting World Championship this October.

Item Balance

Last preseason's item system overhaul brought a lot of change to Summoner’s Rift. Since launch, we've been monitoring the purchase rates and power levels of items across the system from a class-focused perspective (Mage, Fighter, Marksman, Tank, Assassin, and Enchanter). Much of our initial work focused on addressing over- and under-performing outliers among Mythics like Moonstone Renewer, Stridebreaker, Everfrost, and Sunfire Aegis. In some cases, we've looked to fill gaps in a class’s item choices. This happened during the summer when we added in Anathema’s Chains and Hullbreaker for top lane fighters and tanks.

We’ve also adjusted our goal that every champion should have multiple item paths to choose from. While this is still true for Mythics, we think it’s okay if some champions build the same Legendary item in most games if it’s a perfect fit. But some champs don’t have that yet, so this year's preseason will include more new offerings.

Champion Releases

We’ve had five major releases this year so far: Viego, Gwen, Dr. Mundo, Akshan, and Vex.

9242021_StateofGameplayArticle_Viego.jpg

We started the year with Viego, the AD jungle assassin who possesses the abilities of his defeated enemies. Release balance was within our targets given Viego’s learning curve, at 42% on Day 1 and 47.5% on Day 7. Since then, his performance outside of the jungle has been something we've needed to keep a close eye on. In particular, we had to tone down Viego's lane sustain since it, combined with being a resource-less champion, was making it too hard to shove him out of lane. The other notable thing is that the underlying code that allows Viego to use enemy abilities caused quite a few bugs in the game and it took us a while to track them down and squash them.

9242021_StateofGameplayArticle_Gwen.jpg

Gwen released in April, and our goal was to bring a lighter thematic to the AP top lane fighter roster. She released slightly stronger than we had anticipated, at 42% on Day 1 and 48% on Day 7, but these are still healthier numbers than past seasons' new champs. What we did have to adjust was Gwen’s Pro presence: She was more skewed to pro play than we anticipated, and we had to nerf her early laning as a result. She’s in a better state now (we even gave her a small buff in 11.19) but we need to monitor her performance to make sure she’s still delivering on her fighter fantasy.

9242021_StateofGameplayArticle_Mundo.jpg

June saw the return of Dr Mundo where he’s been gleefully “helping” his “patients” in top and jungle. We wanted to maintain his super-regen and super-tank identity with a low complexity kit, while adding more depth and a lot of visual and thematic upgrades. His writing and voice acting were really well received by players! Less successful was Mundo’s initial balance where he hit a 51% winrate by day 7 in average play, so we lowered some of his baseline durability. On the other hand, we were too conservative with his strength in the jungle, so we've also been gradually bumping up his effectiveness against monsters to let the gud doctor more easily go where he pleases.

9242021_StateofGameplayArticle_Akshan.jpg

Akshan swooped in to save the day in July, as an AD marksman/assassin hybrid designed for the midlane. Akshan’s character and gameplay are all about being the hero and being flashy while you do it, and his abilities are indeed flashy. When it came to launch balance, we released Akshan in a pretty weak state as we weren’t sure just how powerful his ally resurrection ability would be. We’d tested it for months but with such a new mechanic we couldn’t be too sure how it’d play out across all levels of play. We were mindful to make sure he had definite weaknesses across his kit, such as the largest camouflage reveal radius in the game, and short-for-a-marksman attack range. Overall, Akshan struggled to win in his first few patches with a Day 1 winrate of 39%, which only rose up to 45% by Day 7. We’ve been slowly adding power to Akshan as he continues to settle in, though one important callout: We won’t be seeing Akshan this year at Worlds, since he released late in the season. We’ll be keeping a close eye on his pro play performance in next year's spring split to see what changes he may need for that audience.

9242021_StateofGameplayArticle_Vex.jpg

And finally we have our newest champion, Vex, the midlane shadow-magic-wielding Yordle. We originally envisioned Vex as an artillery mage, but pivoted to a closer range anti-mobility playstyle instead (redoing the VFX of every missile ability in League was too high of a cost). She hasn't been released at time of writing, though she should be live by the time this blog goes up. I’m really happy with where she ended up and look forward to hearing what you think of her and her shadow.

Finally, I want to share a discussion we're having internally. By the numbers, new champions have released pretty balanced in 2021, but we hear many of you telling us that new champions are still OP. Clearly, something's going on that isn't captured just by winrate.

First is that when we introduce new mechanics, like Zoe’s summoner spell stealing, Akshan’s ally revive or Sylas’ ultimate stealing, they can look overpowered at first glance. First-look pieces like Champion Trailers are meant to get you excited to play the new champion, and showcasing all the ways you can fail or opponents can take advantage of counterplay isn't exactly hype-building. But we acknowledge that we're basically leaning into that risk of OP feels.

Second is it can be hard to understand how to play against new abilities or kits with a lot of moving parts, even if the sum of those parts isn't technically OP. For example, we saw with Aphelios that players didn’t know how to play against him. We didn't need to make his weaknesses more meaningful—we needed to make them clearer. Since then we’ve added new design guidelines around the playing-against experience, and while we’ll continue to still make complex champions from time to time, we're committed to making their counterplay clearer. Our newest champion, Vex, has a kit with clear counterplay; and we strive to continue in that direction especially for higher-complexity releases down the road.

We’ll have a new Champion Roadmap releasing shortly, so look forward to Reav3 telling you more about our future plans.

Competitive Gameplay & Behavioral Systems

9242021_StateofGameplayArticle_Compet_Behavioral.jpg

These spaces have been slowly converging over time, which reflects the kind of feedback we hear from you. Game-ruining behavior isn't just game-ruining because it's unpleasant to experience—it damages your chances of actually winning games, which takes us into the Competitive space. We like to look at this space as an ecosystem of around-game experiences that directly influence Gameplay before or after you’re actually in a match.

Dodging

I know this one is controversial. We treat dodging as a problem because one person's decision not to play out a lobby throws nine other players back into queue; especially at high MMR where queues are already long, this leads to "play" sessions where you're in queue or champ select for hours and only get one game in. (This also literally costs players money in regions where PC cafes are prevalent.) But at the same time, some players feel dodging is their only option when lobbies are being held hostage—when a teammate straight up says "I'm going to int if no one dodges". Neither of these are experiences we want for any player, but we don't think challenges in one area should prevent us from making improvements to the other.

There's also a gray zone you and we may never agree on. If a teammate is on a losing streak, or picks a champion they have a low win rate on, or a champion we just nerfed, or your comp has bad matchups, we believe you should still be expected to play that match out, but some players disagree. To us, champ select is a skill test that can set your team ahead or behind even when you can't control the actions of your teammates, like how many in-game skill tests also rely on being able to coordinate with those same teammates.

Philosophy aside, here's where we're at with results: We recently upped the penalties for players who dodge 3+ times a day, leading to a notable increase in successful lobbies. Which is good! But reducing dodging requires a more systemic solution than just queue time punishments. The reality is that dodging feels like a winning strategy at high tiers of play, and in order to change that we have to address the underlying causes that encourage it. We believe this requires a shift in how the lobby experience is presented and plays out. I’m leaving this vague for the moment because we’re still working on designs. But we’ve heard your feedback and are investing more in time and energy into addressing dodging systemically in the coming months.

Matchmaking

We’ll never stop working on matchmaking until we get it right. So I can safely say we’ll never stop working on matchmaking. Here's what's going well this season:

  • At all ranks, players are getting their Primary role more than 90% of the time
  • At all ranks, matchmade teammates are within less than a full tier of each other in 99% of games.
  • At all ranks, queue times have decreased relative to 2020

We’ve also seen positive results on removing smurfs (or those that are significantly misplaced) from the general play population, but there are a few specific cases that interact oddly with our "did we tag your skill level correctly?" detection. We'd like to dig into questions like:

  • How does when you start playing Ranked in the season affect where youre placed or how you're matched against others?
  • How are players that have taken a break from League represented in the system when they return?
  • What behavior drives players at the top of the ranked ladder to decay quickly, causing them to be wildly misplaced?
  • How well are we matching new players against other new players? And, what does it look like when a fresh level 30 joins players who are statistically at the same skill level—but have been playing for years?

All these point us in the direction of better calibrating what qualifies as smurfing and how it impacts matchmaking. There’s certainly more work to do here to ensure that League is engaging and fair, especially if you're coming back from a long break or on a win-streak.

Disruptive Behavior

Over the past year we’ve made some short-term improvements to our Player Behavior systems. We’re now alerting when a player is punished after you report them, so you have better visibility into certain punishments. Our automated systems catch and punish more AFKs and Inters than in years past, and we want to make them even better. Our ideal state is one where every player is playing to win, and no one experiences harassment.

One thing we haven’t talked much about is that all of Riot's games share this goal, and we’ve been building a centralized technology to help us get there. (It's not technically a League of Legends topic so we weren't sure if you'd be interested.) This centralized system is taking time to create, but we’re over a year in and expect to start seeing the results of the new capabilities soon. These could include better voice communication options, better and faster detection of hate speech and griefing, and more flexibility in how we issue penalties. 

The reason we're bringing this up is because we can only do so much with the old reporting system, although we do think we can still draw out a little more value in the short term. First and foremost is continuing to iterate on the reporting confidence work we’ve done—which increases our automated systems' confidence in whether on-the-line behavior should be penalized, and whether that on-the-line behavior is actually just griefing. Continued level-ups here will help each report you issue make a bigger impact.

Two more things I wanted to quickly mention. First is Honor. This feature has been falling short for a long time. We’re currently discussing options for revamping this system into a more effective one in 2022. Second, in case nobody has ever tried this, let me ask straight up: please be cool to each other. Play to win and be the kind of player you want on your team (problem solved now, right?)!

Preseason

9242021_StateofGameplayArticle_Preseason.jpg

As we initially discussed, this year's preseason gameplay changes will upgrade existing systems rather than overhaul one specific system from scratch. We want to give you a preview of one of the specific systems that we have some very exciting additions planned for: items.

New Mage Items

The first new mage item is a defensive Mythic in the Lost Chapter space. Some mages struggle to make an impact around the map because they can be killed so easily, and Everfrost settled more into the role of locking enemies down than keeping them off you. This new Mythic will guard mages by reducing incoming champion damage for a brief period of time when they're first attacked. Some of its combat power falls off after the protection expires, to ensure this item is best in the hands of champions who sling spells from a safe distance rather than ones who trade blows in closer-range brawls.

The second new mage item is a legendary that'll help mages tear through the shields of their enemies. It increases magic penetration against low-health and recently-shielded enemies, and should be enticing for burst mages that want to start snowballing early in their build.

New Tank Items

Support tanks are often pinned to Locket of the Iron Solari (and therefore its defensive strengths). We want to give support tanks a more offensive Mythic that rewards them for getting into the middle of the enemy team and causing havoc. When the user immobilizes an enemy champion, all nearby enemies take increased damage. If this sounds like Mega Abyssal Mask.... that's because it is. Abyssal Mask itself will be changing to avoid the overlap, but no spoilers for now.

Looking beyond just bot lane, another common refrain we've heard is that there aren't enough mana options for spellcasting tanks. We're adding a new Legendary tank item that gives users a large chunk of mana and a new effect that allows them to dip into that mana for additional shields every time they immobilize or slow enemies. Like most mana items, we expect it to compete for the first or second Legendary a tank buys after their Mythic, before late game mana pools come online.

New Assassin Item

A lot of today's Legendary-tier assassin itemization is focused on concentrated burst damage, and we wanted to create an option for users that are interested in more frequent spell rotations rather than singular burst output. This new legendary comes with a chunk of ability haste and a new effect that refunds some ultimate cooldown whenever an enemy champion dies within 3 seconds of you damaging them.


More on Preseason soon, with the last LoL Pls of the year!