/dev diary: The Music of Arcane
We sit down with Christian Linke and Toa Dunn to dive into the music of Arcane.
Welcome to the second of a four-part blog series from the Summoner’s Rift Team! Last year, we ran the Life of a Patch campaign, sharing insight on how we put together a League of Legends balance patch (here's a summary!). This time, we're going to tackle a few major, persistent balance topics in more depth than we can get to through patch notes, social media, or Meddler and Scruffy's weekly posts.
Here's what you can expect over the next several months:
New champion releases and VGUs are unique because, unlike general champion balance, we don’t have millions of games worth of data indicating how strong a champion is. Tracking patch-over-patch win rates takes time, and during this window, players are still learning the ins and outs of the champion. This means that a champion’s early win rate alone doesn’t necessarily reflect their “real” power level.
We use a champion’s projected mastery curve to help set balance targets for new releases. Mastery curves are a way of looking at how much more effective a player becomes on a champion as they play them more, as represented by how much their win rate increases over games played. They look like this:
Even an easy-to-learn champion with a low mastery curve will likely increase in win rate by 4% over the first ~15 games played. Most of the time, that learning curve extends for many games beyond 15, so players learning new champions will sometimes continue to see growth up to (or beyond) their 100th game—champions like Yasuo, Katarina, and Nidalee all have a long tail of continued mastery.
Because we know that a player’s win rate on a champion is going to improve with games played, we have a difficult balance decision to make: Do we release new champions at what we believe to be their long-term balanced power level, leaving them weaker than the rest of the pack while players progress along their mastery curve? Or do we release them artificially powered up for short-term balance, and then nerf accordingly as players master them over time?
For the last couple years, our approach was to release champions that were balanced for the short-term. A couple months ago, we decided to change this approach:
We now release new champions and VGUs at what we believe to be their long-term balance state.
This means they’ll likely seem weaker at first while players learn new mechanics and builds, but they won’t have to undergo nearly as many nerf cycles or mechanics adjustments following release.
Let’s dig a little deeper into our previous approach and why we decided to change it.
For a while now, we’ve released new champions such that their win rates were on par with other champs once they’d been live for two weeks. For some champions—like Mordekaiser and Neeko—this was pretty reasonable, as they don’t have massive mastery curves. (And in Neeko’s case, there was even a pronounced “playing against the champion” learning curve, where we actually saw her start to lose win rate after her first week as people learned how her passive and other abilities worked.)
We decided to release new champs and VGUs that perform on-par with the rest of the champion pool for a few reasons:
In the past, we’d released champions that won well below 40% of their games at launch, and there was definitely some player outcry. Bard, Ivern, and even Ornn presented unique playstyles that were hard to grasp immediately—in other words, they all had (relatively) steep mastery curves. As a result, players sometimes assumed that a champion was a “failure” due to their weak release state. We were worried that releasing a champion that was perceived this way could hurt the long-term playrate for the champion, but we no longer believe this to be the case.
Next, this approach allowed everyone to get their new (or updated) champions and start playing them immediately without necessarily compromising the experience of everyone else in the game. Even normal games can feel ruined when you put in a champion winning less than 40% of their games. (But as we’ll later address, releasing champs who are too strong results in a host of other game quality issues.)
The pain of “this champion is so weak” is particularly felt with VGUs, since players somewhat expect to understand the champion even after we’ve changed some of (or much of) their kit. By giving an updated champion “training wheels” in the form of temporary strength, it helped players feel as if they could still find success on their favorite champion, even as they’re remastering them.
Releasing new champions and VGUs that are balanced around short-term success does help them feel more viable as the player base progresses along their mastery curves. But ultimately, we felt this approach was causing more harm than good.
We’ve recently seen a growing perception that new champions are too strong out the gate. Just as we were worried that champions being weak on release would lead to long-term perception that a champion is bad, the perception of “X champion is OP” has a lot of staying power and persists long after nerfs and mechanics changes.
But beyond perception, we believe releasing champions that are balanced for the long-term is healthier for the overall state of the game. Let’s jump into why.
Starting with Fiddlesticks, we’re now releasing champions and VGUs in what we believe to be their long-term balance state. Let’s talk about our hopes for this new system.
We don’t want players to worry about us releasing new and updated champions that consistently dominate like they have in the recent past. Without a pressure to make these champs win at the same rate as the rest of the champion pool on release, we can better ensure they’re not running away with games as players become more experienced on them. Champions winning 55% of their games within the first patch shouldn't happen, and honestly, new champs winning even 45% of their games over the first few days could indicate a long-term balance issue if they have a long tail of mastery.
On the flip side, new champions shouldn’t fail more than they need to. If we have indication that champions truly are weak on release, we’ll still react with buffs, as we did with Fiddlesticks and Volibear.
The previous approach wildly benefited players who wanted to grind out mastery on new champions. If you got 30 games in on a new champ during their first patch, you were massively outpacing the average player. So while win rate data might show these champions winning 50% of their games, the heavily-invested grinder was undoubtedly winning at a much higher percent.
When it comes to solo queue, the above behavior isn’t that problematic. It exists for most dedicated one-tricks—you can get an advantage by diving super deep into mastering a champion. But this same concept extended to pro players becomes a bit more problematic: They're encouraged to put hundreds of games on these new champions not just to give their team an advantage, but to keep up to speed with their competition. So if we shortcut the mastery curve on a champion by making them stronger than they should be long-term, the near-guaranteed result is that pros will master them, and then that champ will be stronger than other options in pro play.
Ideally with our updated approach, new champions will no longer become an instant staple of the professional scene.
The next goal benefits players who are hoping to become masters of the latest new champion or VGU. By releasing champs at their long-term balance state, we should no longer need to continually modify and nerf new champions to “make up” for their stronger initial release state.
Player sentiment is a factor here as well. We’ve been hesitant to heavily nerf new champions that look close to balanced (or only mildly overpowered) in their first patch or two, even when we suspect they'll still continue to climb a lot. To players, heavy nerfs here would come off like we're holding new champs to unfair standards compared to other champs who share similar win rates. This eventually leads to patch-over-patch nerfs, which is a bad experience for those playing these champions, as they constantly have to adjust to balance and mechanics changes. It’s also not great for other players in the game, as they have to keep up with the speed of the changes.
By aiming for long-term balance, we hope to avoid the extended cycle of nerfs that often persist for 6–12 months after the release of a champion.
There are many reasons we may change a champions’ mechanics, but generally with new champs, they fall into one of two categories.
This first is that we realize a mechanic is just not particularly healthy for the game. In a perfect world, this never happens. But sometimes, a problematic mechanic only becomes clear once the biggest group of playtesters (all of you!) get their hands on a champion. A recent example of this would be Akali’s true invisibility. Adjusting the way we approach new champion balance doesn’t really address this type of issue, and we may still need to remove patterns that prove to be unhealthy.
The second reason we may adjust a new champion’s mechanics is because we later realize that a tool we gave them to help soften a weakness isn’t actually necessary for a champ's long-term balance, as evidenced by players progressing along the mastery curve. For instance, Akali’s healing on her Five Point Strike (Q) was a tool to help her stay in lane, especially for players who hadn’t mastered her other patterns. But once players mastered her, that healing made it near-impossible to force her to back, even before she'd begun to scale and purchase a Gunblade. Now that we're releasing champions at long-term balance, mechanics will be put under more scrutiny to determine whether they're truly needed for a champion's kit to function.
Hopefully this will reduce the likelihood of removing a mechanic you love at a later date.
Our expectation for this new approach is that we can reach a truly balanced state within one patch after we have Pro data without substantial follow-up afterward. This obviously doesn’t mean there won’t be future buffs or nerfs, but the expectation is that new champions and VGUs will only show up in patch notes as much as an average champion. One exception is that a champion designer may do follow-up mechanics changes for a few patches after the champion goes live, even if they’re perfectly balanced according to the new approach. Fiddlestick’s changes to Harmless Scarecrow in patch 10.11 is an example of this.
While we believe this approach is better for overall game health, there are still drawbacks:
Champions with unique playstyles take time to learn, and we fully expect to hear some “this champ is garbage” along the way. This won’t affect all champs equally, as lower mastery champions with more familiar mechanics will avoid some of the struggle. But it will definitely be a feels bad (both for us and players) when we’re not being more responsive if we believe a champion is long-term balanced, even if their win rate doesn’t show it yet.
Sorry, ((REDACTED)) mains.
Without the power advantage gained by quickly learning and mastering new champions, it may take longer for them to appear in the professional scene. Having less experience playing as (and with) a champ who’s at the same power level as the rest of the roster means there’s less of an obvious incentive to pick them. That said, we think it'll be a relief for players to not have to worry as often about their champ being heavily nerfed due to pro play. We're also excited for the prospect of only seeing a new champion in pro once teams have figured out how to strategize around their unique strengths.
We have two VGUs under our belt following our new principles—Fiddlesticks and Volibear—and will be shepherding our first new champs through the updated process soon. While there have been some bumps in the road, we’re confident in continuing along this path as neither Fiddle or Voli have needed repeat-nerf patches, nor have they lost any major mechanics. The next area we want to improve is getting better reads on predicted day one win rates so new releases aren't weaker than their long-term balance state.
Volibear and Fiddlesticks had medium to high (respectively) relearning curves, but our next couple of champions may put us (and you!) to a more challenging test. Our upcoming dreamy jungler and masked assassin are both likely to appeal to players looking for more novel experiences and require a pretty heavy investment to get over the initial learning curve. While we’re confident that mastering these champions will be well worth the effort, there will likely be some bonus “feels bad” feels, especially in comparison to simpler future champion releases.
As with most of our principles, we will actively iterate to improve upon this process such that we can minimize player pain and create a more positive overall experience with new champions and VGUs.
We hope this blog series gives you more insight into how our teams approach the challenges of balancing League, and hopefully this article helps give our most up-to-date information to continue the balance dialogue with us and other players. We'll be back in August with our next post, covering how we approach balance for major pro tournaments like Worlds!