Summary so far:
- In order to have a better system reconnect and replay and spectate, does the game need a "new engine", or is it otherwise impossible to make those systems significantly better?
There is some validity to what you are saying, but I don’t think “impossible” is the right word.
Heroes of the Storm uses a peer-to-peer networking architecture. This architecture offers a lot of benefits, but also makes certain things more difficult. Unfortunately, reconnect is one of those things. Having said that, it is not impossible to make improvements to the system. We’ve actually improved the system on a couple occasions since the original launch of the game. But we’re absolutely with you guys that the system is far from perfect, and could use more love.
This again comes down to prioritization. The same people who would work on improving the reconnect system are the same engineers working on matchmaking and ranked improvements. After looking at the reconnect stats, we decided that matchmaking and ranked system improvements would benefit far more players right now.
Regarding the pause idea: We do have the ability to pause the game which we use for esports scenarios. So far we've decided that this would be too disruptive to use in normal games if a single player had a bad connection/computer.
- Did you ever consider a full MMR reset with all the issues and effects coming with that?
We tend to consider everything when looking at solutions and, yes, we’ve considered what an MMR reset would mean. Man…it’s UGLY. The utopian view is that a reset would be a short period of utter chaos where everyone starts out equal and is essentially tossed into a giant thunderdome where the weak are slaughtered by the strong until everyone is sorted properly.
More realistically, it would be an extended period of utter chaos long after placements as those placement games would be almost completely arbitrary. With no starting MMR to use to match players up, it would be entirely luck-of-the-draw for team comps and where you end up after placements would come down to chance more than anything.
From there, the ranks would have to slowly sort themselves out as the GMs who ended up in silver/gold due to being matched repeatedly with teams full of bronze/silver players dominate those games where the bronze players who found themselves in platinum due to being in games filled with masters end up throwing most of their games as they slowly work their way back down the ranks. In the process, the GMs are inflating the win rate of the low rank players they’re playing with and the bronze players are tanking the win rate of the ones they’re playing with making it more difficult for everyone to end up at their deserved rank.
In short, it would be expected to be an awful experience for everyone.
- Quick Match has been a hot topic in the community lately. Whats your take on it?
Quick Match is still our most popular game mode in Heroes of the Storm by far. We don’t think the right move is to shunt Quick Match aside, we would rather improve upon it. The problem that we see is that team compositions rarely feel balanced or competitive. In the blog post that we posted yesterday we discussed making a shift away focusing on getting players into matches quickly, and instead to focus more on getting players into better quality matches. That is absolutely applicable to Quickmatch and serves as the plan of action here.
The first improvement we’re going to make is to strictly enforce role compositions for the matchmaker. A team will not be constructed without a balanced team composition on its own. This means that assassins or specialists may potentially have longer queue times, while some tank-capable warriors and healers may have shorter ones.
- Does this mean every quickmatch team will, at some point in the future, have at least 1 tank-warrior and at least 1 full healer, with 3 flex picks?
Blizzard's answer
In our current thinking we believe that it'd require 1 tank-warrior, 1 healer and 1 assassin, so 2 flex picks. Pending testing and the queues not erupting in molten lava, yes. I think it likely that initially this will be the new standard for the vast majority of QM compositions.
The plan then is to follow up on this change by adding more incentives for queuing as roles that are currently underrepresented in the matchmaking queue which will then make this a hard requirement for every team.
- An API for individual and global statistics has been another topic since forever
Releasing a public API has been something the dev team has also wanted for a long time. We have an initial version of this API working internally that is used by our HGC and Heroes Game websites. The problem is that it isn’t complete, and it isn’t setup in a way to support the needs of a public facing API (i.e. reliability).
Unfortunately, the same people who would work on finishing this public API are the same people that are working on features like improving matchmaking and the ranked game mode. Because of that, we simply haven’t been able to justify prioritizing it above those features. As of right now, we don’t have engineers working on this, and until we get through a lot of the player facing improvements (matchmaking, ranked, etc), we wouldn’t prioritize this.
Regarding our philosophy of what we’re ok with having in an API, I would say that we’re ok with having almost everything you can see in your own in-game profile. On the match history side, we’d like to include basically every game stat that we track for a match. Everything you see on the scorescreen, and probably a bunch of stuff that we track but don't show on the score screen. We’ve also ok with even adding “aggregate” stats across the whole player base like hero pick/win rates, talent pick/win rates, etc. Again, this is the philosophy of what we’d like to get in, now it’s just a matter of when we can dedicate resources to it.
- "Teaching the community" is another big topic lately. Do you have thoughts towards that that you can share?
There are some big improvements we can make in the realm of educating our players and preparing them for competitive play. One thing we’re working on currently is a revised Hero Selection system that provides far more detail on things like their playstyle and main mechanics behind different heroes. This is something that is still a little ways out, but it will help players identify weaknesses in their draft composition and potential for synergies and counters.
As we bring the Player-based Matchmaking system back for ranked play we’re considering new ways to contextualize player performance based upon the stats we currently examine. It’s important that we celebrate areas where players have been successful, but even more important to identify opportunities for improvement.
Beyond these things we’re looking for more opportunities to provide actionable feedback to players through the end of match sequence. Right now we’re considering an “XP Missed” stat that tracks missed team opportunities for experience generation. This should help reinforce the importance of laning. There are places here for improvement, and if you have suggestions we’re happy to consider them.
We’ve also been creating additional educational content on places like the Heroes of the Storm esports website for some time now to help expose players to things like pro player insight. Every time a new hero is released now, for instance, we work with an HGC player to create a first impressions build guide that you can find on our HGC website. Here’s the latest one for Fenix, or example: https://esports.heroesofthestorm.com/en-us/news/21659396/fenix-tips-from-yoda. We’re also starting to do more things like Hero spotlight reworks to keep all our players better informed of incoming changes to heroes’ balance or playstyle.
- Updates on classes, and what about quests that require specific goals in game to try and teach things?
In terms of new roles, this is something we want to do. Our current design leaning is to keep it to your first 6 (Tank, Bruiser, Healer, Support, Melee & Ranged Assassins). We’ve always felt that the Specialist category is a little weird. In the best cases, a lot of characters like Nazeebo and Zagaras fit nicely into the Assassin categories. That does leave some oddities, such as Lost Vikings potentially fitting into the “Support” category (they do provide a lot of indirect benefit for your team). Curious what your thoughts are on this, and the rest of the communities thoughts, surrounding placing these odd heroes into more defined categories.
In terms of timeline for this: it is absolutely something we want to do. We think it’s correct for the game moving forward. Currently our priority is around improving the matchmaking and ranked experiences though, so this will be on hold for a little while longer.
For Quests: Many years back, we had a design very similar to this (most likely internal only?). We ultimately moved away from this as it encouraged players to farm out matches they had already won: “I need another 10k healing to finish my quest, don’t end the match!”. We also tried some quests such as “Land X Hooks” but that also encouraged weird player behavior. Ultimately, we stuck with a few simple quests of essentially “Play The Game” but with different requirements. We would love to update the quest system in the future and maybe we can explore some ideas here that could help teach players, as you suggest.
For updated data on bosses, scaling, etc. This is something the balance team has internally. I’ll make a note for us to send out the current numbers either here or somewhere else visible to the community. I will say things like Bosses and other mercenary camps rarely change unless we have a specific design or balance reason that we’re trying to address.
- Has there been a conscious design decision to give recent heroes a lot of tools for their kits (waveclear, mobility, etc) to make them more generalist heroes vs. designing them to excel in very niche situations?
The short answer is yes. For a more detailed explanation:
In the past the design team went through a phase where we wanted to make sure every hero had a really sharp role. Internally we discussed this a lot, but we wanted to make sure that there was a unique gameplay reason to pick every hero at some point. As an example, we wanted there to be a unique decision point of why to pick Tychus over Valla, or vice versa, without players simply choosing whichever character has a higher win rate. This is ultimately what led to Tychus’ updated design that he deals bonus % damage on his base kit. We designed his role to be strong against lots of high health targets. We also did a lot of this on our Healers: Malfurion wasn’t supposed to have burst healing as he was designed as a sustained healer, Uther was intended to be a weak sustained healer but strong with burst, and so on. For the healers, we looked at not just their healing mechanics but everything on their kit.
Ultimately we ended up backing off from this. We found that it made the game feel like there was a lot more hard counters and you could win or lose in draft. We still want to have good design and gameplay reasons for drafting each hero in a game, but we’re letting those be a lot softer.
- Details on third ban and and putting it at the start of the draft?
We’ve been discussing adding third bans since around the middle of last year. The idea to place the third ban in the mid-ban phase came from feedback from both the community and the pros. The initial requests for the third ban started coming up because there were now enough heroes in the pool that could fill similar roles that a counter-ban in the mid-ban phase was losing effectiveness. Adding an additional ban at that point would bring teeth back to mid-draft bans.
Taking a step back, the core idea is that the mid-ban is a strategic ban based on how the draft is unfolding, while the first ban is primarily a meta ban. At the highest levels of play, some strategy comes into play during the first ban phase, as HGC teams have done significant research into their opponents, but for most players, the first ban gets used to remove whatever hero tends to be on top of the meta at that moment. In higher ranks, it gets a bit more variety as it will sometimes be the hero that is on top of the meta for that battleground, but that’s not the typical situation.
That being said, sentiment shifts over time, and we’re open to revisiting this and adding the third ban as a first-ban instead. The feeling is that doing so would be mainly addressing a more short-term concern, the release of overpowered heroes, where an additional mid-ban is more interesting long-term. We’ve been watching the community response and are interested to see if folks still feel strongly about adding it as a first-ban after seeing the reasoning behind the mid-ban.
- Any plans to update the Observer UI?
Yeah, we agree that our current observer UI is long in the tooth. As you mentioned we’ve wanted to bring stuff like the current in-game party panel into the observer UI. As we’ve gone on, our observer UI has diverged from the normal UI a bit, and it’s part of the reason why you’ve seen some bugs in the observer UI lately. That’s no excuse and we need to fix it, but I just wanted to provide some context.
The current plan is to better unify the in-game UI and the observer UI. We’re going to be using the same party panel UI that appears at the top of an actual game. Once we have that unification, we can start to consider more functionality that would benefit esports viewers.
- Thoughts on vision mechanics:
Vision is an interesting part of the game that we don’t really augment too often. We do some bonuses to vision that you mention, and have some heroes (Dehaka) who can also limit enemy vision. Overall, I think it’s a super interesting mechanic that can add fun and strategic depth when designed correctly.
In terms of power level, we typically start internally as a ‘feel’ thing, and then eventually we can use player data to validate those decisions. Many of the talents players are choosing from can offer a lot of sheer power. It’s hard to make a vision talent compelling to choose over 25% more damage, without at the same time making that vision talent feel unfair or unfun to play against. This is typically why we will add secondary bonuses to these abilities or talents. Vision can also be considered situational, so by adding damage as an example, we’re guaranteeing some baseline power on the ability/talent.
In general, I think we want vision abilities to alert the enemy that they have been spotted in some way. As an example, you can visually see Zagara’s creep, so there is likely to be a Creep Tumor near by. For Lunara’s Wisp, this is an interesting one. I think we could alert you that a Wisp has spotted you, but it does lose some of its power as a result. We are working on a Lunara update that has some changes to her Wisp and related talents, so I’m looking forward to seeing how that plays out. We can revisit any of these rules though!
- Why do some heroes seem to get vision or mobility abilities that are core to their kit and others seem to just be flavorful throw ins?
Just to touch on this quickly: We absolutely involve a lot of the design team as early into the process as we can. The live/balance designers are paired with a hero designer from the beginning. One thing the balance designers will try to surface early on in the hero design process is if they feel they have tuning knobs for each talent and ability. In the past we might design a talent that simply increases charges from 2->3 on an ability. If this is too strong, we don't have a lot of tuning knobs without adding a kiss/curse mechanic or something else.
- Why do lots of new heroes have AoE cleave auto attacks, where this was often a talent on older heroes? Any thoughts on changing this?
I agree that a lot of those recent examples have very strong AOE. In general, we try and look at the design of the kit and balance of the hero kit overall. Every hero will have some obvious strengths and weaknesses, but are hopefully balanced within the larger overall picture. For the Junkrat example specifically, I agree that his wave clear is insanely strong! Hopefully however his overall power is about in line with what makes sense. We of course have the ability to tune this down in a number of ways including the splash size, and the fallout of the damage (i.e. less damage further from the center).
I will say we will absolutely revisit old heroes and even these newer ones and make adjustments where needed. Nova has never been intended to have strong wave clear, so I’m not sure we will make changes there. It may be possible that we need to tone down some of the newer heroes wave clear, bring some of the older heroes up (Kael’thas in this example), or meet somewhere in the middle.
Hero Swaps and it's alternatives
Unlike a lot of the things we’re discussing here today, hero swaps aren’t a clear win for the majority of the community so while it is something we’re investigating, it’s a lower priority item.
To be effective, swaps would need their own phase, extending the time it takes for drafts to complete. They also open the door to additional toxicity and, although the core of the feature is available in custom games, it’s reliant on the players on the teams trusting each other. For it to come to other draft modes, there’s a significant development effort involved to add a lot of validation between players. You wouldn’t want someone grabbing your first-pick treasure without your permission, for example, which you can do with the implementation in custom games.
On top of that, for swaps to be effective, they require significant upfront communication by the players. That’s certainly something we want to encourage, but that also means the feature wouldn’t be used by a lot of players beyond the top end of ranked play.
So, right now, we’d rather focus our development efforts on features that will more clearly be beneficial to all players.
For the alternatives, such as trying out first-come first-served (FCFS) drafting which provides a similar benefit, we can do that relatively quickly. There’s some skepticism about whether FCFS can work outside of a team environment, though. We had the same concerns when we allowed 2s and 3s in team league and were pleasantly surprised by the results. It’s an area where we want to gauge community interest and if the feeling is that players would prefer FCFS over nothing, we’d try it out in Unranked Draft first.
- As of now, Team League seems to have a lot of issues to the point where barely anyone plays the mode. Do you have thoughts towards that?
This is a particularly difficult question to answer. Ultimately the population for this mode is lower than others due in part to it being difficult to organize a team to play regularly. Many of the ranked improvements we’re making will improve Team League without resolving this underlying concern. We believe that Team League needs some of the upcoming major social features, such as clans and group finder, to truly flourish. Since adding these social features would result in a significant update not only to Team League but to many other areas of the game as well, they’re still a ways out in terms of when we believe we can successfully deliver them. More on social features here: https://www.reddit.com/r/heroesofthestorm/comments/8bzsup/ama_with_heroes_developers_april_13_2018/dxb9duv/
We originally reduced division requirements for Team construction as a means of increasing Team League participation, which had a small effect on the population of the mode. I don’t believe reintroducing harsher requirements will alleviate the issue, but I’ll take it back to the team to discuss at more length. In the interim if there are other suggestions for Team League I’d like to hear them.
- Do you have plans about social features like a Clan System, a Party Finder or similar?
Lots of plans. As you’ll likely hear a lot today, it’s a matter of priorities and how we spend our resources. The plan is to start with an updated party finder which significantly improves that system. The major changes would be to decouple the party finder from chat channels, which would significantly expand the pool of available players, and allow players to look for others based on the game mode and role they want to play.
We feel that’s the important first step. From there, we can build upon that with clans, which would be a great addition to the game as well.
But, again, you can see the list of things we’re working on currently and we feel those items are higher priority than the social features right now. So, while we’d love to get to them, and they’re coming, they’re further out.
- What about adding descriptor tags like Mobile, Sustain damage, burst damage, global, healer, ect?
As part of the role update we will be introducing a tag system similar to what you're suggesting. This definitely helps provide more context and information beyond what a higher-level role category can provide.
- What are your long term plans for support viability?
I would say that we want every healer to be viable in at least some situations as a solo healer. Some healers will of course be better at preventing sustained damage, burst damage, or even keeping themselves safe. But overall, yes, ideally there isn’t a healer that you feel like you could never pick without another full-on healer.
- You’ve said that Hanamura is coming back soon with big changes. Are there changes coming for maps like Haunted Mines or Blackheart’s Bay?
Map Update: The map team is currently working on revising a number of our existing maps. The big focuses right now include large updates to Hanamura and Gardens of Terror. We also have some smaller updates planned for a few maps but these are more akin to balance changes. Braxis Holdout, as an example, we’re planning some small changes to address how snowbally that map can feel. We would love to make larger changes to Haunted Mines & Blackheart’s Bay, but those are both further off ☹
- Can you speak to any opinions about the current state of Ana?
The rework was intending to help address the idea that she couldn’t solo heal because she had no easy way to heal herself back up. I know internally we played with a wide range of values on this, and even had a playtest or two where she could outduel everyone in the game because her self-healing was way too high. Ana’s overall win rate is a little on the low side so we have room to bring her up a little bit. I’m hoping we can accomplish what you’re looking for via tuning though, or potentially some slight mechanic tweaks. I will say we love Ana as a healer because of how much skill there goes into doing her healing well.
- Thoughts on Medivh's current state?
Overall we consider the rework a success. It is interesting to see how much more popular he is now (which is a great thing!) but it does mean that players who feel he is frustrating to play against will see him more often now. With our rework, we did attempt to reduce some of the frustrating aspects of his kit: There is more downtime between using Portals & Force of Will, which ideally gives the enemy team more options for counterplay. This is an area we could easily explore and try to do more with.
- Thoughts on heroes the community considers frustrating to play against?
Chromie and Genji are both being looked at internally. Genji has some small stuff that will be coming out really soon (similar to Tracer). Chromie is something we’re evaluating a lot more internally. We’ve done a number of changes such as reducing the range of her Q & W abilities so she would have to be within enemy vision range (you could see it coming and try to dodge), showing the splat for her W ability, and so on. Honestly, it feels a little bit better internally but we might need to do more to really succeed here. Chromie is further off as a result. I don’t think we view Diablo or Garrosh as that frustrating to play against. The recent change to Garrosh where his Q no longer pulls enemies in has helped a lot. In general, every character is going to have some level of bullshit they can bring, but it’s important that you feel like you have counterplay options and ways to outplay your opponent.
- Can we expect a specialist balance pass similar to what you did with the warriors anytime soon?
Most of them actually tend to have higher win rates, so in terms of that, they're not currently struggling. As part of the pass we're working on to reduce frustrating moments with various heroes, there are a number of Specialists who are on that list. Usually its surrounding when they're split pushing and it feels like the enemy team has to constantly respond to them.
- Would you be willing to put together a roadmap similar to what the Destiny team has put together to solve similar community related issues and concerns?
So we definitely want to communicate with you more about what we’re currently working on, and what our view of upcoming priorities are. The blog post and this AMA are some steps in that direction. We're also discussing lots of ideas about how best to communicate with the community going forward.
That being said, here’s some info for you in terms of timing/order of stuff:
- New toxicity report validation system: We should be turning this on within the next couple weeks. This will increase the amount of reports we action by a significant amount. We’re going to put out a dedicated post on this soon.
- Matchmaking improvements: Since the majority of these are on the server side, we’re going to be rolling these out independently of our normal client patches. Some will come quicker than others. For example, we’re going to try and get the hardening of rules to favor match quality more over queue time in the near term.
- Loss Forgiveness / Improved Leaver Punishments: This is deep into implementation and is currently lined up to come out in a patch in the next couple months.
- Ranked improvements: Several of these improvements make the most sense to release with a new season roll. So, we’re trying to get some of these changes into the next season roll (~June). 3rd Ban is high on the list.
- Performance Based Matchmaking: We’re wrapping up the improvements we discussed in the blog post, and are planning to re-introduce this system also with the next season roll. We may initially introduce this as a purely informational system, and then turn on the actual point adjustments after the community has had time to provide feedback. If we do that, it technically doesn’t need to align exactly with a season roll.
- Ranked: Do you plan on changing ranked seasons, so they have the same durations/times as HGC seasons: Two six-month seasons per year? Do you plan on adding a team system, for named teams (in team leagues for example)?
- Regarding season length: We've discussed season length quite a bit on the team. We've actually considered everything from once a month to one season for the whole year. We think the current system is a balance between those, but I would say that we're still open to discussion on this topic.
- Regarding named teams: We actually had this in Heroes originally!
 Team league worked this way when it was initially launched. The issue was that problems we currently face with Team League (which I think one of the other guys answered a question on that already), are actually worse with the fixed named teams. The tl;dr is that it's tough for players to align schedules with their teammates. As a result, we saw even less participation in that system than we see with the current system.
- Summons:
We’re not currently looking at summons as a whole but will likely try and tackle this on a case by case basis. An ability that is really high impact (and frustrating) at lower levels of play but not useful for higher tiers of play is a bit of a red flag for us. That being said, this is kind of the nature of a summoned unit. In some ways, you can think of a summoned unit like an ability with a huge variance in its damage potential, and this largely depends on how the enemy team plays against it. Obviously higher level players can downplay the value of these summoned units much more easily.
- Raynor rework?
Raynor is getting a rework: It’s not the next set of reworks that will be coming out, but we are actively working on it. Overall we wanted to keep the feel and playstyle of Raynor, keeping him simple to play, but we’ve made a number of changes to modernize his base abilities and introduce a lot of new talents. The team is super excited to share more with you guys on this soon (next few months).
- Thoughts on Sgt Hammer?
Hammer: I think we added some fun new stuff to Hammer’s kit, allowing her to access some of her key defining talents earlier on. We haven’t seen a ton of Hammer play, but she’s also a character that probably shouldn’t be core to the meta. We don’t currently have more plans for Hammer but we will likely revisit her again in the future.
- Are you considering allowing to downvote / remove maps, so I can only play the maps I like?
This is something that is brought up occasionally within the team. At the moment our focus is squarely on improving the matchmaking process, and while some sort of map preference system could be interesting it’s not on our radar in the near-term. We believe there are gameplay improvements that can be made for some of our existing battlegrounds in order to make them more enjoyable to play for the majority of our players, so our strategy is to prioritize making all the battlegrounds we have more fun to play rather than giving players more means to opt out of playing them. We have several battleground updates in the works already; Matt discusses them in more detail here
- Are there any plans to improve the bot AI? Currently the bots cannot help win a game if somebody disconnects.
We’ve actually made great strides with our AI tools over the past year. While an AI player will never be able to fully replicate a player there are improvements we have planned to help shore up some of our weaker AIs and ensure that they perform their roles more effectively in the match experience. This is an ongoing process that happens in parallel to other development.
- Thoughts on Chromie having too much burst damage?
This is something that we've talked about a lot internally. The short answer is that the vast majority of Chromie's damage is very unreliable. As such, to keep her win rate balanced, her damage needs to be extremely high when she does land her abilities. This in turn can make it feel frustrating or unfair to play against: "I dodged 90% of what she dished out at me, but the one time she hit me I explode, is that reasonable?" (Very nice, non salty player talking here.)
- Will you consider a surrender option, especially when team mates are AFK or inactive for a long time?
We don’t plan to include a surrender option to the match experience, and instead are focusing on improving our actioning for AFK or inactive players. We feel that the addition of the surrender option will ultimately yield a net negative social experience for players in the match. It serves as a point of contention during the match where the team cooperation can grind to a halt in the face of deciding whether or not to quit the game.
We do however have plans to include a Loss Forgiveness feature for players who are affected by leavers during a ranked match. I intend to write a blog post sometime in the next two weeks to describe this system in more detail.
- Updates on showing MMR in game?
Yep, its coming. Over the years, there’s been a lot of back and forth about whether showing MMR is more valuable than harmful. In the end, we’ve come to feel it’ll be more helpful.
To go a bit more in-depth, it’s good to understand the purpose of both MMR and rank. You touch on this, but I want to use this post to also make sure it’s clear to everyone reading this.
Matchmaking Rating (MMR) is purely a measurement. It’s an indication of your skill as a player which is then used by the matchmaker to put together games. The more accurate MMR, the better the matches that can be made. Gaining, or losing, skill is a slow process that happens over the course of many games, so MMR does not change quickly. It purposefully doesn’t react strongly to short streaks of wins or losses as they’re not necessarily an indication of a skill change so much as a streak of good, or bad, luck. This is where issues come up with showing MMR. Once it’s visible, it draws a lot of attention and gets treated like a reward system. Players tend to focus on figuring out how to make it go up by any means necessary, looking for ways to push it higher as quickly as possible. That’s not a healthy outlook and leads to frustration when it doesn’t happen.
On the other side of the coin, Rank is a reward system. It represents how well you’ve done over the course of a season. If you go on a win streak, your rank spikes accordingly.
The issue we run into is that Rank is currently the only visible indication of skill in the game, so it can’t diverge too much from MMR. When players are put together in a match with players of disparate rank, the assumption is that they are also disparate skill, and that feels bad.
This had led to the need for Rank and MMR to stay relatively close, which is why Personal Rank Adjustment exists. That blunts the ability for rank to be a true reward system. It also means that matchmaking must consider rank, in addition to MMR, when putting together teams. Visible MMR would allow us to get rid of both of those, which we feel is more beneficial than the downsides that are likely to come about with visible MMR.
We’re working out the details currently, but the plan right now is to add it as part of the updates we’re doing for Performance-based Matchmaking, where we’re already working on new ways to provide you with more—and more clear—types of information on how you performed in a game. Joe went into more detail on that here
- Are we getting the ability to watch games of our friends live at some point?
The team is very interested in delivering this functionality to players, but the technology that drives our replay system and it’s peer to peer nature does not make this simple to deliver. Compared to other features we are discussing this is pretty low on the priority list.
- What do you think about the option to show the number of players in the QM queue with a chosen role (assassins, warriors, support, specialists)? There could be a bonus reward for those who choose the least popular role.
Yeah, so we've already got a bonus for playing "Auto-Select" when the matchmaker is getting into trouble (a new hero launches and everyone wants to play that hero). We're looking into potentially expanding that to offer more rewards for under represented roles in QM (e.g. Support).
Showing the percentage of players who have selected particular roles is interesting. I'll take that back to the team and see what they think. Thanks!
- Rain of Destruction:
We're talking about this internally. We do want to do something with this heroic to make it feel less RNG and more competitive with Horrify.
- Zarya:
She's in an okay spot right now. The balance team might make some small tuning changes but no immediately plans.
- Artanis:
Similar to Zarya, no current plans.
- MULE:
We're leaning closer to removing this from the game as it feels really bad when the enemy team resets your progress. We think this mechanic works decent on Abathur since he really needs those safe forward positions, so decent chance that something like this remains on Abathur.
- What about the option to swap heroes in the draft if both players agree?
Lots of validation has to be put in place before swaps could go to the main ranked modes. It's just a non-trivial amount of development time that we need to put towards higher priority issues right now, which is why we're looking at alternatives, like trying out FCFS, that could be done quickly.
|
United States12224 Posts
On April 13 2018 11:53 korona wrote:Here we go again... I still remember Blizzard's overly aggressive MMR decay implementation for SC2 Heart of the Swarm that practically destabilized the ladder. For SC2 this destabilization took several seasons as at the starting point MMRs could still most of the time considered as skill indicators. It took a long time for Blizzard to agree that the MMR decay system was hurting SC2, but eventually they decided to remove it & reset all MMRs. For Heroes of the Storm I feel the MMRs too often don't match skill/game knowledge & this is not just a recent development. All the way up-to master league the match quality has been quite low & depended on "ally lottery". Even at the supposed top 5% of entire MMR spectrum you will often encounter players who seemingly have big gaps regarding very basics - whether it is basic tactics, positioning in team fights, drafting & understanding roles, understanding map specific drafting & tactics, understanding how to play from behind and so on. I am really afraid that if a MMR decay system is implemented on top of the current "mess" the match quality might even further decrease over time. (Currently there is a small scale MMR decay system already in place in Heroes - the so called soft reset system between the seasons. The current system feels mostly fine & not too punishing).
The "ally lottery" is more of a moba staple than a matchmaking failure I believe. The 10 people that you get in each game are all going to have different specializations, different expectations, different familiarizations, and different outlooks. In Dota this is already chaotic and unpredictable, but in HotS it's amplified because you have to own a hero before you can pick it. If your hero pool is small (in the "owned" sense as well as the "comfort" sense) and your limited heroes aren't a good fit for your lineup, you're hindering your team. Across 10 players, that's a lot of chaos.
The main problem with decay in SC2 in my opinion was that it compounded. If you decayed some amount, then played, then lapsed again and decayed some more, you experienced a decay greater than the normal max. It was actually worse to not play for a month, play a game, not play for another month than it was to not play for 2 months straight. A lot of players suggested a decay floor, and I think that would have gone a long way to preserving the integrity of the matchmaking in HotS. Statistically, the decay did work and produced the desired results, but the compounding effect made skill tracking worthless for serial decayers. In HotS, if they have a decay floor and they only apply decay when the confidence of a player's skill rating is high, it will probably be fine.
One thing I'm a little wary of is the performance-based skill tracking. It's a very touchy, very risky thing. SC2 didn't have this because it would have influenced the way people played the game. Overwatch has it, and once it was publicly acknowledged, it started affecting player behavior even though players didn't know what specifically was being tracked. The idea of "playing for stats" rather than helping your team taints the integrity of the performance indicators and ultimately matchmaking.
Josh Menke, the original designer of the SC2 matchmaking system who now works at 343 Studios on Halo, gave a presentation at GDC this year about recent developments in TrueSkill. TrueSkill 2 leverages stat tracking as a post-result verification rather than a direct contributor, and its real impact shines in identifying smurfs and properly ranking new players. For example, if you have a new player with default MMR who plays his first game and goes 20-2 in K/D, then you add as a footnote that this player has potential to be very good, but it needs to be verified against stronger competition. It doesn't impact the player's MMR directly, it provides context which can be beneficial in finding better matches faster. Overall, TrueSkill 2 (and TrueSkill Through Time) converge on a player's skill much more quickly than classic TrueSkill. You just find any measurable metric and remember it. For RTS games, he recommended tracking the resource/income delta over time. For moba games, resource delta applies there too. For HotS in particular, I suppose that would be the XP gap between the teams. What was the XP gap at the 5 minute mark? 10 minute? 15 minute? Was it steady, meaning the game was likely a stomp, or did it fluctuate, tighten, or even flip, meaning it might have been a closer match? For player-specific measurements, you could track XP over time based on that hero's norms for the skill level (which is I'm sure what they're doing, along with things like hero/siege damage), but you don't apply that as a direct modifier to rating. You just track it as a reference point. Announcing the existence of performance-based skill tracking will still inevitably result in "playing for stats", so that's a box that can't be unopened, but you just do your best to not give it a direct impact.
|
Most of the stuff discussed here by the League of Legends devs applies to HOTS:
- Why do some games of League feel imbalanced?
When you encounter imbalanced games of League, we generally chalk it down to three major culprits: snowballing, matchmaking uncertainty, and/or perception.
In terms of design, we want League to be a game where getting a small advantage and running with it is one path to victory. Any small successes that are snowballed can determine early leads, and then sometimes the result of the game. Early Pantheon double kills can feel unstoppable without a miracle teamfight, but there’s always itemization and strategic calls that can turn games around from the brink of defeat.
Within the actual systems, most teams have an expected win rate of 50 +/-1%. This means from all the data we have, we think we’ve made a fair match. But we can’t rely purely on the raw data. What if players are playing positions or champions they aren’t familiar with? What if there’s a large discrepancy in MMRs in a normal premade, or a player hasn’t played enough games to get an accurate rating (looking at you, smurfs)? On top of this, there could be personal factors that matchmaking can’t see—lack of sleep, one too many Graggy Ices, that sorta thing.
Depending on how you’re doing, it could just be your perception. Being wrecked is much more visceral than the joy of dominating your opponent, and the really bad moments can unfortunately be more memorable than the great ones.
And sometimes you’ll just have a bad game.
While we know our matchmaking systems are never going to be perfect, and sometimes you (or your opponent) will end up with a 20 minute win, we try everything we can to prevent imbalance before the game has even started, and consider it a huge priority to ensuring League remains competitive.
— Riot Gortok, Designer, Get in Game Team
- Why do I get worse people in my games when I’m on a hot streak and trying to climb?
This is pretty much an urban myth. Mostly.
There’s nothing in the MMR system that forces you to have lower-skill teammates or disproportionately higher-skill opponents. We expect every game you play to have a 50 +/-1% chance of your team destroying the Nexus.
As your MMR goes up, you’ll stop being the “standout” player and your teammates’ skills will be higher. This can also happen to anyone else in the game, which can give the impression that you have “worse” teammates, when really they might just be newcomers to your skill bracket.
Even if it all goes wrong and you feel like your teammates have been letting you down, you can make use of systems like demotion protection and promo helper to prevent you from being punished for a few “unlucky” games. We think this is the right tradeoff compared to having a volatile rank—being demoted feels rough, and reaching a tier should be a decent indication that you can play at that level.
— Riot Gortok, Designer, Get in Game Team
- Why do I get matched with unranked players if I’m Bronze/Silver/Gold?
When players first start playing ranked, they have to go through a period of placement games where we try to figure out how good they are. We have to place them against players we already have a lot of information about to get the most accurate placement. Since most players are Silver and below, we tend to start looking in Silver and move them up or down based on their performance.
— Riot Gortok, Designer, Get in Game
- Why can’t I see my MMR?
We talked about this when we first released the Leagues system (holy crap, five years ago now). tl;dr: Showing MMR has a lot of downsides in a team-based game like League.
On the plus side, MMR is a more accurate summation of where you are in relation to other players across the entire server, and showing it can be more reassuring that the match you’re in is fair when scouting your opponents out before a game.
Using MMR as the sole mark of achievement in League punishes half of the playerbase as their MMR will decline over the course of the season, which sucks because most of them are gradually getting better at the game—but so is everyone else around them.
Ranked tiers also provide contextual progression and status. Knowing you’re “a Gold player” as opposed to “a 1650 MMR player” or “120,353 on the server” gives you clearer targets to work towards. Moving from 1595 to 1600 MMR is probably not that compelling, but promoting from Silver I to Gold V should give you the knowledge that you’re truly improving.
The Leagues system also gives you a bit of protection from losing a bunch of games in a row and having your MMR plummet as a result. Using demotion protection and promo helper, you can get a few extra lives in rare cases of not getting your preferred position for a few games or just being in a slump. Ranked anxiety is real, and we know there can be a lot of pressure, so having meaningful progression that feels good should hopefully break down a few of those barriers.
— Riot Gortok, Designer, Get in Game
One thing that LoL does that HOTS could consider is queueing with role preference. Where you pick two roles and match maker tries to put you in your preferred role and then your back up choice, and only have to fill other roles if the queue times are really long because of a shortage of a particular role. HOTS would need to redefine and fix their roles before that would be an option.
So some of the "ally lotter" is because of:
- One or more players trying an off-role
- One or more players trying an unfamiliar hero
- An otherwise fair match in which one lane has an unfair matchup
- Games can snowball out of control with early game heroes, or one team has exceptional wave clear and the other team doesn't have enough for example
- Sometimes good players just have bad games. Even pro series can see the exact same teams stomp their opponent in one game, and lose to the same opponent in the next.
- A player may be distracted, intoxicated or have some other outside the game factor causing them to play worse than expected.
- Communication and player attitude. A good shot caller and all your team is on voice chat can raise the level of the whole team, a toxic team mate can make their team play worse.
|
On April 19 2018 05:04 Excalibur_Z wrote:Show nested quote +On April 13 2018 11:53 korona wrote:Here we go again... I still remember Blizzard's overly aggressive MMR decay implementation for SC2 Heart of the Swarm that practically destabilized the ladder. For SC2 this destabilization took several seasons as at the starting point MMRs could still most of the time considered as skill indicators. It took a long time for Blizzard to agree that the MMR decay system was hurting SC2, but eventually they decided to remove it & reset all MMRs. For Heroes of the Storm I feel the MMRs too often don't match skill/game knowledge & this is not just a recent development. All the way up-to master league the match quality has been quite low & depended on "ally lottery". Even at the supposed top 5% of entire MMR spectrum you will often encounter players who seemingly have big gaps regarding very basics - whether it is basic tactics, positioning in team fights, drafting & understanding roles, understanding map specific drafting & tactics, understanding how to play from behind and so on. I am really afraid that if a MMR decay system is implemented on top of the current "mess" the match quality might even further decrease over time. (Currently there is a small scale MMR decay system already in place in Heroes - the so called soft reset system between the seasons. The current system feels mostly fine & not too punishing). The "ally lottery" is more of a moba staple than a matchmaking failure I believe. The 10 people that you get in each game are all going to have different specializations, different expectations, different familiarizations, and different outlooks. In Dota this is already chaotic and unpredictable, but in HotS it's amplified because you have to own a hero before you can pick it. If your hero pool is small (in the "owned" sense as well as the "comfort" sense) and your limited heroes aren't a good fit for your lineup, you're hindering your team. Across 10 players, that's a lot of chaos. The main problem with decay in SC2 in my opinion was that it compounded. If you decayed some amount, then played, then lapsed again and decayed some more, you experienced a decay greater than the normal max. It was actually worse to not play for a month, play a game, not play for another month than it was to not play for 2 months straight. A lot of players suggested a decay floor, and I think that would have gone a long way to preserving the integrity of the matchmaking in HotS. Statistically, the decay did work and produced the desired results, but the compounding effect made skill tracking worthless for serial decayers. In HotS, if they have a decay floor and they only apply decay when the confidence of a player's skill rating is high, it will probably be fine.
It is true that there may be feasible ways to implement a decay system especially if decay floors are used. This announcement was frightening because it seemed sudden & direct response to the community outcry on Reddit. It addressed most of things that the community was "demanding" (decay was one such "demand"). It almost felt like the Heroes team was giving in to the demands. Before this the Heroes lead designer Travis McGeathy had considered that there was no need for decay feature as he thought the "rust" for returning players would last only handful of matches & there could be other ways to handle the problem.
It is likely that the Heroes team includes some senior personnel from the old SC2 team. During the SC2 decay era it seemed that the team did not even try to adjust the decay implementation when it's pitfalls became apparent. Earlier experiences are often big factors when new features are under design. My personal fear is that this is more a "throw in" feature for Heroes & it might be easier for them just to follow the "old route". Time will tell. The Heroes team remained silent about the decay implementation in the Q&A that followed the announcement.
On April 19 2018 05:04 Excalibur_Z wrote:
One thing I'm a little wary of is the performance-based skill tracking. It's a very touchy, very risky thing. SC2 didn't have this because it would have influenced the way people played the game. Overwatch has it, and once it was publicly acknowledged, it started affecting player behavior even though players didn't know what specifically was being tracked. The idea of "playing for stats" rather than helping your team taints the integrity of the performance indicators and ultimately matchmaking.
The performance-based skill tracking is indeed frightening. Based on just the few days the first iteration of the system was used, it seemed to have clear design level problems - it allowed gaming the system, it was heavily depended on playing style & skill choices, it was enabled also for characters that had just had big changes (this was accidental thought) & it had generally too big impact + other things. It has been 4 months since the first iteration of the system was disabled. There has been no communication what kind of changes the community can expect regarding it, except that there will supposedly be more transparency.
Blizzard seems to be determined to try again with the PBM. In this case the community can just wait and see & hope for the best. But if too many features that potentially have a big impact in a long run are implemented simultaneously, it may lead to a bigger 'mess'.
On April 19 2018 05:04 Excalibur_Z wrote:
Josh Menke, the original designer of the SC2 matchmaking system who now works at 343 Studios on Halo, gave a presentation at GDC this year about recent developments in TrueSkill. TrueSkill 2 leverages stat tracking as a post-result verification rather than a direct contributor, and its real impact shines in identifying smurfs and properly ranking new players. For example, if you have a new player with default MMR who plays his first game and goes 20-2 in K/D, then you add as a footnote that this player has potential to be very good, but it needs to be verified against stronger competition. It doesn't impact the player's MMR directly, it provides context which can be beneficial in finding better matches faster. Overall, TrueSkill 2 (and TrueSkill Through Time) converge on a player's skill much more quickly than classic TrueSkill. You just find any measurable metric and remember it. For RTS games, he recommended tracking the resource/income delta over time. For moba games, resource delta applies there too. For HotS in particular, I suppose that would be the XP gap between the teams. What was the XP gap at the 5 minute mark? 10 minute? 15 minute? Was it steady, meaning the game was likely a stomp, or did it fluctuate, tighten, or even flip, meaning it might have been a closer match? For player-specific measurements, you could track XP over time based on that hero's norms for the skill level (which is I'm sure what they're doing, along with things like hero/siege damage), but you don't apply that as a direct modifier to rating. You just track it as a reference point. Announcing the existence of performance-based skill tracking will still inevitably result in "playing for stats", so that's a box that can't be unopened, but you just do your best to not give it a direct impact. Interesting. I checked Josh's 2016 GDC presentation when it was published on Youtube Dec 2017. Hopefully the 2018 presentation will be available from public sources faster than the 2016 presentation was made available. The slides for 2018 presentation are already available thought: https://www.gdcvault.com/play/1025425/Significantly-Improving-Your-Skill-System
|
EPT
