The argument against PC/console cross play

General Discussion
1

Blizzard’s implementation of cross play (specifically - PC/console cross play) has generated quite a bit of controversy. I’m seeing a lot of misrepresentation of what the argument against it is, and a lot of bad counters to this argument. I wanted to explain what people are saying in the most distilled form.

The argument has two basic components.

1. The matchmaker is not as accurate at making balanced matches when there are PC and console players in the same pool.

This is not because console players are worse than PC players. It’s because the numbers that the matchmaking algorithms input (MMR, stats, etc) are not directly comparable between PC and console.

If you make two clones of a given player and place them on opposite sides of a match, one on PC and one on console, then on average the clone on PC will have an advantage over the clone on console. It doesn’t matter if their MMR is the same on both platforms. It doesn’t matter if their stats are the same on both platforms. The average advantage will persist. The matchmaker thinks that it did a good job balancing the match by placing these identically skilled clones on opposite sides, but in reality it’s created a disparity.

Blizzard themselves have let us know in no uncertain terms that they think that the matchmaker will be less accurate when the two pools are mixed. If they thought it could still create balanced matches then they would have put cross play in competitive.

If you think that mixing the pools does not affect the matchmaker’s accuracy, then you should also think that implementing cross play in comp is fine.

Note that for this part of the argument, it doesn’t matter which game mode we’re talking about. The matchmaker’s accuracy is either affected, or it isn’t. You can think that the accuracy doesn’t matter in quick play, but that’s not relevant to part 1.

The matchmaker’s job is to make sure that matches are as close as possible and that one sided matches are as rare as possible. An inaccurate matchmaker will makes one sided matches more common. This leads us to the second part of the argument.

2. A large portion of the player base enjoy close matches more than one sided matches.

This is true for people who take quick play or arcade seriously, for people who use quick play as a warmup, or as a place to learn new heroes, or as a mode for chill games where they try their best but don’t sweat it if they lose. A close match is more fun than a one sided match.

That’s it. That’s the entirety of the argument. People want close matches, PC/console cross play leads to less close matches, people want a solution to be found.

I think that you can’t hold the views that PC/console cross play should not be in comp, that close games are more fun than stomps, and that there is no issue with the current implementation of cross play and remain consistent.

A lot of the counter arguments I’ve seen don’t seem to hold water. Here are some of the common ones.

  1. It’s just quick play, it doesn’t matter. I dislike this argument because it’s so dismissive of other people’s play styles. It also has no bearing on the issue. Most people enjoy close matches more than one sided matches, regardless of the mode. That’s why the game has a matchmaker. It clearly does matter how often the game gives you one sided matches.

  2. You’re just being an elitist PC player. You hate console players and don’t want to play with them. I understand that some people are hearing this, but it’s not what most people are saying. It’s not about console players being worse than PC players, it’s about the matchmaker being worse when the queues are mixed.

  3. Some PC players play with controllers. This is true, but immaterial. As long as they only play in the PC queue, the matchmaker sorts them into the right games.

  4. The matchmaker will sort this out over time. This is false. As long as console players sometimes play in the PC queue and sometimes play in the console queue, the matchmaker can never sort them into the right PC MMR.

  5. Some console players are amazing and will destroy PC players. This is true, but again - it’s immaterial. The accuracy of the matchmaker is a question of averages. The fact that some console players are amazing doesn’t change that.

  6. My friend is top 500 on both console and PC. This may true, but it doesn’t matter. It’s an incorrect comparison. If your friend was cloned and played against themselves (which is closer to the situation at hand), then the clone with the KB+M will, on average, have an advantage against the clone with the controller.

The only counter argument that I can think of that makes sense to me is that the price is worth the benefits.

Time will tell what the price is. My games over the past two days have felt a bit more special, but I’m waiting for a larger sample size to get through my confirmation bias. If it ends up being noticeable, then this will be quite jarring given that many PC players and console only players (who have had their console specific balancing removed) are being asked to pay the price for a feature that they’re not reaping the benefits from. It’s especially jarring since the difficulties could have largely been avoided with some simple tweaks that would have made the controversy disappear.

6 Likes
2

… Why not? That’s the entire purpose of MMR.

You’ve seen this working in a very visible way for a year and a half already with role lock. If you sometimes play Tank and sometimes play Damage, you have two different values for SR/MMR that diverge as you continue to play. Your Tank MMR doesn’t become invalid every time you play a few games as Damage, nor would your PC pool MMR become invalid every time you play in Console pool.

3 Likes
3

No most people are angry that PC are now region locked…

4

This is very different than the role lock situation.

Your tank MMR only applies to your tank matches, your DPS MMR to your dps matches, and your healer MMR to your healer matches. They diverge over time to reflect your true skill in the different roles. The separation of MMR into 3 more specialized MMRs makes the matchmaker more accurate.

The PC/console situation is exactly the opposite - it’s not separating MMR, it’s using the same MMR for two very different situations: the console queue and the PC queue. If they used separate MMR’s for those two queues then there wouldn’t be a problem.

Think about it this way. Imagine a college student who is an avid Overwatch player. They usually play from their dorm, where they have an old laptop with a bad connection, a lousy monitor, a small desk, etc. They’re a good player, but their hardware is holding them back.

The student visits their parent’s house once a month for a weekend. At their parent’s house, they have a fully set up gaming rig with excellent hardware.

When they’re at home, they will play better than they usually do. Their matches will be more one sided than they should. Once the weekend is over, they go back to their dorm and play on their old clunker.

The matchmaker will never be able to find their true home skill and dorm skill in this situation. It’s not built for that. Since they play mostly in their dorm, it will mostly reflect their dorm hardware limitations. Every time they go home, their matches will be unfair.

This is what’s happening with cross play. As long as a player plays any significant portion of their time in the console queue, their MMR will never be accurate on PC.

That’s one thing people are angry about. I’m seeing a lot of anger towards the cross play situation as well.

2 Likes
5

Okay, so your entire premise is based on a blind assumption that the devs for some reason decided to persist the same MMR between these two completely different matchmaking environments, despite all historical evidence indicating that they wouldn’t do that.

You don’t even share the same MMR between Quick Play Tank and Competitive Tank on the same platform, and those are functionally almost identical modes of play, only with different players. Why would you ever assume that you’d share the same MMR between two completely separate matchmaking pools where you don’t even have the same control settings?

A professional developer would never intentionally implement it the way you’re describing, so unless there’s some actual reason to believe that they did, assume that they didn’t, and your problems are solved.

1 Like
6

Can confirm. What makes an excellent player on console can be totally different.

For example: sym and torb have lower damage output on console due to reduced turret damage.

Xbox players have also by and large taught the matchmaker that 1 op carry player exists per team

7

Huh? Most mmr resets have been accidental. Rofl.

8

The discussion isn’t about resets…

9

Those are part of “all historical evidence” fam, you good? Get enough sleep?

10

I don’t understand what you’re getting at; are you just trying to point that the devs make mistakes sometimes? The time they accidentally zeroed MMR for a day a few years ago does not seem particularly relevant here otherwise.

11

I agree with the entire post aside from the assertion that MMR can’t be tweaked over time.

I hope that this calibration is aggressive to result in fair matches sooner.

12

Supposedly the MMR reset with this crossplay patch, I think people assume this because they are getting GMs and bronze in the same QP match.

I think it is because the GMs are in stacks with people well beneath them ruining the match for everyone, rather than the match making, in this instance anyways.

13

You are asserting that all historical evidence would suggest that we have a separate quick play mmr from quick play… For each role.

I just believe that they are using the same mmr data from quick play in quick play. Difference of opinion, I suppose.

14

I disagree that quick play and arcade games are not usually the place to look for close matches since you see players from different ranks since the restrictions is looser for quick play. I mean you could have a gold, diamond, and a master player on the same team.

If you want closer games then I suggest trying out comp.

I see cross play as just adding a mix bag of jelly beans to a pool of mixed jelly beans.

So far my games have been the same as usual and I haven’t seen any glaring dip in quality.

15

I understand you - at the beginning, they don’t have much data to go on so MMR might be applied the same but that is not to say that another MMR can’t be applied from this point for the separate pools.

It would make complete sense and I’ve also posted believing this would be the case.

Your MMR will set you free from peasants

Crossplay problem is a matchmaking one

1 Like
16

You’re welcome to believe that if you want. I’m just trying to figure out what past MMR resets would have to do with any of that. There’s no reset happening here, regardless of whether your assumption that they’re implementing it badly is correct or not.

17

If they had a system for separating the MMR’s that they were confident in then they could:

  1. Implement cross play into competitive.
  2. Tell people about it. They’ve been pretty forthcoming about these things in the past. They had to have known about the controversy their announcement would spark. They could have avoided or quelled it entirely by explaining that there will be a separate MMR system. Instead, they chose to lock PC players into the system and then stayed completely silent about any MMR system that could fix their concerns.

Until I see evidence otherwise, I’m going to work under the suspicion that no such system exists, or if it does - that they’re not confident in it.

It can be tweaked over time - by Blizzard. It won’t adjust itself on its own. This is my preferred solution to the whole cross play issue.

Quick play represents a certain compromise. You put up with a somewhat looser matchmaker in order to gain several other benefits (less pressure, quicker games, etc.)

However, it does have a matchmaker, and this matchmaker produces many more well balanced games than random chance would. You sometimes get gold and masters players in the same game, but you usually don’t.

The problem with messing with the accuracy of the matchmaker is that you get less accurate games and more inaccurate ones. It changes quick play not by nature, but by degree. The compromise becomes worse. If you don’t like licorice flavored jelly beans (and what kind of monster does?), and someone pours a bag of jelly beans with a higher proportion of licorice into a bowl that has a lower proportion of licorice, then your bowl will become worse.

My games seem to be somewhat worse so far, but I’m giving it time.

1 Like
18

It’s possible that they intend to implement it in Competitive later, although I doubt it. As we’ve discussed, there’s a significant ramp period where the system won’t be able to confidently match players in cross-play, because it has no historical data for those players in the same population.

Normal Competitive play was not immediately available for Overwatch, either. They waited a few months to let the MMR distribution establish itself a little, and players were/are required to level to 25 to make sure they have enough individual data to at least make a reasonable guess at their initial placement.

Mostly I think it’s unlikely to appear in Competitive because of conflicting goals between the two features. Crossplay is intended to allow players to play with any of their friends, regardless of skill level, platform, anything. In contrast, Competitive mode prevents players with significant skill differences from grouping together, and doesn’t even allow more than two players per group at all above a certain threshold.

I’d say it’s fair to criticize their communication here. The amount of negative feedback that’s appeared around crossplay, even if most of it is related to major misconceptions of what the feature even does, is evidence enough that it was not explained to the community well enough.

That seems to be a fairly frequent problem for the Overwatch team. I’ve quite often found myself here on the forums trying to clarify details of new features/changes for people. I don’t receive any additional information from Blizzard to enable that, however. My only advantage is that, as an experienced game developer, I usually recognize the underlying reasons for why Blizzard makes the decisions they do.

On this particular point about diverging MMR, though, I can relate. If I were developing the feature for placing console players in the PC pool, the thought never would have occurred to me that the community would just assume I had no idea what I was doing until directly stated otherwise. I wouldn’t have thought to mention that different populations use different MMR, because that’s the only correct way to implement it, so of course I would implement it that way.

But even beyond that, the developers themselves are not typically the ones who write articles to explain things to the community. There are usually other, non-technical staff members responsible for collecting information and writing about it for public consumption, and this question is likely not one they would have asked.

1 Like
19

It’s entirely possible that you are correct - that Blizzard implemented a separate MMR for console players in the PC queue, that they considered implementing it in comp eventually, and that they simply did a bad job communicating it. This is not an impossible theory of reality.

I think this is less likely than you’re making it out to be. Using the same MMR for both populations in quick play is a more parsimonious approach. I don’t think that it’s a good approach, but quite often these sorts of compromises do win out at the end of the day in large development projects.

They knew that PC players were going to be upset about the new system, which is why they purposefully prevented them from opting out of it and never mentioned the reason. They stressed quite strongly that cross play won’t be implemented in comp. They removed aim assist for console players in the PC queue (which they wouldn’t have had to do if there was a separate MMR). To me, this is all clear indication that they realize that there is a problem and don’t have a solution for it.

I think they would have been very likely to acknowledge the issue and tout their solution to it in the developer update, the FAQ, or during the ensuing furor. It would be incredibly easy for them to do so.

I’d very much like to be wrong about this, but I am thoroughly unconvinced that I am.