To contribute to the main discussion: the developers said MMR and the matchmaker as a whole has the goal to create matches that are as close as possible (meaning each team has a as close to 50% possibility of winning as possible).
To me this is not rigging and is proper game design.
I want matches where I can be that 1% that makes my team win.
I’m happy they implemented it this way and I really hope they won’t change its fundamental concept.
I welcome any improvement in the accuracy tho (lately match quality has been quite volatile).
And to be clear, devs or whoever - I consider to have high elo ego when they classify someone’s intelligence based on what rank they are in. Not allowing a 3 stack in gm in fear of exploiting synergies is a form of high elo ego. And to think that’s it’s “fact” or have the reasoning that “well if they knew about it they wouldn’t be in that rank” is so ignorant of so many other variables.
Especially since they’re so much online vods and streamers. Just cause I’m gold doesn’t mean I don’t understand the importance of high ground or toon synergies.
Stating that only masters or gms would only recognize a bad performing range hero, and thinking it’s fact, stems from an ego.
Honestly that’s why I think people are upset when 1 gold can criticize another gold. “Well if you knew better you wouldn’t be gold”. Just ignores and ignorant of all the research I’ve done to recognize a failed play. Me not being able to overcome failed or deficient play (and yes sometimes my own play) is why I can’t do unranked to gm.
But a system that purposefully puts people that have a likelyhood of deficient play (people who do NOT play a lot) with people who DO have consistent play (consistent at a rank and has played a lot), is wrong. And does not conduct within the best interest of the players. It serves the interest of itself to figure out the certainty of a player.
I’ll even go on and say that when matches have only players of certainty it might even throw it out cause the system believes those games are “boring” and it won’t learn anything. But the players in the lobby could have had an incredible game if it were allowed to continue. Imagine queue time decreasing if they allowed more “boring” matches.
The matchmaker also has a responsibility to respect SR as the only tangible measurement of skill that players have. It’s not fair to make a match based on MMR when nobody can even see what their MMR is. Anyone with high MMR compared to their SR is getting rigged by the system.
Every game will usually come down to which team performed better than expected. Being that 1% difference is a separate standard for each player depending on how much they are handicapped.
You could single-handedly provide 50% of the overall value for your team but if the game expected you to provide 52% you will still likely lose. Same goes if you are expected to provide 1% of the overall value to your team and you end up providing 2% instead. In this case, would you really feel like a difference maker who earned the W?
Nothing competitive, in general uses hidden stats for match making. Could you imagine the outrage if the Olympics started using hidden stats instead of the black and white scores? Chess? Professional sports? Amateur sports? Poker? Shall I continue?
Tbh, I like the idea of having the whole mmr/sr thing hidden and I’m really happy it is in discussion for OW2: not having the psycological hurge to keep that number growing would make my games so much less stressfull.
EXACTLY. This is the simple, absolute truth of the matter, boiled down to a single line.
The smirking insolence of Blizzard and its fanboys on this subject is totally intolerable. It’s as if they think competition didn’t exist before the days of computer gaming, Bayesian mathematical statistics, and automated matchmaking. This myopic perspective is why they can’t see why algorithmic handicapping is wrong, or frankly even understand what it is. They’ve never known or imagined anything better.
The problem is, your rating is still based on relativistic performance measurements instead of your ability to win matches, probably even more so. It’s not a true ranking system for competition.
You sound like just the type of player who blizzard targets for engagement by keeping you in close games regardless of your personal performance.
Personally I care more about the 12 players in a match being as close in skill as possible than the overall team balance. Games are 10 times more fun when it isn’t immediately clear who the best and worst players in the game are less than 1 minute into a match.
So there are a few interesting aspects to consider here. The first is the accuracy of ladder rankings. And that’s why the games need to be matched as evenly as possible. If the matches made are more lopsided, that 1% difference in performance is not impactful. If, however, the matches are evenly made, the 1% difference can be decisive.
That’s why, if we want ladder rankings to be as accurate as possible a reflection of skill, we want 50/50 matchmaking- precisely so that we have the best opportunity to make minor differences in performance impactful. In so doing, the ladder rankings are able to respond to players who perform even slightly better than their current ladder ranking would suggest.
And not everyone has to care about that, there are certainly players who would prefer to face lesser competition even if it meant that ladder rankings would be less accurate. But it’s still important to understand why 50/50 matchmaking improves the accuracy of ladder rankings.
The second point to consider is that the scenario you suggest (one in which one player on a team is expected to provide only 1% of the team’s performance, while another is expected to provide 52% of a team’s performance), is overwhelmingly unlikely to actually occur. The matchmaker will not make a match like this intentionally (on either end of the scenario, neither a match where one player is expected to provide little or no output nor where one player is expected to provide half the team’s output). You would only see a scenario remotely like this when players are grouped to the limit allowed. And even then, the scenario is unlikely to occur.
But even if we allow this hypothetical for the sake of argument, the scenario would be one in which the lower rated player is playing in a game that is 1000SR above their rank. So a high diamond player playing in a match with high GM players, for instance, or a low silver player playing in a match with low plat players. And even then, the lower rated player would be responsible for more than 1% of the team’s expected output. But if they overperformed in such a scenario when they were being exploited by much better players on the enemy team (because they were the lowest rated player on their team), the ladder should factor that new information into the rankings.
Players should be rewarded when they outperform the expectations of the average player at their rank. That’s what we want of competitive rankings. And it is less likely to happen if we don’t make 50/50 matches.
The next point to consider in your proposed scenario, is the perspective of the much higher skilled player. And one thing to note there is that competition breeds excellence. Yes, we could put these players into more games that are not competitive matches- games where they would simply steamroll the opposition. But from a competitive standpoint and from a standpoint that recognizes and seeks to foster excellence in play and to increase the overall skill of play, those matches are pointless. They are time wasters.
It’s like having professional players play against middling high school teams. The pro players neither demonstrate their peak skill, nor learn anything from those matches. 50/50 matchmaking is better from this perspective as well.
And I feel like people need to at least understand the reasons for 50/50 matchmaking before any sort of productive conversation can occur. Because there are certainly failure states and issues that arise, but the alternatives that people propose will generally exacerbate those failure states and issues and people cannot see that because they do not understand why the current systems exist and how they operate.
There’s a lot of misinformation swirling about and very little effort to explore and understand.
There’s always uncertainty. No one has sandbags put on them. Every match and every player on the ladder has uncertainty. It’s just that there is higher uncertainty with new accounts and uncertainty increases when someone vastly underperforms or vastly overperforms.
That isn’t something unique to Overwatch. That’s just what uncertainty means in a predictive context.
How bout identifying non casuals with casuals. Call it what you want.
Using averages with pbsr can be improved.
Mixing casuals with non casuals is wrong and frustrating.
If you are going to add uncertainty sandbags, put them on the same roles! Putting them on different roles is another issue, and why stacks are so powerful.
It is my belief if someone who just plays 5 games a season, is not going to have the same game sense with someone who plays 50 games a season, and might practice in custom game or aim train.
But who just play 5 games is not going to randomly paired with another 5 gamer. They are purposefully paired with the 50 gamer.