Matchmaking and the Perceived Problems of Competitive Play, A proposal

Hello everyone, WyomingMyst here. In this discussion, I will be identifying the current "perceived problems" of Competitive Play, the underlying factor that creates these problems, along with the one solution I wish to propose that, in theory, will give players the means to no longer deal with these perceived problems. First, let’s begin with identifying each of the perceived problems of Competitive Play. Feel free to click the title to expand on each perceived problem and learn what it is, what has been done about it, and why it is still seen as a problem.

Leavers/Teammates Disconnecting

The first problem is when players leave a competitive match. While many times when a player leaves it may be due to rage-quitting that match. However, it is far more complicated than that. Players are also marked for leaving a match for reasons including, being pulled away from the game due to personal issues, disconnections from the game server or the internet entirely, computer crashes, and power outages. This means some of these problems are due to causes that the player had no immediate control over. However regardless of the reason, when a player leaves they are still penalized.

This is, unfortunately, the core of the leaver problem. The penalty for leaving a competitive match must start small with only a 10-minute suspension and a match loss with a fixed 50 skill rating deducted from the player. The reason it starts small is that the penalty must account for those who have technical issues such as disconnections, game crashes, and power outages and give them a fair chance to repair the technical issue before they play Competitive again. In fact, a high percentage of our community has encountered at least one technical issue during their time of playing Overwatch.

The system does quickly escalate in penalties with consecutive leaves. In fact, it takes as little as six consecutive leaves to accumulate to a season ban. When a player is season-banned, they lose all end-of-season rewards including bonus competitive points. Furthermore, players who get three season bans will be banned from Competitive Play indefinitely. Season bans and permanent bans from the Competitive Mode of Overwatch cannot be appealed. The system, however, does work to lower suspension time if the player completes a majority of their matches and shows consistent reliability as they participate in the season.

It is because of this precarious balance of penalizing leavers but being fair to give honest players a chance to redeem themselves, that makes the Competitive Leaver Penalty System as balanced as it can reasonably be. Unfortunately, it still allows leavers to be a perceived problem in Competitive Play.


Lack of Team Communication/No Voice Comms

I have seen many posts on these forums that address the lack of team communication, including a lack of using voice chat in Competitive matches. While there is no denying that choosing to not participate in team communication will leave both that player and the team at a possible disadvantage, I do believe it is unreasonable and unfair to all parties involved to force players to participate in team communication. Players may not choose or be unable to participate in team communication for reasons including, not wishing to be intimidated or insulted by other players including for reasons such as discrimination, they do not own a microphone (which is a very common cause on console platforms), or they have technical malfunctions that prevent their voice communication features from working such as playing on a network that deliberately blocks the necessary network ports that allows Overwatch Voice Comms to work properly.

As such players will not be penalized for not participating in team communication. In fact, reports for Abusive Chat and Spam are not accounted for if a player is not participating in team communication when they are reported. The Overwatch team does encourage team communication and that is why they created the “Join Team Chat” button on the Hero Selection Screen for all compatible game modes in January 2017. However, there is no effective way to enforce it on all players and it would be wrong to do so. Making lack of team communication another perceived problem in Competitive Play.


Hero Selection Conflicts/One-Tricking

Hero selection conflicts is a growing perceived problem in Overwatch. Overwatch is a unique team-based shooter where players are allowed to change their heroes in the middle of a match anytime while in the spawn room. This is a fun mechanic that allows teams to be able to counter-play other teams and create strategies to play the best they can. Overwatch also has the elements of a MOBA, where it features a growing pool of diverse characters and each and every one of them are fun to play. However, the nature of each character makes it very easy for a player to become accustomed to a single role in a team composition, allowing them to play their best possible game while only focused on that role. There is nothing wrong with this and in fact, professional teams primarily consist of players that perform their best in their trained role of DPS, Tank, or Support.

Unfortunately, Competitive Play does not effectively allow for players of a specific role to be evenly paired with other players who can effectively fill other roles. Despite some changes have been made to the matchmaker in the last few seasons, it cannot guarantee players of certain roles will be in every team, ever match. This, unfortunately, requires players to possibly play roles they are not comfortable with. Also, with the hero-changing mechanic of Overwatch, team compositions cannot realistically be dictated to specific formations such as 2 DPS, 2 Tanks, and 2 Healers. Therefore, proposed solutions such as a “role queue” would be ineffective solving this or any other perceived problem of Competitive Play.

This is also why there is a perceived problem of players who are considered “one-trick ponies” of a specific hero. It is reasonable to assume a player playing a certain hero only will be able to maximize the performance of that hero and can be effective in many situations during the course of a match. However, many heroes are often easily countered when the enemy team changes to a hero that can completely shut down such a player. So, while it is not a violation of Code-of-Conduct to specifically play a single hero, doing so risks putting the entire team in that match at a possible disadvantage.


Toxicity/Violations of Code-of-Conduct

In any given online video game, toxicity and players who violate that game’s Code-of-Conduct is always one of the biggest challenges for the success and health of that game. Players often become tilted when a match does not play to their expectations (whether they win or lose). Any of the aforementioned problems can attribute to this, and there are other various factors as well. However overall, it is a lack of control that players can feel, that can create toxic behavior.

Over the last two years, the development team for Overwatch has given players the tools and means to report players who are violating Code-of-Conduct. These tools, for the most part, is used effectively and helps Blizzard find players who violate the Code-of-Conduct. However, it is important to note, Blizzard keeps the protocol of how to identify such players secret. This ensures the reporting system can’t be gamed or abused by anyone. This often gives the feel that reports are not effective at times because the reasons a player may be reporting a person is not a violation of the Code-of-Conduct. As such this leaves players frustrated to dealing with what they perceive to be a playing who are toxic. Furthermore, there are no avoid player features in Overwatch anymore as it can unfairly prevent honest players from finding fair matches. In any case, players being unable to pair up with other players that meet their standards is a perceived problem in Competitive Play.

The Unseen Problem: Lack of Teamwork

As you can see these four major perceived problems contribute to one single unseen problem in Competitive Play. This unseen problem is a lack of teamwork. Overwatch is a very team-oriented game, and in most cases, a team that has a better sense of teamwork will have the stronger chance to win the match. It is my opinion that the focus of Teamwork should be the number one factor in being able to gain skill rating and reach the higher levels of the Competitive Play rankings. The best way to achieve good teamwork is by grouping up with other reliable players that share the same agenda. However, right now the matchmaking does not reward this factor.


Matchmaking currently works to pair groups of players into teams that face teams with similar group counts. It does not do this precisely every time, but it is a factor when pairing players. And if it can’t perfectly pair groups together, a group may face against a team of solo-queue or smaller grouped players who have an average SR that exceeds the grouped team. In fact, players at the very ranks find they are unable to even be queued for a match if they group up into a six-stack. On paper, this seems like a fair way to go about it, and for a single match it could be considered fair, but when working to develop a competitive experience that rewards teamwork, grouping up often results in allowing the four perceived problems to appear at all levels of Competitive play. In the following post on the older forums, Principal Director Scott Mercer does explain why the matchmaker currently does use a group priority matchmaking:

For matchmaking groups, there's actually two separate issues that we try to solve. The first issue is "How do we handle groups formed of players with different MMRs?". With season 2 we prevented players of REALLY disparate Skill Rating from grouping, but there's still some variance we need to handle. Over time we've tested different models to try and see what's best and are now using what tested most accurately. (Hint: it's not simply averaging the MMRs)

The other issue is how do we model the synergistic effects of players being together in a group. As you noted, they have access to voice chat. Now here's where things get interesting. This "massive" advantage actually differs based upon the skill rating of the group members. Based upon the data we've seen groups of low to mid SR players don't see that much improvement to their win %. Higher SR players do see more notable improvements, but it's not as huge as you might think. Still, we do take this into account when we predict the win% for each team. Regardless of how the data looks, we do know there's a perception of a large advantage for groups. That's one of the reasons why we explicitly try to match similar sized groups together.

The Proposed Solution: Remove Group Pairing Protocols in Matchmaking

The best possible solution to removing the four perceived problems of Competitive Play and make teamwork the first rewarding factor is to remove group pairing restrictions. This solution on the surface may seem radical, but if it was implemented, would allow any player the means to completely avoid serial leavers and toxic players, have reliable friends who communicate in voice comms, and they can pre-plan team compositions before ever having to start a Competitive Battle. The disadvantage is that there would be matches were 6 solo queue players would possibly face against a grouped 6-stack. However, that would be intended because 6 solo queue players that can beat a 6-stack would prove to have more teamwork than the 6-stack and would gain rank appropriately. Another disadvantage is that it could also be considered more incentive to have players carry lower skilled players into higher ranks, however, I believe this would be made irrelevant in time if the competitive ladder is reorganized where players with good teamwork skills can level up. In theory, this would make grouping up a tactical advantage at all skill tiers, but I think that is not a problem and should be encouraged. See the visual example below:

It would not completely remove the four perceived problems, but this solution will give every honest player the option to completely avoid those problems and be able to enjoy high-quality competitive matches in Overwatch. Yes, players can group up now, but since they are paired with other large groups of players, there is often match imbalance between the two skill levels of each team and it is more difficult to enjoy consistant matches in Overwatch. Players who can’t contribute to teamwork-oriented experience would eventually drop in rank. The Looking-for-Group forums would be effectively used, and it would open the door to bring in features such as a guild/clan system.

If such as system was applied, there would be a bit of a rough transition period. Which might be another disadvantage to this idea. Anyone who does not immediately start to group up would probably see the greatest effect especially if they start falling in rank realizing they can't perform if they are not good in teamwork oriented play. However, in time the overall goal would allow anyone at any rank the ability to group up into an organized team and find a fair match. This is especially important at the highest ranks, where grouping up into a six-stack is near impossible.

Now there may be other reasons why the Overwatch Development Team has designed the matchmaker in both Competitive and casual game modes to favor group priority matchmaking. As I mentioned above, the system does make each individual match feel more balanced. However, if there is any more reasoning to why group priority matchmaking is used, I would love any insight from this. As this is probably the one remaining question that I have about why the game works the way it does.

In conclusion, I would love to see Overwatch apply its matchmaker so that group priority is no longer a factor; which allows everyone the means to completely avoid the perceived problems of matchmaking in all of Overwatch's game modes. Thanks for reading my proposal everyone.

13 Likes

Well thought out and extremely well formatted. I enjoyed the read.

That being said, I take issue with the hidden MMR mechanism. It makes “Skill” rating a misnomer. MMR actually makes SR kind of meaningless aside from a superficial rank identifier. In any given rank - even at the SAME SR VALUE - there is a HUGE variance in actual skill. Given the way matchmaker works, there will certainly be a “most skilled player” on any given team. That player has a higher MMR then all the rest of the teammates. That player’s MMR will be used to balance out team MMRs to “make a fair match”. Personally, I think there is nothing fair about being expected to carry your team. It results in a lot of huge SR gains and losses and kills any consistency for skilled players.

I would very much like to see an end to MMR and let a flat, organic SR be the ONLY rating. It would mean chaos as players started to find their proper SR, but at least SR would have a meaning. Players would be able to see what all their time and hard work really means. Having the true skill value hidden is a huge insult.

It would also be nice to see an actual scoreboard on COMPETITIVE mode. Seriously.

5 Likes

Are you saying that for any given MMR/SR there is likely someone that is the best, as in a mistake by the MMR system, or do you mean there is a large range of MMR/SR and the one that is high by several hundred is expected to carry?

I can kinda agree with the first, but the system doesn’t make decisions on it’s measurement errors, just the measurements themselves as it can’t know of the error.

The second, though, is avoided by the system and only happens when there are low player pools. No one is expected to carry, the average SR is the same on both sides (though probably randomly distributed from there, as long as it’s equal I doubt they look into it further.)

I mean, no one is expected to carry the team. That would be unnecessarily complicated matchmaking. It’s just a bunch of people who the system thinks are approximate equals and sometimes it’s wrong.

Do you mean having SR go up or down based strictly on the average of the opposing team? They’ve said repeatedly that this is the majority of how the MMR works and MMR tracks closely with SR. You climb with a 50% win rate if you consistently beat players that are better than you…and don’t lose against those who are worse.

Well from what I understand, the hidden MMR is like an Elo rating and it can climb or drop to infinite levels (there is no end cap unlike the presented skill ratings). Now a basic ELO formula looks something like this:

E = (O + 400(W - L))/G

E = Elo Rating
O= Sum of Opponents Ratings
W=Total Number of Wins
L=Total Number of Losses
G=Total Number of Games

So a basic presentation of our MMR would present a very confusing number to work around what would be an even more complex formula. This is why skill rating is a very simple to understand the whole number for us to read.

So I am not sure if the display of skill rating versus actual hidden MMR is anything wrong right now. But lets face it… I hate math and rather not be obsessed over it.

1 Like

Very good post, I apreciate that the OP spent a lot of time on it to make your arguments detailed and clear.

I agree with most of what was said in the post however I feel like it gives teamwork a tad too much weight compared to mechanical skill. Right now, of course, it’s the other way around, with mechanical skill most of the time out doing coordination, so increasing teamwork’s value in the MM is needed, but I would be careful so that the importantance of mechnical skill is lost.

Why don’t hardstuck players like yourself worry about bettering your own gameplay instead of trying to change an elo system that correctly assesses your ability to play this game?

1 Like

That is possibly true, however lets take a recent match of the Overwatch League to make a comparison. Last night, Dallas Fuel had a daunting task of playing against the New York Excelsior. (Click to watch.) Now as many of you know, Dallas Fuel has well… performed very poorly and not to their original expectations when the season first started. However, last night was different. Their player Seagull who is a known DPS main took up the off-tank role, and Mickey their off-tank D.Va specialist, started playing the main tank role, and to everyone’s suprised, it worked surprisingly well taking the match to a Map 5. Now NYXL still beat Dallas last night, but Dallas, regarded as the second worst team in the league, held really strong by taking a more teamwork based approach rather than relying on the mechanical skill and best heroes of each player. This may not be the best example, but it demonstrates why teamwork skill is equally important to mechanical skill.

My proposal would benefit what you want, everyone to be able to play this game properly, and it would give the opportunity for everyone to have consistent matches if they are both mechanically skilled and teamwork oriented, whether it be solo queue or six-stack.

1 Like

I have a question.

Regarding the second scenario (3 vs 4 stacks), what is it about the current system that makes you think that the second version isn’t actually the way it works now? I mean, why do you think the system draws the 5 solo q players in a way that there is a 80 SR average difference rather than a 1 SR difference?

I mean…I get the first example. I just don’t understand the second. I’m just not sure your second example is realistic or, if it is, where you get the idea that they would group that particular person in there with better options available.

It may not be perfectly realistic in the drawn-out example, most of my matches where I have observed a 3-stack versus a 4-stack has anywhere from 10 SR difference to as much as 100 SR. My example is assuming a far smaller pool of players versus what may actually be in the queue at any given time.

1 Like

I was just wondering if there was something that I missed in your example or explanation. I generally agree that teamwork is the overall issue people have with Overwatch and that most just don’t realize it. I’ve written a large paper on the subject. I’ll link if you’re interested.

I suspect the actual queues to be far smaller than most people really think, if you account for ping and limit the MMR range to =/- 50. There’s clearly enough people to make a match of 12 in a few minutes, but the games are only 15 minutes long. You need only 96 people in queue to pull that off. Not that I think there’s only 96, but you don’t need that many to see the results we do see.

Please do, since you likely don’t have direct linking abilities, just place the URL in-between two tick marks (`) to show properly.

https://www.reddit.com/r/Overwatch/comments/7y87hy/jeff_and_the_overwatch_team_admit_they_have_a/?sort=new&st=jekniude&sh=465f45db

Yeah, I learned how to get around that restriction pretty quick. Thanks!

2 Likes

I’ve queued as a 6 stack maybe 30 times, 10 of those time we ran up against 6 solo queuers. This happened after being in queue for 7 to 8 minutes. The current group restrictions you speak of, may only be for the first 6 minutes that you are in the queue.

Correct, the system doesn’t always force same stacks together if it can’t find an appropriate pairing, but it tries to very often or if necessary, cancel the queue in extreme cases. When it does pair very unevenly, there is always a significant skill rating difference that favors the team with smaller group counts. Again from, a single-match perspective it could be considered “fair” but in terms of trying to form a team that can play consistently together, it often results in awkward match quality and eventually decentivizes keeping a team together. Especially one that decides to group up after a successful match as solo queues.

1 Like

At first, I am surprised I missed that reddit, but then realized you originally posted when I was on vacation in California… watching OWL.

This is a good read.

Yeah. It got no traction here when I tried to x-post it. Of course, x-posting isn’t allowed here either, I don’t think .

The big problem with the pictures that represented your thinking is that how is 6 solo q’s going to be at the same communication and coordination of a six stack. This system is a start but I don’t think it’s the right one.

No they are not going to be at the same level, and that is the problem we have now with the current system, and why tools like the “Stay as Group” button is not effective. Now in theory with this proposed system, solo queues can eventually climb in ranks if they can think and function like a member of a grouped party. It won’t be perfect, but I believe it is possible.

This may not be the ultimate solution, and realistically even with this system grouping up can only go so far before other factors such as mechanical skill and technological advantages (both of which already exist in our game) need to apply. However it is my hope that if such a matchmaking protocol was implimented it would encourage more grouping up, formation of long lasting teams, and far more high quality matches the higher the ranks go. None of which is found in our current version of the Competitive Mode.

2 Likes

The game does a terrible job of separating the ranks. IF that’s even their goal. I feel like their goal is to mix noobs in with experienced players, for some reason.

Because i keep seeing low levels who don’t even know the game. Why am i on my 400 hour Main hero trying to play my absolute best in unwinnable games with a level 26 attack Torb? Not sure.

But since i’ve started keeping track, any player under level 50 has a record of…
4-24. Yep.

That’s just ONE problem with this entire MM and SR system.
It’s really becoming unplayable at this point.

As long as they refuse to let the matches be truly random, stop trying to control things, and let the players fall where they may… the game will be a toxic cesspool.

1 Like

I love this post, thank you.