Ben Bowen's Blog [Xenoprimate] • Home / Blog • • About •

Balance in Overwatch: A Critique 

It's been about 9 months since Overwatch (Blizzard's FPS-cum-MOBA) was released, and I feel like it's been a long enough time now that we can start to make serious critiques on the balance of the game, especially in a competitive setting. I'm not going to be talking especially about any particular "level" of competitive play, and though I did finally get to Grandmaster rank at the beginning of Season 3 I think the points I'll be making below are more or less universally applicable.

Quite frankly I feel that the game is a mess in terms of balance. In every other way it's extremely well polished (which is very much Blizzard's style); and that alone counts for a lot. After all, we couldn't really be talking about gameplay balance if there were bigger issues like chronic lagging, poor optimisation, or ugly/messy visuals (though the particle count can get distractingly high on occasion- but let's leave that for another post).

But in terms of pure balance the game is rather stale. I actually stopped playing competitive about 2 months ago and haven't looked back. I'm going to discuss the issues as I see them in no particular order below, and elucidate why I think they're problematic as well as what I think could be done to ameliorate them. Of course, with anything like this there's a fair amount of subjectivity, but I hope I'll at least make a cogent enough point to start some discussions if nothing more. So... Let's begin!

The Issues 

The Usable Hero Pool Is Woefully Small 

Currently Overwatch has 23 heroes. Now, it would be extraordinary if every single hero had equal playtime and usefulness; but ultimately in an ideal world each iteration on the gameplay should aim to get us closer to that ideal. In other words, perfection is impossible but getting a little better every time is not.

Season 1 certainly had some issues (first there was the Widow meta, then Pharmercy, and of course the awful 5xTracer+Lucio KOTH lineup). But I feel like the overall variety between those bits was much greater. Season 2 gave us the "Beyblade"; a meta characterised by overuse of an unbalanced support ult combined with a powerful attack ult. Season 3 has started with the "Triple Tank" meta, now evolving further in to tedium with the "Quad Tank"- sometimes seasoned with a little Soldier:76.

The problem with this progression is that each meta has become more bland and "gamified" than the one before it. The first balacing mistakes in Season 1 were characterised by damage values being too high or heals being too strong or abilities just needing a buff or nerf. There were some hugely broken balance errors but they could almost all be summed up by "people were better with this character than we anticipated" or perhaps "we made this character a bit too strong". Fast-forward to Season 3's "Quad Tank" and it's nothing to do with individual character tweaks. Instead the tank meta has exposed a fundamental flaw in the underlying design of an entire class of hero.

Let's look at the current hero pool and their viability in competitive play. A player by the name of "CaptainPlanet" has been compiling a "meta report" based on the usage of various heroes in PC tournaments since near Overwatch's release. Here's the latest graph detailing the overall usage of each hero from the most recent professional play (credit overbuff.com):


Something that strikes me is the huge imbalance here. Yes I know getting a perfectly even spread is impossible, but surely there's room for improvement here. In a game where every team must pick six different characters, only six of the twenty-three options are being picked more than 25% of the time. Or, to put it another way, this means most games are the same six heroes on each side fighting each other. That is not a healthy state for the game to be in, especially considering the whole selling point of a hero-based FPS is the draw of being able to make intelligent hero choices to outplay your enemy. That's completely gone out of the window! You could cull the bottom 17 heroes and the competitive scene wouldn't notice much.

But it gets worse...

The Most Viable Heroes are also the Blandest or Easiest to Use 

It's not a hard-and-fast rule, but it seems that the more interesting and higher skill-ceiling heroes are tending towards the lower-end of the 'viability graph'; whereas the more generic or 'boring' characters see more frequent usage. It's not a perfect correlation- after all, Ana is a great character to play and sees practically mandatory use. But she's more the exception than the rule.

As an example, take the attack class heroes:

Genji: High skill-ceiling, versatile kit involving movement skills and a mix of attack options. Usage somewhere around 10%.
McCree: Skilled character, requires aim. Hitscan aim is a somewhat bland skill in shooters these days but at least it's still a skill. Usage below 10%.
Pharah: The unenlightened would call her a "spam" character, but they probably haven't played any arena shooters. ;) The fact is she has an interesting kit with some nice movement options and a skill requirement of being able to hit direct rockets at longer distances to make her really work. Usage is usually below 10%.
Reaper: Very awareness and decision-making centered. Requires smart play to use well. Very easy to read the situation wrong with Reaper and end up in enemy territory with your Wraith on cooldown. His usage has dropped off a cliff since Beyblade was patched out, usually ~5% usage.
Sombra: Another interesting kit with esoteric and interesting possibilities. She's a newer character so we'll let it slide a little (and she is definitely underpowered at the moment), but usage is maybe 1%.
Tracer: A bit like Reaper- very dependent on awareness and decision-making to use effectively: A fun character. She's seen a rise in usage lately but take out KOTH maps and that usage drops to about 15%.
Soldier: ...And finally we come to Soldier. He's a literal CoD hero with the easiest to use weapon and a self-heal. Even his character is deliberately designed to be generic- and his kit reflects it. There's very little situational awareness required with Soldier, very little decision making (only which target to prioritise), and much less skill than any of the other attack heroes. Usage is the highest of any attack hero currently at ~25%, varying slightly from week to week.

So why is the lowest-skill-ceiling, least interesting attack hero also the one with the highest average usage rate? That is a fundamental failure in making exciting gameplay. Even worse, the two other 'S-tier' mandatory picks for competitive play at the moment (other than Ana) are Reinhardt and Lucio: Both arguably the least interesting character to play in their class. It's a basic rule of competitive game design that the heroes with the highest skill ceiling should generally be the ones that you want people to gravitate to- because those are the heroes that are the most challenging and therefore most rewarding to play.

If Overwatch wants to make itself stand out from the generic AAA shooter crowd it needs to start getting its more esoteric, unique, and interesting characters (from a gameplay sense) some airtime! Season 3 started with the "Triple Tank" meta and things have only got worse as we move to "Quad Tank".

Quad Tank... 


(Image credit Laeski)

The tank class is presumably meant to serve the purpose of being on the front-line, soaking up damage, and creating a distraction to allow the attack or defence heroes to cause havok; as well as protecting weaker "healer"/support classes. They do serve this role, but they also have way, way too much offensive capability. Why take Reaper when you can instead pick Roadhog, who has over double the health and whose weapon deals more damage per shot- as well as having the ability to pull enemies out of formation instead of requiring you to flank them? Or, why take McCree when you can take D.Va who has more health and more mobility and a better ult and the ability to absorb enemy damage/ults? The only thing Tanks tend to lack in Overwatch at the moment is long-range damage. But that's not really a problem because firstly not all maps provide room for keeping distance, secondly all but two tanks have some way of dealing with long-range threats, and thirdly no one's playing long-range damage characters anyway right now.

What's worse is the problem compounds itself. Because tanks have so much health, ordinary attack characters (except soldier) can't deal enough damage to 'soak through' the healing. This wouldn't be a problem if those tanks had reduced offensive capability and/or mobility: Fights would be long and drawn out but ultimately the mobility of attack classes would come out on top. But that isn't the case, so instead the only counter to 3 or 4 tanks is... 3 or 4 tanks.

The problem in my opinion is that Blizzard (and many others) seem to think that "Tank" is synonymous with just "lots of survivability/HP and a big hitbox". But a big hitbox is nowhere near enough of a negative to the positive of otherwise greatly increased survivability. It wouldn't be an issue but when the tanks can do most of the damage-dealing as well, you end up with triple/quad tank metas. Ask any non-tank main whether they'd trade doubling their favourite character's hitbox in size for at least doubling the health and I expect most would answer "wow, yes please!". I know I would.

So then, what's the fix for all this?

The Fixes (Potentially) 

Work Out What Each Role is Meant to do, and Design for that 

This problem manifests on the 'class' lines (e.g. attack/defence/tank/support) and between heroes within those classes. Let's start with the classes:

The 'Classes' are Meaningless 

So far for example I didn't even speak about the dismal state of the 'Defence' class usage, but it's clear that no one's interested in those heroes on the comp scene. The problem there is that a lot of the defence classes don't actually do anything particularly 'defensive' except having less mobility (which obviously on its own is hardly a benefit). The few defensive abilities that the classes actually do possess, such as Junkrat's trap, are usually easily circumvented and are not worth the tradeoff of otherwise reduced mobilitiy and burst damage potential.

In the current state of things, this is how the classes feel:

Offence: In general they have decent mobility. They do have decent damage output but arguably the tanks have at least equal: It's not uncommon to see Zaryas or Roadhogs taking gold damage.
Defence: More or less useless. Still less damage than the offence class and less mobility too.
Tank: Slightly less mobility but more everything else. More health and comparable or more damage than any other class. There's a reason "quad tank" is becoming the meta, and it's because this class is just far too capable.
Support: Admittedly this class is in a good place at the moment- even if the individual heroes that make it up are not well inter-balanced. Still, support generally means "healing" as far as the overwatch meta is concerned.

Now I'm not the one who designed Overwatch so I can't really say "that's clearly wrong", but here's what I believe a much healthier design would look like. I split the defence class in to two subsets, because I feel that there genuinely is two different types of defending:

Offence: These guys should be the ones who break up a defence or front-line. Their combination of mobility and burst damage should make it possible for them to flank and wreak havok behind entrenched defences. They should not be the highest max-DPS class, but should consistently be scoring at least 50% more damage than the next best tank! Otherwise, they're rendered completely obsolete.
Defence Type A: These guys should trade mobility (which they currently do) in order to be the highest damage dealers (which is far from the current state of things)! The reason for this is that attackers can always poke-and-retreat, but defenders have to stay close to the objective. Offense classes on the attacking side should be trying to take out supports and then keep popping up and chipping away at defences; while the less mobile defenders can punish mistakes and keep the rest of the enemy (attacking) team at bay. At least, that's the way it should be. These guys should be impossible to rush past or ignore! Straying even slightly in to their territory without distracting or handling them somehow should be a death sentence. That's the basic offence/defence paradigm at its most simple and most effective. Imagine that ideal and then think of, say, Torbjorn as he currently is and you can hopefully see the discrepancy.
Defence Type B: The other 'type' of defence character is one who disrupts, holds up, and generally divides the attacking team. Splitting a single attacker off from the rest of their team can be lethal; and Mei's wall is a great demonstration of this. Unfortunately the rest of Mei's kit has pretty much no defensive implementation whatsoever and she ends up, like many defence heroes, being regarded as just a second-rate DPS. This kind of defence hero should be a nightmare for a different reason. "They might not get the kill but as long as they're alive it's still a very bad idea to push the attack!" would be a great way to summarise an ideal character of this class: They punish you by breaking your co-ordination, allowing the Type As to turn you in to mincemeat.
Tank: Tanks should be there to distract, protect, and hold the front line. To this end giving them a huge hitbox and loads of survivability is a good thing. This class definitely should have the highest-health. But where my ideal differs from the current reality is that this class also should have very low DPS and lower mobility. That doesn't mean they should all just be human walls and nothing else- there's way more that a decent tank could do than "stand infront of hurty". Zarya's bubble shield is a great example of the 'protect' remit, and I'd like to see her kit reworked from "deal damage first, protect second" to "protect first, deal damage second". Winston is a great distracter, and I think he would see more usage if the other tanks weren't so overpowered. On the other hand for example, Roadhog is absolutely not what I would consider a tank: He's simply a DPS char with a lot of health.
Support: The name of this class is support but it may aswell be called healer: What I'd really like to see is more room for non-healer supports in the game. The Symmetra rework was interesting but so far hasn't made any impact in her usage whatsoever in competitive play. Personally I put that down to healers across the board being a bit too strong at the moment. Ana certainly has insane healing potential but even Lucio and Mercy are maybe too strong. For example, it only takes about 1.5 seconds of Mercy's healing stream to heal a half-health 200HP char back to full. Somewhat counter-intuitively this actually cheapens a good Mercy's role in the game- it reduces the importance of prioritisation and good positioning if you can just fly around team-mates topping them back up with ease the minute any of them take even 1 damage. I'd like to see it taking somewhere closer to double that (e.g. 3 seconds) or even more- and for other healers to be brought in line of course.

With those roles I honestly believe we'd see at the very least an instant dumping of the quad-tank meta; and hopefully even see a much more consistent spread across class types, opening the game back up again to interesting strategies and counterplays. With those class archetypes in mind, we can move on to thinking about the heroes that comprise them:

The Individual Heroes' Kits are not Designed with a Consistent and Unique Use Case 


(Image credit Kotaku)

A lot, but not all, of the heroes have a very strange mix of abilities that people are certainly used to but when looked at under the microscrope have very little self-consistency or clear driving gameplay design. Others have clear use-cases but are completely overshadowed by other characters with identical ones who do the job better.

Take for example McCree... He has a mid-to-long-range primary fire mode with a secondary mode that is all but useless and a short-range flashbang, topped off with a combat roll for some reason. What exactly is his purpose? One could argue that he's literally just meant to fill the "all-rounder" role but then Soldier does that a lot, lot better. The flashbang would suggest anti-flanker, but then the rest of his kit has nothing to do with that purpose and a single-shot hitscan weapon is probably the worst weapon for taking down a Genji bouncing around all over the place in proximity (unless you're an insanely good shot). It's either inconsistent or redundant, and neither is good design.

If you're particularly cynical you could say that Blizzard designed each character to design to a specific demographic first, and then tried to work out what exactly their kit was about as an after-thought. Whatever you believe, here's a rundown of the problems I see with some of the characters in the game at the moment:

McCree I already mentioned, but just to start the list I'll elaborate a little further: His kit has no clearly designed use-case and the abilities are all at odds with each other. He saw some high usage in Season 2 because of his high potential DPS, but that doesn't make his kit any more self-consistent: It just meant his weapon was the best at dealing damage for a while. The proof of that can be seen in the fact that he's been completely dumped now that Soldier's gun is better.
Soldier isn't inconsistent, he's meant to be a generalist I think and his kit suits that: Sprint, heal, shoot. But generalists are meant to be good at everything, outstanding at nothing. Unfortunately, he's outstanding at everything at the moment. But more on that below.
Pharah's kit is overall well designed, but her ult needs some work. It doesn't make sense for a character that spends the entire game moving and keeping her distance to suddenly come in close and stop in midair for her best attack. I have to say actually I'm a Pharah main and I regularly go entire games without using her ult. It's barely better than straight rockets for one or two enemies, and is a suicide button against any more. It's just the complete antithesis to the rest of her design.
Hanzo is tough to even define in terms of exactly what niche he's meant to fill. One of his abilities is more support than anything (sonic arrow), and the rest of his kit is basically leaning towards "sniper". Personally I'd scrap the scatter arrow and give him something to complement his supposed defence role, such as a fire-arrow that ignites the area where it lands and keeps it burning for a number of seconds, denying ground-access. It would also make it great when fired in to a tight chokepoint as attackers are moving through.
Junkrat's kit is in a complete state. Does he set traps or does he spam chokepoints? The answer is mostly the latter because his traps are fairly useless, but I'd like to see a character with an emphasis on traps so I'd rework his kit towards the former. To make him useful I'd make his concussion mine invulnerable to enemy damage: I know it feels 'unbalanced' but all you have to do is kill him to eliminate the threat or distract him long enough so he doesn't set it off at the right time. The beartrap might be better as a net-thrower that can be stuck to a wall or celing, making it harder to spot (It would throw a net over an enemy when they step within range, trapping them in a similar fashion to his beartrap now). His ult is also completely inconsistent currently. One idea that would gel with the "trap setter" theme could be something like an "ultimate sticky proximity mine"- it would explode for 200+ damage when anyone approaches but would be destructible.
Mei I like as a defender who can split up enemy attacks and isolate single enemies; only her wall and her ult really work to that effect however. I'd remove the 'self heal glacier' ability and give her something else to break up enemy attacks- maybe a snowball she can throw that will temporarily freeze the first enemy it hits, sealing them in ice for ~5 seconds where they can not perform any action or take any damage? And finally I would give her gun's secondary (the ice dart) the same effect as the primary: Initially slowing people it hits for a moment, and freezing them after successive hits, balanced by lowering its damage. These changes would make her the queen of intefering with enemy attacks, instead of the strange half-disruptor half-DPS she is now.
Torbjorn is probably the biggest mess of unrelated ideas of all. He's a builder (defence) that throws armour for team-mates (support) and has a gun capable of reasonable damage (DPS). Firstly, I'd make his gun heal his buildings when he hits them and abolish the hammer that was ripped from TF2 anyway. This will free him up to not stay so close to his buildings all the time and allow him to split attackers' attention between him and his turrets. Get rid of the armour throwing, and turn him in to a proper builder by letting him build a second type of building- maybe an armour dispenser. His ult is okay I guess but kind of boring: Why not complete the builder theme by having his ult let him place a static barrier, like rein's shield but immobile, with ~4000HP or so? This would again fill the defence role and make him more useful.
Roadhog actually functions okay as a 'vanguard' tank; someone you don't want to get too close to. His weapon is short-to-medium-range at best and his ult fills that role too (it keeps enemies from getting too close). The problem is really his hook- and I'm not talking about the massive hitbox here, but just the design in general. It makes him way too versatile because now not only does he have brutal close-range damage, he can pick enemies out from far away as well. Even if he doesn't get the kill after a hook, the dragging of an enemy out of position is usually a death sentence anyway. As a hard-front-line tank I think the self-heal ability is okay, so I'd just alter the hook somehow. Perhaps a reverse ability would work best, so he can hook someone at close range and then fling them far away with it. Currently he's dangerous to anyone within line of sight, but a change like that would make him very difficult to ignore at close range but less potent at medium+ distances; instead of the master-of-all that he is now.
Ana has a fairly consistent kit, but she's just too strong; which makes her edge out pretty much every other support in competitive play. I think she simply needs some numbers tweaking, I wouldn't change anything about her abilities.
Symmetra makes very little sense as-is and reaches almost Torbjorn-levels of mixed-up kit. Her recent rework was meant to make her viable on attack (thereby fulfilling her role as 'support' and not just 'defence') but I really don't see it. Yes, the new "photon barrier" can be useful but if anything that's more of a tank or defence ability. If she's to be a support character then give her something to actually support her team other than her new ult (the shield generator). Ideally I'd say scrap the turrets and the photon barrier and make her kit work around shields and armour. Maybe give her something like an 'armour grenade' akin to Ana's that gives friendlies +75 armour and removes all armour from enemies. Her second ability could be something where she drains her shields to momentarily 'power up' friendlies' damage or speed (100 shield depleted for +50% damage for 4 seconds, or something). These are just two ideas and are probably not perfect but they would seem much more consistent with a 'support' role, in my eyes.

Of course all of the changes above are just ideas, and very subjective, but I hope that I'm demonstrating my point with them: The current kits for a lot of characters are very scattered and with a little creativity and thought could be much more self-consistent. The benefit of this would be each character having a very specific use case: This makes them all more likely to be picked (as none does any job better than the other) and makes them easier to balance, because each character would exist in more of a vaccuum relative to the others.

Make the Most Interesting Heroes the Most Useful 


(Image credit Backwallpapers)

The final thing that I'd want to see before I'd ever go back to competitive play would be making the 'fun' characters the ones I can play without being shouted at by five team-mates before the game's even begun. Of the seven 'viable' picks (Soldier, Rein, Lucio, Zarya, Roadhog, Dva, Ana) I only really have any interest in playing two of them (Zarya and Ana): Four of the other five are very basic in FPS/shooter/strategy terms.

Soldier is obviously a generic "AAA CoD-style shooter", Reinhardt's shield is his most interesting feature and even that isn't as original as you might think, Lucio is literally playable braindead (yes I know people like DSPStanky put a lot of effort in being a great Lucio, but it's not required), and Roadhog is just point-and-click. D.Va I don't play often enough to give a qualified opinion.

Decent game design dictates that the characters with the blandest kits should be slightly underpowered to offset the fact that they're easier to play. Because heroes with higher-skill-ceilings demand more perfection, the rewards for playing them should be higher. But unfortunately the attack character with the best consistent DPS at the moment is also the blandest one.

It's not hard to make this mistake, after all: Of course Soldier will be the best DPS character if you buff him too far because he's also the easiest to play. There's a fundamental failure to understand somehow that 'generalist' characters need to be slightly underwhelming otherwise they're the only ones who will be picked. People will argue that you can technically get higher DPS out of McCree and it might be true, but it requires too much perfection- even the pros can't do it and that's why Soldier is the default pick.

I was playing some Pharah on quick play the other night and something occured to me after fighting a Soldier at medium range and losing. Although I fired two and hit two direct rockets he ended up killing me because he was stood in his own biotic field, and his assault rifle still deals 20 damage per bullet at that range. That means in the time that one of my rockets had travelled a fair distance (and I'd led his movement and hit him square in the face), he'd equalled my damage by hitting me with only 6 bullets from an automatic weapon. There's something wrong here: Hitting direct rockets is a universally harder skill than hitting someone with automatic weapon fire. You could argue that maybe someone with very good aim should be comparable here but only needing to land six bullets to stay EVEN in damage with a PERFECT Pharah player is frankly ridiculous: Why would you ever pick Pharah, or any other character that requires more skill?

Look at some of the characters with <= 10% usage rate in competitive at the moment: Hanzo, Reaper, McCree, Mei, Pharah, Genji. All much more interesting to play than Soldier, yet almost completely untenable from a competitive point of view. We can make the same case for the supports: Would you rather play one of Zenyatta, Mercy, or Symmetra instead of Lucio? I know I would.

The same discrepancy of input to output that we see with Soldier can be applied to Lucio: Lucio is just too invaluable for the amount of effort he requires. He's just too good at healing everyone without even trying. Personally I'd either keep the automatic line-of-sight heal but let it only heal one person at a time (prioritzing the weakest) or keep it as AoE but drastically reduce the range.

Until I'm not forced to try and quickly pick Ana or Zarya in competitive in order to have fun, I'm going to stick to quick play where it doesn't matter if I play a character that I know is not the optimal choice. Some may say that that's what quickplay is for but I disagree: I like competitive gaming, but I don't want to make a laborious 'job' out of it- the term, after all, is competitive gaming. :)