What Did Blizzard Change?
- The maximum Skill Rating (SR) difference between the highest SR player on a team and the lowest SR player on the same team will be more restrictive.
- Personal Performance Skill Rating Adjustments removed for Diamond-tier players and above.
- Full Dev post here.
When Do Changes Take Place?
Season 8. Jan 2nd, 2018.
OWL Pro Player Thoughts
Félix "xQc" Lengyel (FUEL Tank): "I think we'll get better matches on the ladder, of course, that's the whole point. But I think it'll make longer queue times. Obviously I'm a streamer, so I think about that stuff.
But I think it’ll make a more reliable practice environment for the players, because once you’re done with scrims, you play comp, and sometimes it’s late at night so you'll get really unfortunate games. On the West coast, you'll get like Diamond/Master games when you're like 4,500. So that's currently not optimal for practice. But I think it’ll get better. And like I said, it’s a more reliable environment, so I’m looking forward to it."
Aleksi "Zuppeh" Kuntsi (MAYHEM Support): "I personally think it’s a good change. There’s a lot of one-tricks right now, so it kind of will drop them out.
So, the comp ladder will be more competitive, which is always good for the esports scene. If we get the public ladder to be really good, everybody will play, then everyone will tryhard,
Johan "CWoosH" Klingestedt (MAYHEM Flex): "It’s especially good for our tank players because I haven’t been getting a lot of SR and now I can finally get some!"
Ted "Silkthread" Want (VALIANT DPS): "I think that's actually a really good change by Blizzard.
I think performance based SR gain is a huge mistake in any team based game, because there's too many variables for statistics to be able to actually genuinely mean something. And so to play for statistics, to reduce the amount of SR you lose per game, is just a selfish way of playing and I don't think it's how you should play a team game like Overwatch is and ultimately with that I think it will benefit the competitive system overall and you'll see more people able to climb leaderboards more consistently and you'll see more talent because right now it's hard to tell real talent from people who just cheese their way to the top."
Sebastian "Numlocked" Barton (VALIANT Tank): "At the start, it's gonna be a bit rough as the one-tricks kind of fade out of the high ranks, because performance-based SR has been giving them a boost. But, eventually is going to kind of even the playing field and it's gonna be a lot better at high levels."
Josh (Lead Overwatch Producer): These are solid steps in the right direction.
Considering how heavily Blizzard has been emphasizing teamwork since the announcement of Overwatch, it seemed really counterintuitive to have a ranked system in-place that rewarded selfish play based on individual stats.
Everybody’s going to have bad games from time to time - it’s a natural part of a competitive system. Punishing players for downswings was probably one of the worst ways to encourage consistent engagement. “Oh, you played against a few teams that were REALLY good at diving supports? Yeah, that’s totally your fault. That’ll be 30 points, please!”
I’m very glad that we’re moving to a system that emphasizes winning over making yourself look good. Every game of OW is different, and it’s not always the top fraggers who carry the game.
After all, even if you’re decked out with four gold medals, an L is an L.
Sharon (Editor-in-Chief): Overwatch's ranking system is based to some extent on the Elo system (shoutout to Arpad Elo and all you other Hungarian physicists). The idea is to give more points to an underdog winner, and fewer points if the winner is the... overdog.
This algorithm gets coupled with a couple statistical principles (massive simplifications incoming):
- If you are better than the players in your rank, you'll win most of your games until you reach a tier of proper skill level.
- You have teammates, but if you always play consistently, your team has 5 (low) chances to derp, and the other team has 6. Inversely, the opposing team has 6 (low) chances to have a superstar, but if you're better than you're ranking, you're a guarunteed superstar, and your team is more likely to have more stars.
This stuff gets muddled in small scale moments (losing streaks, rage quitters, etc.), but with statistics, it's all about likelihoods and large sample sizes. In theory, if you keep playing consistently, you'll end up where you should be. Most people don't enjoy that day-to-day experience, though, and that's largely why companies employ personal performance ratings. But they ain't perfect. Right now, the system isn't smart enough to capture actual merit in a lot of cases, and at the worst, players are attempting to game the more simple parts of the algorithm.
Nobody has perfected a ranked system. The original Elo system even has its flaws (sad Arpad). In that case, you try to employ whatever works best for each section of your ladder (systems perform differently at different ranks for a lot of complicated stat and human reasons). So yay to change, yay to getting data on whether or not it helped, and yay to developing something better in the meantime.