Algorithmic Handicapping (MMR/PBSR) is WRONG for Overwatch 2

General Discussion
61

I believe the portion of the community that actually cared about this left quite a while ago when they realized blizzard was never going to do anything about it. Blizzards core design philosophy is to make games addicting and the patents you shed light on prove that. They’ve been prioritizing addiction over competitive integrity from the start and I’m not surprised.

The devs are making changes to the competitive system for OW2 so we’ll see how that turns out. Could be quite interesting with the move to Microsoft as well. You’ve done the community a great service, it’s a shame not everyone is able to see that. Thank you for all you’ve done and thank you for continuing to fight for the clueless community that has no idea what’s going on.

2 Likes
62

In this system there are periodic learning phases, these are needed to estimate how likely it is that one team can win against another team. This is no secret and has been officially confirmed several times.

In theory, this is a good idea, as long as a certain number of players don’t misbehave.
(cheating, throwing, toxicity)

  1. Remove the expected score based on a rank’s training data and replace it with actual data for each player. (Start with ZERO data, that would be an honest approach and reset every season)

The main reason why players don’t always want to win is that they don’t have fun or motivation. This may be because of a teammate’s toxicity or because they are forced to do things in the game that they don’t really want to do.

Once players are having more fun again, matchmaking will automatically improve as a result.
Still, not a good idea for an e-sports game.

1 Like
63

This isn’t a thing though. The whole game is a learning phase. Every match is a learning match. When you play a match, the game predicts the outcome before the match begins, but the outcome of the match is used to adjust the prediction for the next match.

So it’s not that there are periodic learning phases. It’s that every match provides new information. They did do a soft reset of the player data at one point. But that introduced a lot of volatility into the system. There’s no way they’d do that periodically. (Though they may decide to do something like that with the official launch of OW2- it would be the best point to do it, if they were going to do it again.)

This would exacerbate all of the problems people have with the current system. It would be the equivalent of forcing people to smurf every season, which would have two major outcomes- matches would feel terrible for much of the time, only stabilizing to the point we are at now in the latter half of the comp season (and even that outcome would only happen if the majority of players played a bunch of matches during that initial terrible part of the season), and the overall rankings would be less accurate.

1 Like
64

Wouldn’t doubt it one bit.

65

You again?

Information > Fake News

66

How many threads are you going to make saying the same thing no one cares about but just worded differently. Go write a school essay or something with the time and energy you put into these useless forum posts.

5 Likes
67

My supporters are generally more coherent and well-spoken than my detractors, it’s true. I won’t even call them ‘critics,’ because they say nothing substantively critical. All they do is screechy ad hominem attacks and mewling cries for censorship. Nothing but haters.

But I have never made an alt account to play Overwatch or post on this forum. You are observing a real correlation between the position that players take in this debate, and their post quality. Most of the detractors have not even read the original post, so how could they possibly write about it?

2 Likes
68

It is clear from the participation in said threads there are actually many forum members who care about this

1 Like
69

Because it’s the same thing, and said by others than you in both far elegant/crude manner which all comes down to the same thing.

There isn’t any handicapping, and at most, the matchmaking is doing its best to ensure matches are often at 50/50. Nor are there any of the force wins and force loss.

A force win/loss would mean that in the case even if you win/lose, the system will go ‘No, the number says you lose/win so you’ll LOSE/WIN PERIOD’ regardless of the actual results of whether you actually winning/losing the match. (I.E, you win the match but force loss means it’ll say it’s defeat, completely overwriting your results regardless of your actual result.).

You can get terrible teammates, and terrible synergy (Vice Versa), also smurfs and such in the current system. It’s not perfect, as it is something the matchmaking can’t account for, nor can it really solve the problem (Even other bigger, better and more experienced companies struggle to solve this issue.). Mitigate, yes, but it still exist.

However, the system put in place always try to ensure that players get an even match as much as possible. Which is shown as eventually, WR always come down to 50% since you climb as good as you can, then when you lose or face people of equal skill, it stabilize until you improve or drop. Once drop, you either maintain or climb again until you’re back to your skill range until you hit a peak before the cycle continue again.

Way I see it, you’re not asking for a fair and square matchmaking system, you’re looking for an actual rigged system that gives you more wins than losses.

Am I bias? Yes, absolutely, a side effect from hanging in the forums perhaps but I’ll flat out admit I’m biased because an actual, wild west, no SR/MMR system is far more chaotic than the current system done in place by OW. In which, I can guarantee you that you’ll be screaming to high heavens to fix because of how chaotic it is (Or maybe not, since with how wide the difference is, you might actually be noob stomping all the time.).

Also, if you feel that strongly. Join blizz and go give your hand in trying out to actually fix this thing. What better way to fix something than being behind the scenes and part of it. You could do your analysis, objections and search for justice while getting paid for it. Unless you want to add in the spiel of how terrible it is to work for blizz (Which TBF, makes sense, won’t say it’s not terrible.) or the system.

Maybe you could actually put those 5 years to good use instead of just spamming and cluttering this forum.

(Also, trying to correct my usage of english? Get off your damn high horse already.)

2 Likes
70

My Thoughts are. They should STOP Focusing on making MISERABLE matches and go with the Business plan that has a P&L statement which reflects making money off of FAN stuff.

If they hype this game to the MAX. They can make everyone happy and make this game a triple A title that is what we want. This means SELLING Cosmetics.

Lifeguard Mercy would be a MASSIVE success. I could see them asking 1 dollar for just this cosmetic alone. In the blizzard store and making BANK. At the game’s height of success. There were over 30 million players world wide. So that means at LEAST 20 million BUCKS would be made. Literally. a massive success. and they don’t want 20 million dollars in their pockets?!

BCRFC mercy - set an "x amount percentage of proceeds would go to the foundation. Yearly.

1 Like
71

How can you say this, and then almost immediately say:

The 50% odds you describe are precisely and literally what handicapping is meant to accomplish. Making these odds is the explicit purpose of Match Making Rating, as explained by Principal Overwatch Designer Scott Mercer.

Colloquially speaking, individual Overwatch players are never ‘forced to lose;’ it is always possible for them to win. But more strictly and accurately speaking, wins and losses are always forced by Match Making Rating and algorithmic handicapping. The algorithms do not merely predict match odds, they determine match odds based on said predictions.

But that is effectively and exactly what happens in some cases with Performance-Based Skill Rating Adjustment (PBSR). The system can decide whether or not you gain/lose SR and it decides how much you gain/lose on a per match basis.

It’s all relative, and it depends what rank you are competing at. So why do you think matches are made based on MMR? Why aren’t they based on rank?

Absolutely WRONG, I have said nothing like this. Lazy, bad faith interpretation. Still I applaud your effort, much better than your peers on the wrong side of this debate.

2 Likes
72

This and even without PBMMR handicapping in terms of simply having highly varying skilled players skews the game up heavily in OW favoring those with the better more staple averages vs. those with higher ends and lower lows. Drag these ends enough and it cripples the teams heavily, not to mention makes the gameplay experience miserable for that team almost always

1 Like
73

It is not true that the algorithms determine the odds of the match. It is the relative skills of the players that determine the odds of the match. What you are suggesting when you say that the algorithms determine the match odds is that the algorithms are somehow deterministic. And this is an extraordinary claim to make. If you could somehow support this assertion it would be revolutionary in terms of our understanding about the underlying laws of physics which govern our reality.

Assuming, however, for the moment that you have not somehow discovered such a revolutionary truth that would put you in contention for the most impactful thinker in human history, it may be that you simply do not have a particularly full understanding of how the OW matchmaker operates and how Elo systems more generally operate.

In any Elo system, a prediction of the outcome of the match is made prior to the match being played. This prediction is then compared to the outcome of the match in order to refine our understanding of the relative skill levels of the players in the match. It’s not unlike a scientific theory in that it must both make predictions and test those predictions in order to demonstrate the validity of the theory.

What I find most interesting about your statement, however, is that it is (unintentionally I’m sure) predicated on the idea that the matchmaker has, in fact, accurately assessed the relative skills of the players in the match. That’s the whole point of predicting the outcome ahead of the match and then using that match to verify the prediction.

The only way for those predictions to be accurate (as you suggest they are) is for the matchmaker to be operating on an accurate assessment of the relative skills of the players in the match. The only question, then, is what sort of matches the matchmaker should make. And the reason 50/50 matches are preferable is because these are the matches most susceptible to having their outcome disrupted by the players in the match.

A 50/50 match, in fact, must have it’s outcome disrupted because some team will, in fact, win. And that is of benefit because it maximizes the chances that we get new data for future predictions.

In a 70/30 match, for instance, the most likely outcome is that the team favored to win actually wins. What new information have we gained in that case? No, far better to make a 50/50 match and maximize the chances that the players in that match will upset our current assessment of their relative skills.

We want to learn more about the player’s skills. That’s the whole point of having a competitive ladder and having players compete on that competitive ladder.

2 Likes
74

So if the game was balanced to your preference, without an algorithmic handicap, but you continued to lose 50% of your matches… what would your opinion be then?

1 Like
75

Possibly because everyone has made their point in your 3000 strong set of replies to your original post.

Now people are just amazed that despite the overwhelming evidence that you think it’s a bigger problem than almost everyone else does, you are continuing down this path.

What do you do for a living brother? I would honestly hire you for this single-minded determination alone if you have even a tiny bit of talent.

1 Like
76

It is ABSOLUTELY TRUE. Match Making Rating determines the odds of every match by deciding who plays in it and which team they play on. Please, try not to lie in your first sentence when posting.

This hypothetical is only interesting to you. Also, I don’t play Overwatch anymore.

1 Like
77

So if the game was balanced to your preference, without an algorithmic handicap, but you continued to lose 50% of your matches… what would your opinion be then?

78

Right, this is the answer I expected.

2 Likes
79

You are just assuming the answer because you don’t know me, and you happen to be wrong. Removing handicapping would benefit experienced and skilled players. Handicapping benefits inexperienced and unskilled players. I am a better-than-average Overwatch player, and on that basis I’m willing to be that I’m better than you anyway.

And if you want to get technically specific on the basis of that assumption, then we can say that removing handicapping would benefit me, as an above average player. I would win more matches. Removing handicapping would hurt you, as a below average player. You would lose more matches.

So there’s the answer to your stupid hypothetical.

1 Like
80

This is true and also exactly the way it should be. Inexperienced players need more help whereas more invested, skilled players are more likely to tolerate issues that arise from this.

Also, please refrain from waving your wingwang in public, its embarrassing.