Overwatch: Competitive Play Update May Not Have Stopped Leaver Trolling

Overwatch [official site] devs have patched the game to stop that situation in competitive play where opponents could leave mid-match if they thought they were going to lose and reduce your Skill Rating gains at the end of a match to a sliver of what they would have been. The only thing is, the community is still reporting that the issue exists.

The idea that led to the problems was that when you’re playing competitive mode the game calculates the probability of each team winning the match – it’s part of working out how fair or even the match is. That then gets used after the match to adjust the measurement called your Skill Rating. If the game was a tough match-up and your team wasn’t expected to win then winning will net you more in the way of Skill Ranking gain and you’ll lose less if it goes badly. The reverse is true for the favoured team in the matchup.

As per a Blizzard community manager post on how the system worked before the change:

“Previously, if a player from either team left mid-match, the game would automatically recalculate each team’s probability of winning, taking a look at not only who left the match, but also when they left. The assumption was that if a team was a person down, the odds of them winning would go down, too, and that the most accurate thing we could do was account for this when adjusting each player’s Skill Rating at the end of the match.”

Thing is, that works when people are using the system in the way it was intended, but knowing how the systems of a game work impacts how people tend to use them. Thus you could reduce the winnings of another team by leaving a match partway through forcing a recalculation of the odds of winning. Heck, even if it was unintentional and you just have a dodgy connection it could still feel a bit rubbish for the winners, knowing that their win was worth less in the way of Skill Ranking just because of someone leaving.

Blizzard ended up deciding it was more important to combat the incentivised negative behaviour than enforce the accuracy of that particular system so Skill Rating gains or losses will be set according to the probabilities set at the start of the game.

The post continues:

“From a technical perspective, this will make Skill Rating adjustments a little less accurate for that specific match, but it should ultimately feel better for the players who stay and finish it out (plus, your Skill Rating will correct itself the next time you complete a match with no leavers). We also hope this will remove the incentive to “troll” a winning team by leaving, since it will no longer have an impact on anyone’s Skill Rating.”

That post was actually from last week but I’m writing about it now because Blizzard have now confirmed that there was a bug so the change didn’t actually come into effect until a day later, hence a whole bunch of confusion about what was going on and whether the change had happened or not.

Players are still reporting that leavers are having an effect on their Skill Rating gains though. Reading through the posts it’s an interesting situation where either bugs still exist/the change isn’t working as intended or there are other factors at work that are contributing to that perception of unfairness.

I remember interviewing Jeffrey Lin from Riot and he told me: “there was a time when we nerfed a particular Champion and saw the win rate of the Champion plummet a few percentage points. However, we found out after that the nerf didn’t get implemented into the patch. The simple fact that the patch notes described a nerf to the champion had a weird psychological effect on the play patterns of players and resulted in a drop in win rates.”

The Overwatch update post notes that other information is factored into those competitive match calculations :

“There are other factors that still have an impact on Skill Rating, though. Things like your team’s probability of success, your individual performance, and win/loss streaks can all affect how your Skill Rating is adjusted at the end of a match. These factors are there for a reason (to help make Skill Rating as accurate as possible), but we know that it’s not always clear right now why your Skill Rating is adjusted the way it is.

“That’s a concern for us, because we not only want Skill Rating to be accurate, but also for it to make sense. The fact that it doesn’t is good feedback for us, and seeing how we can make Skill Rating feel better/be clearer is definitely something we’ll be keeping in mind as we look at ways to improve Competitive Play for future seasons.”

Without having access to the Overwatch data it’s impossible to know which situation is actually occurring, but given it’s leading to frustration and negative behaviour regardless of the cause (i.e. if people are leaving games because they think they can reduce their enemy’s winnings REGARDLESS of whether that’s actually true) I’d say Blizzard need to find a way to offer more transparency about the way at least some aspects of competitive play are communicated.


  1. Meat Circus says:

    Hey Blizzard. Far be it from me that you’re not actually new to online, competitive gaming.

    How did you manage to screw up Season 0 of competitive mode so badly? Overwatch is great, but competitive mode is a big screaming pile of donkey balls.


  2. Unruly says:

    If someone leaves before a match ends, penalize them by exactly as much as they would have earned. Don’t just zero out their earnings, deduct their earnings from their existing total. If someone is leaving once in a while because of connection issues, then it isn’t really a problem and while they might get penalized it wouldn’t be enough to do much damage. But if someone is always leaving because their team is losing, they’ll quickly have their skill rating drop like a rock. And since those same people are the ones who often massively pride themselves on such frivolous ratings and use theirs to berate people with lower ones, it also takes that away from them.

    I’d even look at a person’s track record and possibly apply multiples modifiers to the penalty. Say you average 10% of all games as leaving before the end. In that case, you’re doing it fairly infrequently so you stay at the normal penalty rate. But if your leave percentage hits 30%? You get quadruple penalized, because now it’s at the point where you’re most likely intentionally leaving the games. Start doing that after someone has at least 100 games under their belt, so you can get a decent sample size, and maybe make it an average of just their last 100 games so people can clean up their act and see a rather rapid return to normal.

    But actively penalize that kind of crap in a way that takes effort to rebuild. Not earning more points isn’t the same as actually losing already earned points, so the people who leave like that will look at it as a zero-sum game until they start taking further losses than they would otherwise.

    • Alfy says:

      I don’t play Overwatch, but I completely agree with the principle you’re laying here. I have connections problems at times, but as you said, if that happens infrequently, is not that big of a deal: I’d like competitive games to start applying real pressure to quitters, and I’d be happy to pay the price when my own connection acts up once in a while.

      • Unruly says:

        I don’t play Overwatch either, and my truly competitive days are behind me. But people that ragequit like what’s described above still annoy me to no end. I played League of Legends in a mostly casual fashion a few years back, and the people who ragequit were also the ones most likely to do nothing but scream at people for everything that they thought they were doing wrong. Something as minor as not last-hitting perfectly? “You piece of shit go kill yourself noob, fuck off and die faggot.”

        And that’s how it seems to be in every game with a team-based competitive element. Even Battlefield, which has much less to do with individual players and everything to do with actual teamwork. And that’s a game I DO play, or at least I did before 4 came out and was too much for my rig.

    • Neutrino says:


    • ThinkMcFlyThink says:

      The problem this introduces is that people can easily “quit-down” in rapid periods to de-level and then wreck lower-ranked competition. Not everyone would mind a reduced skill rating. This could perhaps be combatted by increasing play-time suspensions penalties for routine quitting in a given time period.

      • KevinLew says:

        What you described is a ridiculously rare occurrence. There’s plenty of proof that the exact opposite is happening in almost every other competitive game out there. The primary problem with smurf accounts in games like CS:GO and Dota 2 is because when people lose their “pro level” status, they get mad and immediately create a new account so they can trick the placement system into putting them into a high level again.

        My gosh there’s three Primary Rules in Competitive Video Games:
        1) I always deserve a higher MMR/ELO/Rank than what the system states.
        2) I’m losing because I’m playing with dumb teammates.
        3) The automated cheat detection system banned me for no reason and I am being framed.

      • Unruly says:

        And if that happens, and it becomes a pattern, then you permanently place them into a separate hell of players just like them. Sort of like how some games have started moving cheaters to cheater-only servers. No ranks, only insufferable assholes that whine like no tomorrow when things aren’t going their way.

        Let them pay for and create a new account if they want to play with the rest of the people again. And when they revert to their old ways, it happens again.

    • Scelous says:

      I thought if you quit enough timed you get banned from the competitive season. So wouldn’t that resolve the issue?

    • Lamb Chop says:

      The disconnect problem would be entirely resolved if they removed the 60 sec leave option and allowed players to rejoin outside of the window. I played on a spotty connection two nights ago and lost 3-4 ranks from d/c’s that where the client restart took too long. It’s in every player’s interest to allow people to reconnect. Instead apply the leaver penalty at the moment the match ends with a certain percentage of disconnected time.

      Also, a skill rating loss doesn’t mean much if the rating is elastic enough that your higher skill will allow you to snap back. Briefly frustrating but if you play a high impact character, you can easily shift your win rate above 50% against worse players.

      • Lamb Chop says:

        Incidentally, the leave penalty is definitely still broken as of last night and is easy to tell. You get an order of magnitude less skill rating for a win against a team of 5 than a team of 6, far below the floor of what you would ever get even in a heavily favored match. If the rating change range for a non-leaver match is say, 100-200 points, I got roughly 30 points for a win against a leaver, exactly the same size change as you’d get last week. Also, Kaplan is right that what determines that 100-200 point variance is totally opaque.

  3. Monggerel says:

    Hey Lizard,
    the absolute best and brightest part of your community is the part that makes all the Tracer porn.
    The rest is gar-bagel.

  4. Krazen says:

    What I like best about Overwatch is that it’s completely team dependant but that is also it’s worst problem in Ranked Games. Unlike other most other MOBA/FPS games it is neigh-on-impossible to carry a weaker team as a solo so winning or losing seems completely random.

    • asmodemus says:

      At this point it feels almost like Blizzard policy to smooth out the competitive edges so much on games that a good player simply can’t make enough of a difference to lift up a weak team.

      Heroes of the Storm has done this to a massive extent and even comes with an 80s racing game mechanic that hugely aids the team that is behind in terms of trying to even out XP differences.

      The policy feels like something Blizzard came up with to favour weaker players and make games more casual friendly but in actuality what it means is that the first 3/4 of the game is almost irrelevant if a team has a strong showing at the end and that stronger players can dominate all game but still not make enough of a difference in a final teamfight if one player on their team simply fails to do a job. It’s actually LESS forgiving of players who don’t do a basic job.

  5. DiiGiiTAL says:

    It’s an awful system. It’s fucking soul-destroying putting in a good shift only for your team to fail massively and you get just as penalized as they do.

    • Scelous says:

      On the one hand, I agree with that sentiment completely and have been in that position before. On the other hand, Overwatch IS a team game.

      • DiiGiiTAL says:

        Yeah, totally. I just find it infuriating that they chose to do it like how you ladder in Hearthstone (lose, go down. Win, go up) when there are so many more factors to take into consideration. I can’t see me playing this anymore than a few rounds for 30mins.

        Also, not one single epic or legendary skin from free loot boxes so far, and I got several legendarys when I bought some crates. Trying not to be cynical, but meh.

  6. -Spooky- says:

    “win/loss streaks” <- with a full pub group. no one has a clue about the maps, neither the objectives. i miss the old wolfenstein: enemy territory times :/

  7. Mansen says:

    That’s incredibly sad to hear – The punishments for midgame leavers was the only incentive I had to go competitive.

  8. Kalamity222 says:

    CTRL+ALT+DELETE …. Close Program . NOTHING BEATS THAT. Blizzard cant fix this even if they wanted to.

  9. macek677 says:

    I still fail to understand how people can play competetive game which is 20ticks per second.
    Outcome of each fight is decided by computer/server not player actions, in most of the cases.

    Going even further… getting angry because of competetive play that is on 20 ticks per second servers…is plain daft. People must have boring life.