HEROES OF THE STORM DEVELOPMENT UPDATES [@iHeroes]
@iHeroes
As many of you are aware, there’s been a lot of feedback and community conversation about topics like matchmaking, ranked play, new Hero balance, and player toxicity. It’s clear to us that some players in the Heroes community are unsure if their concerns are being heard. We take your feedback seriously, and we definitely haven’t done enough lately to acknowledge and respond to it. We can do better, so we’d like to shed some more light on a few of these topics today, starting with Matchmaking.
MMR and Matchmaking
Our goal with matchmaking is to ensure your Heroes of the Storm games are fun, fair, and challenging. While we’ve made progress in some areas, we don’t think we’ve hit that goal in all of them. We’re working on some matchmaking improvements that will bring us closer to our objective and hopefully give players more positive gameplay experiences.
Queue Time vs. Match Quality
Matchmaking in Heroes boils down to striking a balance between creating the best possible matches in terms of player skill, and initiating games in a reasonable amount of time. Currently, the matchmaker is always seeking to make the best possible matches it can based on a set of criteria it uses to find equally skilled allies and opponents. The longer players wait in the queue, the looser these restrictions become in order to get players into a match as quickly as possible. We originally set up these somewhat elastic matchmaking rules to speed up wait times. Unfortunately, decreasing wait times sometimes comes at the expense of match quality. This can be especially noticeable at the highest skill levels, where there are fewer players in the queue at any given time. When this situation occurs, players can find themselves in games that include allies and enemies outside of their skill range.
Given all your feedback, it’s clear we’ve been favoring reducing queue times a little too much. Moving forward, we’re going to harden these rules in order to enforce higher quality matches. As a result of the changes we plan to implement, higher-skilled players may see longer queue times in exchange for higher quality matches.
MMR and Rank Decay
Your matchmaking rating (MMR) is an internal numerical representation of your skill that is adjusted based on the outcome of the games you play. This value is most accurate when you regularly play within the current meta against other active members of the player base. When players take an extended break from the game, or only play a few ranked games each season, it can create situations where a player’s current skill no longer matches their MMR (if only temporarily). We’ve heard from those of you who feel it’s unfair to be matched with players who are in this situation.
To alleviate this, and to help improve the accuracy of a player’s MMR and rank, we’re working to institute both MMR and rank decay. Players who have not played any games for a while will have their MMR slowly adjusted to compensate. Likewise, players who do not regularly participate in a given season will see a decline in their rank points over time.
All of this said, it’s also very important to us that the decay rate is not overly punishing. We’re still working on the specifics of this decay system, and we’re interested in hearing any feedback you might have.
Performance-Based Matchmaking
Performance-Based Matchmaking is a system we introduced late last year, which aimed to improve MMR accuracy by taking your individual performance into account when adjusting your MMR. Its release unfortunately coincided with some unrelated issues that occurred during the new ranked season. As a result, we disabled performance-based matchmaking so that we could focus on addressing those issues. While it was active, we absorbed your feedback and took note of additional improvements we wanted to make.
When we bring this feature back, we intend to add additional context about how point adjustments are determined. This should help you better understand how you’re performing and where you can improve in the future. Keep an eye out for those details in the near future.
We also received feedback that performance-based matchmaking isn’t as effective at the highest levels of ranked play. This subset of players is relatively small, which can result in too few players setting the standard for in-game performance in specific situations. As a result, when performance-based matchmaking returns, it will no longer adjust point totals for Grandmaster and Master league players.
Though we had disabled the performance-based matchmaking system’s ability to adjust players’ ranked points, it has continued to gather live player data and improve upon itself over time. We believe it is ready to be turned back on soon, and we’ll have more details to share with you in the coming future.
Ranked Play
We know that many of our most dedicated players have been curious to hear about any updates we might have in store for ranked play. We are working on several improvements for ranked, and we’d like to share a few of them with you today.
Placement Matches
Placement matches were primarily intended as a way for us to gain more information about your skill level in the absence of previous match history. However, the more games you play during a season, the less necessary it becomes to reassess you at the start of the next. Going forward, those who play enough ranked games during the current season will only need to complete as little as one placement match when the next season begins. After completing your placement, you will be placed based on your MMR at the end of the previous season.
New and Returning Player Seeding
The first time a player steps into ranked play, we use the knowledge we have about their performance in other game modes as a general starting point for assessing their skill. We also do this if a player hasn’t played ranked in a long time, and their MMR in other modes is now significantly different than their existing ranked MMR. Currently, players who enter ranked play for the first time, or after a long absence, can place as high as Diamond 3 after completing their placements.
We recognize that being placed highly when a player is new to ranked can lead to some less than positive experiences if they lack some of the more nuanced game knowledge necessary for success in that league. As a result, new players can no longer place higher than Platinum 5 at the end of their placement matches going forward.
Additionally, we believe the introduction of MMR and rank decay will help account for players who take long breaks from ranked play. As such, these returning players will no longer be re-seeded into placements based on their skill in other game modes.
Hero Bans and Swaps
We’re going to add a third ban for each team during drafts. We’ve been considering this addition for some time, but we wanted to wait until our Hero roster grew to an appropriate size. We believe we’ve now reached that threshold, and adding a third ban should provide teams with another layer of strategy when choosing their Heroes for a given match. We’re currently planning to add this ban to the existing mid-ban phase rather than create an entirely new third phase. This will enable us to keep drafts moving quickly, so that players can get to the action. It also allows teams to use that additional mid-ban to react to the draft as it unfolds.
Second, and more nebulous, is the addition of a Hero swap feature. We’ve seen a lot of mixed discussion on this topic in the community, and there are many things to consider when introducing this system to Hero League, including increased draft time, complexity, and player conflict during the draft. Because of these complexities, Hero Swaps are lower on our priority list compared with some of the other improvements we’re discussing today.
Despite these concerns, we’re still exploring a few options. We’ve considered alterative ideas like position swaps, in which players could rearrange their team’s pick order. We’ve also considered simply using the same “first come, first served” pick method that exists in Team League and Custom Games, and we may bring some of these ideas to Unranked Draft as a test.
We’d love to hear your feedback on this topic, so let us know what you think!
New Hero Balance
We do our best to provide fun and satisfying experiences with every new Hero we release. This goal can take many forms, from offering exciting new abilities, to bringing the fantasy of a character to life, and making sure they’re fun to play as, with, and against. It’s also important that each new Hero feels impactful in players’ hands and has the potential to shift the meta. We’d like to share how we look after our new Heroes in the days following release.
New Hero Balance Changes
We received a lot of feedback from players who felt that both Maiev and Fenix were too powerful on release, and we agree. In the past, many of our Hero releases have been considered balanced, or even weaker than they should be. When a new Hero is weak, it tends to see minimal competitive and ranked play, which limits how much live Hero data we’re able to quickly gather and doesn’t evolve the metagame. We want new Heroes to feel compelling for players at launch, but we also want to do that while maintaining the core gameplay balance of Heroes of the Storm. We’ll continue to work toward improving on this going forward.
Over time, we’ve developed a cadence of releasing balance updates for new Heroes every two weeks. This gives us enough time to collect data and allows players to learn how to play with and against the latest additions to the Nexus.
Occasionally, we’ll release a balance update for a new Hero outside of this time frame, as we recently did with Maiev (and Zarya even further back). This isn’t ideal, as it doesn’t give the community enough time to actually explore a character’s strengths and weaknesses aside from initial gut reactions. It also means we have to make these decisions with very little data and live play experience. Our goals are to avoid knee-jerk reactions, to give players enough time to learn new Heroes, and to gather data before we make any balance adjustments.
New Heroes in Ranked
We understand that some players in the community would like us to lock new Heroes out of ranked modes for a couple of weeks following release, and we’ve discussed this internally. Despite this concern, Hero League data is critical for us to fine-tune a new Hero and hit their two-week patch timing. We hope the addition of a third ban in ranked play will help relieve the pressure when expending a ban on any new or reworked Hero that may be considered too strong.
Toxicity and Reporting
A competitive game of Heroes can really get the blood pumping, but we understand that just one player’s negative behavior can ruin the fun for all 10 players in a match. We want to make this clear: Toxic behavior will not be tolerated. We’ve been working on improvements to the ways we detect negative behavior, mitigate detrimental effects on others when it occurs, and message when we take corrective action.
Ramping Up Actions
We’ve been working on a new system that uses machine learning to help improve the validation of player reports. Our internal testing shows the new system allows us to take action against many times the number of reports than we could using our previous methods, while also maintaining an even higher level of confidence in those actions. We’re still doing some additional tuning, but we expect to be able to roll this new system out in the very near future.
Repeat Offender Penalties
Along with the improvements around detecting toxic behavior, we’re also revisiting punishments for repeat offenders. Currently, repeated abusive chat penalties result in an escalating silence duration. This has proven to be enough for most players to get the message, but a subset of players continually return to their toxic behavior, continuing to ruin games for everyone around them. Rather than simply issue them even longer silences, we’re going to start escalating penalties from silences, to suspensions, and even bans as repeat offenses stack up.
Non-Participation and Intentionally Dying
Going AFK during a match, leaving matches early, and dying intentionally are all behaviors that harm the game and compromise a match. We’ve been working on ways to address these behaviors while limiting the negative impact they have on others. As part of this endeavor, we’re deep into development of a loss forgiveness feature, which will mitigate ranked point deductions for players who lose a game due another’s failure to participate. The leaver will be penalized for the loss and repeated offenses will result in harsher penalties. The system will initially only forgive losses caused by leavers, but we plan to expand it to include AFK players and intentional dying in the future.
Communication
One of the challenging parts about combating toxicity is that most of the action happens behind the scenes, so it can seem like nothing is being done. A few months ago, we started sending e-mails to players when their reports on non-participating or intentionally dying players lead to action being taken against the offender’s account. We eventually want to bring those notifications in-game where they’ll be more visible. In the meantime, we’re going to increase visibility for the suspension and ban waves we issue by bolstering our broader communications. This is something we’ve done sporadically in the past, but we’ll be moving to a more consistent cadence going forward.