Image for post
Image for post

The issue of current front-facing quantitative performance metrics in Blizzard’s Overwatch (and other topical balance issues)

Nicholas Walsh
Jul 8, 2016 · 28 min read

Depriving players of critical information does more harm than good.

From a holistic level, Blizzard’s approach to quantitative metrics in Overwatch exhibits a profound identity crisis that when dissected, seems to be a tale of two conflicting design philosophies. A duality presents itself that struggles between wanting to project significant and detailed in-game player statistics (a necessity in any competitive game) and seemingly promoting a more cooperative environment for players.

Image for post
Image for post

However, in implementation, this becomes fairly counter intuitive when (what I assume Blizzard intended to be) a ‘safe space’ is promoted via reduced transparency of performance metrics, rather than dealing with the core fundamental issue, which is how players actually behave (and as an extension, why they behave that way). In its’ current iteration, I do not envision a world in which both statistical transparency and censorship can coexist while still asserting the game as truly competitive. The very core of competitive games is to pit teams against one another, where a winner is determined and rewarded for playing better than their opponents, not just in mechanical skill, but in strategy and meta-game. Inhibiting access to teammates/opponents real-time performance fundamentally undermines much of the strategy that competitive gaming is predicated on — Overwatch isn’t Team Deathmatch.

Stripped of the ability to analyze the performance of individuals in a match (both internally within a team, and externally against the other team), it becomes nigh on impossible to adapt mid-game for players that aren’t unreasonably cognizant of their own performance relative to those around them. This plagues not only solo/group competitive queue, but even professional teams as well. Not only are vital statistics that are knowingly recorded not being shared, but there are much more meaningful statistics that are currently not recorded — statistics that can much more accurately construe meaning in regards to certain heroes’ performance.

I could go on and name countless scenarios where a simple match-wide scoreboard would actually resolve many currently toxic*, standoffish scenarios that are a result solely of Blizzard refusing to show these metrics in the status quo.

*Footnote: It’s critical to understand that Blizzard will set the standard for what is and what isn’t toxic. For me, asking a player to switch a character if they aren’t performing well is not toxicity. Asking for your team to communicate is not toxicity. Pinpointing mistakes your teammates are making do not constitute toxicity.

There are many things such as threats, racism, and more that don’t need to be mentioned here that explicitly and unarguably fall under toxicity. However, the confusing, gray areas of toxicity are not typically defined by what is being said, but typically in how it is being said. Berating somebody by telling them that they are bad at the game with no evidence, in a blind rage, is fundamentally different than being able to use numbers and physical evidence to make constructive critiques and suggestions, but without realtime statistics, this becomes difficult to achieve.

To name a few scenarios where improved statistics could improve the current game,

  • Which DPS hero isn’t pulling their weight?
Image for post
Image for post
Transparent data allows us to be friends, not enemies.

These are all questions whose answers are imperative to mid-match adaptation, yet devolve into a verbal massacre over comms when no physical evidence can be supplied, causing players to get scapegoated or berated unnecessarily. Needless to say, in both solo/group competitive play, this makes actually improving at the game unnecessarily difficult, not to mention infuriating. I can’t even begin to count the number of times where I’ve lost games and have had nothing to say other than ‘I don’t even know what we did wrong, what do we need to improve?’. Accountability is tantamount to improvement, whether that be as an individual’s decision making within a single round, team composition changes within a match, or a player’s overall hero pool or skill level over the course of their career.

Scoreboard (or lack thereof)

The lack of a scoreboard is a fundamental issue with performance metrics in Overwatch. In its’ stead, we have the medals system. At its’ core, the medal system has good intentions — reward people who are leading their team in a particular statistic by giving them an aesthetic (albeit meaningless) gold, silver, or bronze medal for their corresponding team ranking per stat.

Sadly, in practice, this system results in far more confusion, misappropriation, and general disillusionment than originally intended. People are often seen using the medals as false justification for positive performance. Great examples are how healers will obviously always have the most healing on their team, tanks will always have most objective time, and heroes that are spamming damage from afar/spamming a Reinhardt shield will have the most damage. The most important comparison to be made there is not how those values compared within your team, but to the other team’s players. I’ve seen a similarly disillusioned issue with the Play of the Game/Match: it seems to lull offensive or burst damage hero players into a false sense of security as to their skill level. Bastion, Hanzo (ult), Genji (ult) are traditionally often receiving Play of the Games, as their ults will always get extremely high kill counts. From a balance perspective, Hanzo/Genji are seemingly predicated on relying on their ultimates for big plays, but the issue at hand is that many of these players are completely underperforming and/or being generally useless all round until they get their ultimate, which isn’t how they’re designed. Hell, harassment and poking is a skill within itself that won’t even show up in a performance metric (unless Blizzard broke down damage done to each member of the enemy team), but the point still stands that countless players use PoTG as a false sense of justification for their performance within a game despite them consistently underperforming. The current obfuscation of statistics only further feeds into this disillusionment as to how effective some of these players are actually being.

Image for post
Image for post
My face when I found out Blizzard didn’t put a scoreboard in Overwatch


However, stepping back for a moment, none of these ‘medals’ actually show how the team is doing relative to important, known standards (relative to other players on that same hero at that rank, or more importantly, compared to the other team).

Eliminations: Blizzard took a bit of a contrary approach to the classic ‘Killing Blows’ statistic here, but I don’t know if it is playing out as well as they would’ve liked. Eliminations, to my understanding, are essentially awarded to any player who dealt damage to an enemy that was then killed in the near future.

Image for post
Image for post
I guess this is a bug

Because of the large amount of general healing in this game, one of the most important things I’ve actually been interested in seeing has not just been who is able to participate in kills, but rather who is able to actually complete kills by securing executions. A common theory is that Blizzard took this approach to lessen the emphasis of ‘Kill Stealing’ that happens in other games, where it could potentially promote toxicity amongst players who wait to secure kills simply to pad their own stats. Again, good intentions, but I think Blizzard really put the cart before the horse here and has tried to preempt toxicity unnecessarily, only creating more problems in the process.

Does it really make sense for a team’s Lucio that has <8% of his team’s damage to have an extremely high number of eliminations by just senselessly spamming his blaster that has little to no impact? What legitimacy does that give this statistic? Implement an execution statistic and award it for when he can successfully finish someone off, as that is a much better representation of what he’s actually doing.

Make Killing Blows, Assists, and Damage Done three separate statistics, as they are all unique and tell very different yet vital stories as to how a match progressed.

Damage Done: Extremely important statistic, but the medals simply aren’t enough. Often times people who have silver or gold damage in a match utilize medals as a false justification of good performance to ignore any advice given by a team. If a team is collectively doing badly, being the ‘cream of the crap’ doesn’t necessarily mean that any particular individual is immune from making changes that could help the team overall, but false accolades promote this mentality. A scoreboard needs to be implemented to put actual numbers into context both within a team’s performance, and against the opponents’.

It’s easy to lie without numbers. It’s easier to lie with them.

It’s critical to understand that certain heroes are going to naturally put up more damage than others. A great example of this is how Tracer or Genji may not have a huge amount of damage done, but are extremely potent at assassinating key targets such as Mercy or Bastion, which can almost single-handedly win a fight. By being able to see scoreboards, the community can begin to garner more realistic expectations for how individual heroes perform in regards to one another, and stop this obsession over arbitrary and borderline meaningless medals.

Healing Done: This is a fairly useless medal. I understand that Blizzard doesn’t want to enforce a strict team composition/meta, but in practice, teams will almost always run 1 or 2 healers, and it doesn’t necessarily matter which healer is ‘healing more’, as the only two healing heroes in the game (Mercy and Lucio) have extremely different types of heals that don’t necessarily need to compete with one another. I have a lot of gripes with raw healing done as a performance statistic in general, but I’ll get into that later.

Objective Time: I’m glad this exists, but again, it becomes a bit of a moot point, similar to healing.

Payload? Your main tank should have gold, always. Maybe a Lucio, but this is something that anybody on one’s team can see based on fundamental positioning. This results in Reinhardt getting a medal for having 90% of the game time on the objective, with silver, bronze, and everything short of those being meaningless.

King of the hill? I care about finding the member of my team who isn’t spending time on the objective, which this doesn’t accomplish.

Attack/Defend? Fairly useless statistic in this context, see King of the Hill. Often on Attack/Defend, I can see who isn’t fighting on the site when someone gets to spectate after death, at which point they get told to politely get their ass on the objective, because this game isn’t Deathmatch.

Objective Kills: It’s unclear as to what exactly constitutes an objective kill across each gametype, as it’s never described in game, but kills that involve the shooter being on the site for Attack/Defend or Control Points definitely count. I’m a huge proponent of Blizzard incentivizing fighting on the objective (as many players seem to think this is just team deathmatch and just watch as the enemy team retakes a site), but I’m not quite sure if the way this one is calculated is perfect yet. Shooting enemies that are standing on a payload/site should be part of the calculation if it isn’t already.

Rank System

In general, the system needs to be better explained in a detailed blog post or video by the algorithm devs. How does the underdog/favorite effect the match? Blizzard, my question to you — do teams that vary by 1–2 skill points actually have any sort of meaningful difference in skill level, at least to the point where ELO changes should be altered? I understand that you want to use algorithms to modify ELO changes based on in-match performance, and I love the idea, but, without meaningful performance metrics or even a proven way to evaluate the ones you already utilize (which seem to have fairly low victory causality across the board), it’s doing far more harm than good. To be honest, statistics very rarely explain some of the most pivotal reasons why certain teams either won or lost games, and as such, shouldn’t nearly continue to influence ELO changes to the extent which they currently do.

There are too many commonplace scenarios in which statistics not only don’t tell the full story, but in turn, could tell the opposite story, that using this current system is criminally negligent. The biggest case of late has been in relation to healers, particularly Mercy due to the fact that she is incapable of acquiring a sizable number of eliminations. Clearly, the algorithm needs to be tweaked, and you’ve acknowledged that, but I don’t think it’s a bug so much as it is a fundamental design philosophy. This wasn’t a bug, the system wasn’t behaving in a way it wasn’t supposed to, it simply wasn’t tested or even designed properly. How does this happen for an irreplacable hero that has a near 100% pick rate? Mercy arguably has three aspects to her kit, damage boosting, healing, and resurrections, the first of which doesn’t even have a recorded statistic, which is absurdity. If you want a way to incorporate true performance metrics for Mercy, Blizzard needs to find a way to count this either towards eliminations or damage dealt. Further, in countless situations, healing done is extremely situational. DPS’ers have 6 targets to deal damage to at any given time, bound by their opponents’ positioning, whereas a healer is bound by their own teammates. Healers are uniquely punished for teammates’ inability to position properly in a way that any damage dealing hero is not.

Image for post
Image for post
If Dr. Harold Winston was a Data Scientist, I’d love to hear his thoughts on these matters

To extend this, in any sort of one sided match, healers get shafted regardless of winning or losing in the performance algorithm Blizzard is using now. When on defense, teammates that instantly die due to offensive blitzes have no way to be healed, so even if you try to factor in Amount Healed divided by Damage Taken, the healers will still score extremely low. In games where your team blows the other team out and little damage is taken or people heal themselves with health packs as the primary source of regeneration, healers are also getting shafted.

I appreciate trying to incorporate performance metrics into account, but they’re truly broken and without actually pooling large amounts of data and coupling it with the necessary machine learning to get statistically meaningful models, the algorithm’s factor weightings are nothing more than arbitrary, which are why we are seeing such absurd and unrecognizable gains and losses for ELO. Personally, I’d like to see this feature removed until more data and models can be produced. Save it for Season 2 or 3.

Second, the group matchmaking has some serious avenues for abuse. From what I’ve observed, group rating is taken by calculating the simple mean of the group (Sum of Ranks divided by Number of Players). This wouldn’t be as much of a big deal if skill differences didn’t translate to exponentially different performance rates. Ex: A level 65 isn’t simply “10% better” than a 55, the difference is much greater than that. Alas, A 65 paired with a 45 has the same mean rank as two 55’s, but the level 65 playing a high skillcap offensive hero (Tracer/Genji/Soldier/ex Widow/ex McCree) would simply roll over an entire team, with the lowest level on the team typically filling a role that mitigates their lower level (Lucio as I typically have seen). This isn’t saying that I want every class to be equally as difficult, as presumably ‘harder’ classes have a high-risk high-reward complex, however in instances where the skill differential is so much greater, the capacity for a higher ranked individual to carry a game singlehandedly just becomes too much to overcome at times, especially when there aren’t direct hard counters to certain heroes (Tracer/Reaper come to mind). Again, Blizzard, you have the data for this. I’d love to collaborate to be able to prove or disprove many of my hypotheses, but these are trends that seem to appear constantly across large queue 58–70 ELO games.

The potential for abuse extends further, however, because ELO gains and losses are calculated not based on how the individual’s ELO compares to the opponents, but rather how their group’s average compares. Level 65’s should not be gaining the same XP beating a team of players 7–10 ranks lower than them, this is just completely wrong. If this isn’t addressed, you’ll see people duo queue boosting with extremely low level players (sandbagging), or what I’ve more commonly seen, having a single high level player hard carrying in a 5–6 man queue game where their highest rank player’s rank is diluted in the average because of the other number of players on the team.

I suggest either increasing the weighting that the highest ranked member of a team has on the group’s MM rank and/or implementing some form of maximum differential for group queuing.

Lastly, the current leaver system is egregious. I’m beating a dead horse by saying it at this point since you’ve said you’re ‘working on it’, but I don’t see how this even made it out of the PTR. Leavers go relatively unpenalized, Even if your team overcomes all odds, the team that wins 5v6 gets marginalized ELO gain.

Played Time (for Heroes)

Personally, I’m conflicted with being able to see other peoples’ played time on heroes. The biggest thing here are the expectations (both positive and negative) that are produced.

First, the negative:

Someone sees a mercy/lucio main on their team? They’ll only ever get to play that in ranked. At this point in time, I’d venture to say the average number of healing mains per team is at or below 1, so the sight of a healer main is an instant role designation for this player regardless of their personal hero preferences at the moment or their experience on other heroes (I’d venture to say the third promotion player with Pharah as their fourth most played hero at 15 hours may be far more useful as a DPS than the Genji only player that just hit level 25 with a whopping 6 total hours, but that presents another issue entirely as well).

When someone is seen playing a hero in competitive that they have little to no experience on, it gives an incentive for people to be hostile towards them. I’m even torn on how I feel about this, because to be honest, competitive shouldn’t be an arena for players to try out new heroes. However, in the status quo, griefers ruin games because of the lack of restriction on who players can play in competitive. I truly believe griefers would still find a way to be useless even on a hero they had a lot of hours on, so this isn’t necessarily the fix to that, but even people who aren’t griefing will play far below their expected skill level when playing a hero which they are very inexperienced on.

Listen, I’m all for ensuring that every single player is playing their best and/or preferred heroes, giving their team an optimal chance at victory, but this comes down to some basic Economics 101. Under the assumption that the skill levels across both teams are normally distributed and homogeneous (which your matchmaking system should do, right? [see: matchmaking group queue abuse]), true optimization occurs when the distribution of players and roles are designated in such a way that the highest amount of cumulative utility is achieved for the team, which does not necessarily mean that every individual member is on their own personal best hero.

In simple terms, play what’s best for the team. A simple concept, but something that goes completely unenforced by Blizzard, which is sad when there are a handful of potential avenues to go about solving this issue.

Now, the good:

Seeing peoples’ profiles makes for a decent way to try to work out a team composition or make hero change suggestions based on each person’s ‘hero preferences’ (but if this was intended, there could be a much easier way to demonstrate this rather than having to click on every person’s individual profile, or implement a role queue system). Further, if someone is playing a hero that they have very little played time on, I will suggest them to play one of their more preferred ones.

Being able to see individual’s play time allows for people to become leaders in competitive, to make meaningful suggestions that are backed by information that will hopefully enable win-win scenarios: the player may come to the realization that they agree and should play their more experienced hero, as well as the team performance hopefully taking a turn for the better.

Potential Solutions & General Critiques

As a pretense, I understand that many of these issues may not exist to an equal fervor across all ELO brackets, however, by opening the discussion in a constructive, detailed, and dissecting manner, appropriate actions can be taken to address the issues in a way that is ideal for all parties involved.

  1. Add a scoreboard (and preferably a post-match breakdown screen, even if it’s only in competitive). It’s egregious that this isn’t a feature to begin with, as its’ disappearance has caused a multitude of other problems to stem and ferment. With a scoreboard, individuals can reflect and assess their personal performance in order to make corrections to their playstyles and hero choices in the long term, rather than perpetuating the current standoff of ignorance and misunderstanding caused by the lack of transparency. To extend this, when players aren’t playing well, they likely will go on tilt and won’t be able to be as introspective on their own performance as usual — this is totally normal and happens in any video game, not just Overwatch. By having a scoreboard, teammates can communicate constructive suggestions that can improve the experience for both that individual as well as the team as a whole.

Ex: “Hey, I know you’ve been having a hard time putting out damage as offensive Hanzo this game, maybe switch over to a Widow and pick off their healer?”

This allows for aligning of values while accessing a win-win scenario: the player still gets to play a Sniper (their desired role), but should have more utility.

2. Role queuing system/stricter competitive requirements, similar to what already exists in World of Warcraft for dungeons and raids with LFR/LFG and iLevel gear requirements. Again, I understand that you don’t want to enforce any particular type of meta, and you don’t “want Overwatch to be like any other game that’s already out there”, but there are simply too many players who only play one particular character (dare I even say, one role) to actually tap into the entire realm of team composition strategy that you want to exist in this game, and the gameplay suffers a lot because of it. You’ve got the data, so you know as well as any of us that the number of one trick pony Genji and Hanzo mains is far too high. In League of Legends and DoTA, there exist over a hundred heroes, meaning that for every hero that one could potentially go up against, there are a handful of different viable individual heroes that can be an even or favorable matchup, allowing people to have true choice and preference while still abiding by the conventions of counterplay. Due primarily to the relatively small hero pool in Overwatch, individual matchups become less of a suggestion and more of a requirement to win or even be competitive. This game balancing strategy is perfectly fine, so long as precautionary steps are taken to facilitate truly proper matchmaking of players. See a Winson? Play Reaper and you’ll shred him, otherwise, good luck as he runs around your backline disrupting everyone short of a Zarya. Going against any decent Tracer worth her weight in salt? If you don’t have at least one McCree to stun and kill her, you’re in for a nearly impossible round (who am I kidding, Tracer is way too strong at the moment). This rock-paper-scissors balancing mechanic works when teammates are capable of playing a large number of the heroes, but in practice, matchmaking has come down to teams getting completely rolled over as soon as the opposing team presents a composition whose counters aren’t within the other team’s wheelhouse (I won’t even get into the issues with ultimate charging speed and offensive sided maps). This leads me to my next point, which is my hypothesis as to why MOST games come out as stomps in either direction.

Individuals can have extremely variable match histories not necessarily because their skill has a huge variation, but rather, that ‘one trick ponies’, are incapable of adapting to the given situation. Certain players, especially the niche hero mains(Genji/Hanzo/D.VA), will have extremely inconsistent outcomes based on the strength of the team composition matchup in each game, getting one of these players on your team feels just like losing a game to a coinflip. It’s fine for heroes to be high-risk high-reward, but your luck of getting a particular player in a game who either carries or is useless based almost entirely on the random map, game type, and enemy team comp is simply unacceptable and should not be promoted. These players, who aren’t just unwilling, but incapable, of playing other heroes, literally take their team’s chance at victory hostage. I believe this to be a perfectly logical/plausible hypothesis as to why this phenomena seems to occur, as an extrinsic factor BEYOND actual game balancing (offensive/defense sidedness, individual character balance, etc).

3. Hero stacking NEEDS to be addressed, either a hero limit or some form of diminishing returns on particular abilities. I understand you’ve said in the past that hero stacking is an intended mechanic, but I think another look needs to be taken at how this is playing out in professional and high ELO play, which will only trickle down to quickplay/competitive as exposure increases. The most frustrating aspect of double/triple heroes is that stacking essentially masks the fundamental weakness the hero possesses through sheer overwhelming, something that is typically noninteractive and not in any way enjoyable to play against. I’m warning you now Blizzard, if you continue to allow double or triple hero stacking, balancing this game will become a nightmare. You will never be able to playtest every iteration of team composition and will constantly be playing catchup to address FoTM team comps that arise from imbalanced stacking interactions that slip through the cracks.

It’s time for another look at this philosophy.

Blizzard, are you really worried about teams not being able to come up with innovative compositions? Allowing hero stacking creates a bit of a cop out answer, and here’s why (time to put my Applied Math & Statistics degree to good use). There are currently 21 heroes in the game, with 6 players per team. This is known as a combination problem (rather than permutation, where we would care about which player was playing each hero.

That means with a 1 hero limit, there exist 21*20*19*18*17*16= 54,264 potential unique team compositions in this game without hero stacking.

If team compositions are too rigid in the status quo, maybe the issue is with balancing of individual heroes, in which case, allowing hero stacking only exacerbates such imbalances. Great examples of why these systems feel so rigid are due to such balance issues with individual roles, particularly how invaluable certain abilities or roles are that have no clear alternative. There is no alternative to a Mercy resurrection or a Zarya Gravitron Charge. There is no alternative to Mercy for single target healing (Zenyatta? haha) or Lucio for AoE heals and a speedup (in addition to a pre-emptive defense boosting ultimate). That number of unique team compositions starts to quickly dwindle when there exist only two viable healer choices (Mercy and Lucio, where at least 1 Mercy is required on Attack/Defend or Payload), Reinhardt and/or Zarya nearly being a necessity on Attack/Defend and Payload, as well as Symmetra being a necessity on Attack/Defend maps.

With these core values in place, the conservative estimate of number of viable unique compositions quickly changes (I even included ‘liberal’ comps like 3+ tanks):

2(Reinhardt or Zarya)* 2(Mercy or Lucio)* 2(Lucio or Off Tank)* 19(All other heroes minus Lucio/Mercy)* 19*19= 54,872

On defense for Attack/Defense, Symmetra is a borderline necessity for the first site defense team, further lowering the number of comps to 2888. Now take relative strengths of offensive heroes into account and you’ll see that there are far fewer than this number of compositions being played consistently in solo queue. It’s not a mystery, it’s nothing that can’t be explained by some simple math.

It goes without saying that it isn’t impossible to win with comps that don’t abide by these formats, but in the status quo, deviations from this template represent the vast minority of games whose results can most likely be attributed to players being disproportionately good at a particularly unconventional hero for their rank (see: one trick pony argument), or reasons that extend beyond team composition (bad in game decision making or performance by an opposing team, leavers, etc).

Image for post
Image for post

Honestly, this is due in part to every role being so good at their job that there really aren’t alternatives to not picking up a team composition that is balanced around that. There is no alternative to a Reinhardt wall. Resurrection is an unparalleled ultimate in the same way Zarya ultimate is, however there are countless DPS heroes who all have ults that ‘just kill everything’, making them interchangeable in the context of balance. I don’t necessarily agree with the large number of ultimates that result in easy team wiping, but I digress. The game is designed in such a way to facilitate balanced team compositions but at the same time offers little flexibility for choosing within those parameters outside of the DPS role, completely not accounting for the relative balance state of the offensive heroes, which limits this further.

Many heroes can be fairly balanced in isolation, but become unnecessarily imbalanced when stacked. This also results in teams getting cheesed in Sudden Death defenses, where the defense essentially goes in blind and has no way to adapt to the offense’s composition due to the longer run back time, whereas the offense can afford to wipe once and adjust. I understand you’ve said you’ll get rid of the coin flip, but hero stacking collectively still needs to be addressed.

Image for post
Image for post

Reaper: At least in competitive and specifically, KoTH, one of the big elephants in the room at the moment. He is arguably the best duelist in the game, able to kill essentially anybody at close range, with a facerolling’ly easy escape via wraith form. Personally, I think 8 bullets is far too much for his clip, as he is capable of securing kills while missing too many shots, or simply running around with his left mouse button taped down facerolling people. 250 health is also quite a lot. Make reaper require some semblance of a brain stem beyond deciding when to drop down and use his team wiping ultimate. I don’t know how you should choose to go about it, but as of right now, Reaper has been seen as extremely powerful mainly because he has no clear counters (post-McCree nerf). The issue with Reaper being duplicated stems less from the interactions between two of them, but rather that he is simply one of the strongest, and most un-counterable heroes at the moment.

Image for post
Image for post

Tracer: No, everybody isn’t a professional level tracer who can one-clip kill any non-tank in the game, I understand that. Despite this, even in mid to high ELO & professional play, doubling up on tracers presents two hyper-mobile, high damaging, relatively uncounterable death balls. In the current ‘mobility’ meta, it makes double tracer a crippling composition to play against, especially considering how little they need to actually coordinate with their team to be maximally effective. With only one enemy tracer it becomes possible to find her or track her movements/rewind positions, but with two, it doesn’t just become doubly difficult, it becomes exponentially more difficult. Maybe nerf the amount of health regenerated on rewind (50–75% of damage recovered instead of 100%?) to make her slightly more dependent on her team, at least for heals. Again, they weren’t as much of a problem until after the McCree nerfs, but the issue with McCree as a counter for double tracer is that tracer can simply choose to not interact with the McCree, avoiding him and still being extremely powerful, while McCree is stuck in his currently gimped form trying to get kills on her. In competitive tournaments with no hero limit, Tracer has over a 130% current pick rate per team, but she only has a ~30% pick rate with 1 hero caps. Further, tracer in general is noninteractive when it comes to KoTH site control, as she only needs to tap it once. It’s the reason the only viable KoTH comp right now consists of 2 Lucio/2 Winston/2 Tracer. KoTH and site holding mechanics need to be re-examined regardless of how you change Tracer, but duplicates are significantly more of a problem than individuals.

Image for post
Image for post

Lucio: Blizzard’s already acknowledged how strong Lucio stacking is by making his auras unstackable, so I know that you’re cognizant of how overpowered some of these hero stacking abilities truly can be. His slipperiness, coupled with the constant speedup/skilless group healing becomes a fairly obnoxious combination on KoTH maps with the current ‘1-man in, full site hold’ scoring system.

Image for post
Image for post

Zarya: She is very powerful right now, able to do an extremely high amount of ranged damage with a high health pool, shield regen, and personal/ally shields with an invaluable AoE CC ultimate. I believe the stun protection from her shield as well as the duration/amount of the shield she is able to give to others is simply too high in general. When there are two Zaryas on any team, there can at any given moment be 4 invulnerable, CC-immune players, making preventing ultimates insanely difficult.

Image for post
Image for post
Photo Credit

Winston: It doesn’t really matter if a Reaper or Bastion is the counter to Winston, if 2 or more jump on anybody, they will die, instantly. Their shields also stack with one another. This mainly effects first site attack/defend, payload, and sudden death offensive cheese. It doesn’t help that Winston’s damage is easy and reliable, meaning that anybody they jump on has essentially no way to escape.

Image for post
Image for post

Soldier76: As an individual, I believe Soldier to be one of the most balanced heroes in the entire game, however in doubles/triplets, becomes too overwhelming to deal with. The combination of (essentially unlimited range) hitscan damage, coupled with controlled burst in the Helix Rockets melts squishies and tanks alike. The AoE healing should have some sort of diminishing returns. Soldier ult lasts fairly long to begin with, but when multiplied by 2x/3x, either concurrently or consecutively, becomes an unparalleled zoning tool that has no true counterplay. We started to see this same issue with McCree right before he got nerfed.

Image for post
Image for post

Mercy: Not currently a problem due to the popularity of solo Mercy or Lucio, Mercy/Lucio, or Lucio/Lucio, but I foreshadow this as being a potential problem in the future. Resurrection is an immensely impactful ultimate that should come with some form of post-rez debuff that prevents another rez within a certain amount of time. Double mercy may not be an issue currently due to how immensely powerful Lucio is, but this is something that I could see becoming an issue in the future.

DPS stacking collectively is probably the most forward facing offense(Formerly: McCree/Widow, Presently: Tracer and Reaper, to a lesser extent), as they are essentially ‘best in slot’ for their respective role (killing people). This is a pervasive issue that Blizzard will need to figure out how to deal with should they choose to not implement a hero limit. When the role of a DPS is to ‘deal damage’, and there are multiple ways to perform the same exact task (heroes), the one who does it better than their peers will see a disproportionately large amount of play time (Look at KOTH in the pro scene, it’s 2x Lucio/2x Winston/2x Tracer literally every game). In a game with so few heroes, it will only become natural for one of the very few heroes to bubble up to the surface midway through each balance patch.

4. Increase the number of role labels.

You already acknowledge the presence of snipers, builders, and healers, so why not start breaking down the damage dealers? The difference between constant DPS (Soldier, Pharah) and assassins (Genji, Tracer) is fairly evident and is necessary to diversify if one wishes to have a well varied team composition, however, many in the community seem to think as if damage dealing characters have all been created equally. Recommending a balanced offensive composition, or at the very least, acknowledging that different types of roles exist, could help to curb the stigma of uncooperative offensive players at the root.

5. Add match history.

If the idea behind viewing hero played time is to hold true, adding match history can extend the same principals to allow optimal gameplay. It feels strange to only be be able to go back and view arbitrary highlights from a game that the ever-perfect PoTG algorithm has hand picked for me. I want to see stats about the game

6. Overhaul or remove the medals system.

As per my prior critiques, in their current form, they do more harm than good.

7. Add more heroes.

Obviously, this will only take place over a long period of time. With more heroes in place, the rock-paper-scissors countering system will seem to increasingly fade as players are both afforded more heroes they could find preferable, are given more time to play all of the heroes, and find ways to play around their counters on their favorite heroes.

8. Add a match data API.

Blizzard, please, add a match data API. I would love you forever. As a bonafide data junkie, I would love to be able to be able to use data to prove/disprove some of my own sentiments, or to at least be able to quantitatively analyze things with a high end player’s point of view. Data is an invaluable tool in the arena of game balancing, and I know you’re using it, but I have a few critiques I know you’d be interested in hearing. Ex: Patchtimers for coin flip sudden death hit today, and I don’t believe lowering the timer will change much about the offense bias when the games are usually extremely one sided due to how quickly ultimates charge in this game. PS: You still gotta work on how overtime is triggered, Tracer and Lucio have a few things to say about it.


To be completely honest, if Blizzard wants their newest brainchild to be taken seriously as a competitive game, these are issues that need to be fundamentally overhauled.

For a professional team, analyzing a match’s performance shouldn’t require all members of a team to go and write down their own performance numbers before a match ends (since there is no scoreboard, post match stat breakdown, or match history). They can’t even see their opponents’ stats. No scoreboard will also greatly stifle the growth of the eSports scene, as it creates a poor viewing experience.

For any competitive player looking to improve, real time performance is critical, and with a handful of tasteful changes to the matchmaking system (queue requirements [role and group abuse], addressing hero stacking, change coin flip sudden death), the experience can be improved immensely.

For a casual gamer playing quickplay, all of these concepts still trickle down and apply — they will care about them, but will not be nearly as vocal as some of the other players.

As an arena FPS, Overwatch is an extremely new venture for Blizzard, so I completely understand why certain things like intricate hero balance or nuanced team compositions aren’t perfect in their current form. However, a few of the things I’ve discussed here are deliberate overarching design decisions that should be seriously reconsidered, or at least revisited to see if they are adequately serving the role for which they were intended.

Blizzard, you have built extremely passionate communities who care immensely about the games you’ve created, and Overwatch is no exception. I’ve spent years of my life playing your games and I don’t think any individual imbalanced patch or release will change that, but just remember that through any of the hatred and frustration the community voices, it’s only because they believe it will make a difference, that they will be heard and that the company that introduced such a wonderful experience to them will be able to work with them to make the problems go away. I’ll end this with a quote from you that ended up hitting me a lot harder than I thought it would.

Image for post
Image for post

Image for post
Image for post
The world could always use more heroes

Hey, I’m Nick Walsh. I’m a Technical Evangelist for Wolfram Research as well as a Hackathon Coach for Major League Hacking, with a passion for technology, eSports, and empowering developers & hackers. Data analysis and APIs excite me greatly.

I’m currently in the process of building a website dedicated to Overwatch statistics, matchups, hero guides, and pro analysis.

Welcome to a place where words matter. On Medium, smart voices and original ideas take center stage - with no ads in sight. Watch

Follow all the topics you care about, and we’ll deliver the best stories for you to your homepage and inbox. Explore

Get unlimited access to the best stories on Medium — and support writers while you’re at it. Just $5/month. Upgrade

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store