Wow, it was super weird to stumble on this thread.
In early December, we were brainstorming ways to shorten DPS queue times and the idea was proposed to try changing the role queue team composition to be 1 Tank, 3 Damage, and 2 Support. Our matchmaking engineers did some modeling around queue times, expected behaviors etc. and all indications pointed to an overall positive improvement to all queue times under a scenario like that.
From a design perspective we were reluctant for a number of reasons. Before implementing 2/2/2 role queue, we had done experiments with the team compositions being 1 Tank, 1 Support and 4 damage. The playtests under a 4-1-1 comp were terrible. The problem was the solo support. As a solo support, you felt unable to keep the rest of the team alive. Added to that, at all times, you had 5 very aggressive players prioritizing you as the target every time. As soon as the support player on one side died, the entire team would fall like a house of cards.
So when the idea of trying 3-2-1 came up, we were hesitant because of our poor playtests with 4-1-1. But we did something interesting in December and I am really proud of the team for sticking with it.
Internally, for the past two months, we changed the game to be based on 3-2-1 composition. When we queue for our playtests, we only get 1 Tank per team for each match. We decided to “live with it” for December and January because we felt like the easy and natural thing for us all to do, is just say, “hey this is different and here are all the problems with it” and dismiss the system. But by forcing the team to try it out and live with it for so long, it’s challenged us to try to solve some of the problems that have surfaced.
And there are a number of challenges.
First, there is the issue of what players have come to call “off tanks”. The biggest issues center around D.Va, Zarya and Roadhog. The current playstyle of those tanks (if you’re adhering to various metas that have existed) is to pair them with other “main tanks”. Obviously, if we were to change the distribution of roles from 2-2-2 to 3-2-1 it would require some balance changes. But possibly more than that. Roadhog is a great example. Is the correct thing to do for Roadhog under that design to try to make him more “main tanky” or is the correct thing to do to simply move him to the Damage role and balance him as a damage character? We’ve tried both in our months of testing. If you were to come and play Roadhog today with us in our 3-2-1 experiment, he’s moved from the Tank to the Damage category, he only has 400 health, Take a Breather only heals and does not do damage reduction and there were a bunch of DPS changes to his scrap gun. Now, don’t freak out. We’re not doing this for 2-2-2… I am only sharing the design experiment that’s going on here. Maybe a better direction for Roadhog under 3-2-1 is some sort of team damage reduction ability? How do you take what is essentially an “off tank” and make him a “main tank”? That is what we would have to explore with any of these tank characters. Since the Tank roster is already one of the more limited character selections, we’re obviously not crazy about removing choice from Tank players. And if our stated goal is to improve queue times, did going to 1 tank actually remove a bunch of otherwise tank players from the queue? Or were some of these “tank” players actually damage players who wanted a faster queue time so they picked Roadhog? Obviously, complicated questions and it is even more complicated to be confident in an answer.
The other fear around a 3-2-1 comp is the importance it places on the Tank player. In our playtesting, this has manifested itself in two notable ways. 1) Some players feel a lot of pressure to choose the “correct” tank. If there is only one tank, the team has a very strong opinion about who that tank should be. Another fear around this issue is that metas will be even more limited as players tend to take the “path of least resistance” towards their hero choices. If Reinhardt is deemed meta, do we only see Rein from now on? Are we all mad at our tank player if they play the hamster? 2) Some tank players felt a tremendous anxiety about their performance in the match. They felt like being the lone tank put a lot of pressure on them and if they died, it was a really big deal. Some of our tank/support players who would occasionally play tank stopped playing tank during 3-2-1 and only gravitated toward support because they felt intimidated to be “the main tank” and have so much focus on their play. Conversely, there are a number of main tank players on the team who actually enjoy the added spotlight. Traditionally in video games there are different personas that are attracted to roles in games: Tank, Support, Damage etc. We saw this in WoW and we see it in OW. While these are stereotypes (and you have to be really careful assuming too much here), there are many Tank players who love being the “lone tank” and carrying the team to glory. Watching this play out in our weird, internal experiment has been fascinating.
Our support players have given mixed feedback on the experiment as well. On the negative side, people feel bad when the lone tank dies. On the positive side, many support mains have commented that they feel like they have more freedom to focus on other players and not just “try to keep the big bags of hitpoints alive all match”. In general, these matches are more chaotic and (I’ll touch on this later) feel “more FPS-y”. The result is good or bad, depending on the type of player you are. For some of our support players, this makes matches more exciting. EVERYONE is taking a lot more damage under 3-2-1. for some support players, the chaos causes negative anxiety and they prefer just healing tanks.
The damage experience has been overall positive. For one, for those of us (like me) who usually queue for all 3 roles, we’re often pleasantly surprised to actually get to play Damage. Also, many on the team have cited that with 3 damage dealers they feel far less pressure than they do in the current live game with only 2 damage. The compositions have been very interesting. For example, you can have a 2 sniper comp (maybe Widow and Hanzo) AND have a flanker (Genji/Tracer). It’s really opened up the game. If you’re evaluating plusses and minuses, it adds to the chaos, makes the game play more like a traditional FPS (less barriers/damage mitigation going on) but also detracts a bit from teamplay.
Speaking of Teamplay, one of our testers who did not like the experiment made a comment that he felt like 3-2-1 detracted from teamplay too much. I found this fascinating because in 2013/2014 every decision we made was to embrace/encourage/force teamplay at all costs. We put so much effort into putting the focus on team victory/defeat rather than individual performance. But in 2020, I feel like the over-emphasis on teamplay (while great for hyper competitive players and situations such as OWL), causes a lot of psychological pressure for your average player just looking to blow off steam in a video game. I guess what I am saying is, that in 2020, feeling like you can deviate from teamplay a little bit in OW and have some success feels like a good thing, not a bad thing to me. The other analogy I’ve used (and I know you all hate my sports analogies) is that OW – in it’s current evolved form – feels like a football game (American) to me where every match is 4th down on the goal line. The amount of team synergy and execution required to pull off a victory is exciting… but also a little intense. There is something nice about a more loose/skirmishy version of the game – but again, it’s all opinion and perspective. Not everyone on my team agrees with me and that’s a good thing.
The other odd thing about testing this, is that when we started, everyone treated it like this was “one of Jeff’s crazy experiments” and was a super different, challenging (and possibly stupid) idea. I tried to calm people down by reminding them that MOST of the matches in OW history since launch up until the release of role queue, were actually played with one tank… or less. After a few of the early playtests in December, I remember one developer giving feedback that “the game felt like old school Overwatch… like when we first launched.” I tried to point out that the reason they felt that way is because we used to play the game this way all the time… it’s shouldn’t be that surprising or different. Let’s all remember, the surprising thing back then was actually having two tanks…
So it’s been a really interesting and fun experiment. We’re not really confident that it’s the correct thing for the game. It solves a lot of problems but it also introduces a lot of problems (like most things in the world of game design). I am really proud of the OW team for experimenting with it for the past two months. We’re really conflicted on it so it was cool to see you bring up the idea. We’d love to hear your thoughts on and opinions on it. We’ve also been brainstorming if there is a way to bring this experiment to the community somehow (either through the PTR or the live game somehow). We don’t want to freak people out though. Usually when stuff hits the PTR, we intend for it to go live (or some version of it). But this is much more of an early experiment more than anything else.
Anyway, thanks for proposing ideas like this.