"The more certain the matchmaker is about your MMR, the less your MMR will change in either direction based on a win or loss.”
“Less” does not mean “doesn’t move much”. A million is less than two million, but a million is not a small number.
Comparing the movement of MMR in a single win or loss to movement of SR over many games is comparing apples to oranges. MMR is also expected to fluctuate widely over the course of many games.
Scott has said basically the same thing about SR, so your quote does not show any divergence between MMR and SR.
I don’t understand why you all have regressed back to speculating on the details of how it works. It’s like playing “Guess what’s in my pocket”. It has little to do with whether the system is working as intended, and what the intentions are anyways.
It was an example how after just 10 games it could quickly go out of whack. The assumption against this is that SR will converge on MMR “very quickly” which would need more than 10 games to do so.
That is also only 1 player in a team, and 1 of 12 in a match. If MMR and SR can diverge that quickly, and take just as long to converge, then there is no predictability that the players in a team even with the same SR are within a few hundred ESR/MMR of each other. This is where the handicapping argument starts, but not going into that part of it again as I’m just showing that MMR is equivalent to SR is impossible.
What you mean to say is “I don’t believe you” not “that doesn’t happen.” Unless you are willing to take the time (was about 40 hours of messing about in seasons 5 or 6, can’t remember which anymore, not including the leveling up 3 accounts to 25 so they could play comp) then my word is actually massively better than yours. You won’t believe me because if what I said is real, then it blows a lot of your arguments out of the water.
You clearly don’t get it, because you still think I’m arguing at all about climbing. I can’t make it any clearer that I’m not. Also, everything about how you talk about the balancing/handicapping of games only works if there is a small population (so really you are talking about the extremes) while I’m talking about how the MM worked at the mid range.
Your need for every single quote makes it hard to discuss (and is detrimental as it frequently takes aspects out of context). You know there are quotes that say MMR is able to stay still on a win or a loss, and quotes talking about how SR always moves. They have talked about how watching MMR would be a horrible experience, and that SR is better psychologically.
Which brings me to this comment:
While I would love to see the source on that comment, I want to talk about the psychological aspect.
If MMR fluctuates widely over the course of many games (you are implying just like SR) then why would they claim it isn’t as nice a number or horrible to watch? So from logic they can’t move in the same manner.
If it can jump greater distances than SR, then wouldn’t seeing it go high after a win be rewarding? Knowing that it fluctuates a lot and that each game you play can impact it heavily wouldn’t be horrible to watch. It could feel random at first but people would work it out. Also this contradicts the statement that it will become “more sure” over time.
If it moves less than SR, then it would be horrible to watch. Imagine if every win or loss only moved 3 to 5 SR. Many people would quit as you would need to play some 600+ games at 60% winrate to even climb a rank. This is the most likely way it works when combining all of the statements about how it moves.
Scott’s quote is about grouping and SR changes, and that context does make a difference, and in this situation the range he is discussing is already somewhat known from complaints of wins only gaining 5 to 15 SR and losses taking 30 to 40 SR. He’s trying to sway the unreasonable fear of grouping up.
You seem to need a core aspect to fight against while I’m simply pulling down parts of your theory. This is why you can’t get it, and if you keep thinking the way you do you will never get it. It then follows if core aspects of your idea on how handicapping is working in the game then how you think it is balancing games is wrong - I just don’t care for that argument as it’s pointless.
Now, let’s look at a context aspect from my posts - I said I’m happy where I am in Plat because the games are mostly fun for me here - I can play any character (except Ana, I seriously suck at her) and have an impact and have fun. This means that once people get where they are meant to be the system seems to work.
But there is the crux - once people get where they are meant to be
The path to getting where you belong is longer than it needs to be, and the process is frustrating.
The matchmaker as it stands has a lot of inbuilt frustration that could be fixed by removing MMR, or at least creating an option for a player to reset their MMR and SR after 3 seasons with 50 games played in each season. At no point have I been thankful for MMR on any of my accounts, or found that it lead to a better experience.
The source is that MMR is closely linked to SR. And SR is known to move widely. You won’t get away from that blue post. Also, see “MMR works very similarly to SR” below.
Because occasionally MMR has a unappetizing move. Which people would come to the forums and whine about. And it probably moves a bit slower (my data suggests 19 MMR per game, compared to 24 SR per game, but these numbers adjust to keep SR from getting too far away from MMR)
This is the post that you referring to:
MMR works very similarly to SR. There are some minor differences that make it feel worse though, when you just watch that number. For example, it’s possible to win a match and not gain any MMR. We make it so that if you win a match, you always gain SR – even if it’s just a little bit – to feel psychologically rewarding. But MMR’s entire goal is creating fair matches – which isn’t always fun to look at and certainly not “rewarding” for players looking for pats on the back or a sense of progression. So SR “chases” your MMR very closely, except in a rare case of severe SR decay at GM/Masters/Diamond level of play. - Jeff Kaplan Overwatch Forums
Again, it says that MMR works very similarly to SR and SR chases MMR very closely. So, even though there are some small “not rewarding” differences, it is not enough to generate the effects that you describe.
Anyways, I’m exiting again. I just came in to ask for you to back up
But as I knew, there is no such support, and the references you have pulled up don’t do the job.
You speak to my concerns as well. There are online games with smaller communities, where everyone knows each other’s skill levels. In those communities, players usually volunteer to handicap the teams – the best players spread themselves across teams. That makes for a two-sided match.
But Overwatch has a huge community and nobody knows each other. When Match Making Rating is hidden, the best players get no recognition for bearing the handicap; for carrying their teams. That cognitive dissonance makes the community toxic.
I am also really tired of speculating about how Match Making Rating works. I know enough to object to it on principal (see the Scott Mercer quotation in the OP). But users should have complete information about handicapping in the game’s terms of use, so they can decide whether or not to waste their time in this falsely advertised “competitive game.”
Blizzard makes games for their own financial profit, at the vast expense of their users’ time and health (see World of Warcraft). I wouldn’t say they make games for any person, just the almighty dollar.
I feel you like to use the start and beginning of quotes but always leave out the meat filling. I’m just going to use this one to show some issues with the mmr moves heaps assumption.
So it is similar, but also not the same. Like how a car and a bicycle are similar but not the same.
MMR moves differently. It can move not at all. This is important for the next logic step to remember.
These words all come to the focus point - progression.
What is the opposite of progression? Stagnation. Staying the same.
So mmr can both stay still, and lacks progression.
SR chases mmr. But mmr can do 0 movement and stagnates. So to reconcile these points movement of mmr must be much less than sr.
Basically the exact opposite of your interpretation is there.
I can’t really come back for a few weeks. Work is sporadic so I can only post during quiet times. Assuming you are staying out of this for now too I’ll recommend people reread your last post so they can draw their own conclusions.
I said it long time ago it is a communist system to make everyone “have fun”. So it matches noobs with veterans.
It is a business to keep the majority happy so they can buy lootboxes.
Why shouldnt bad players have 35% winrate and stay bronze? Why are they forced to have 50% winrate?
Lol removed premades from solo queue. Also in Lol it seemed a steady grind up as you improved. I never saw plats group with silvers in a gold average sr games.
Why the fornication do you allow players with 800 sr difference to group togheter?
This system caters to emotions and “leave no man behind” mentality bs.
It is a game people it should be a challenge and not a binary system where the flip of coin is decided before the game starts.
This system is terrible.
I totally agree with this AND your sentiment that there should be a solo-only mode. (As an aside I think there could just be a solo-only SR while they keep dynamic queueing, but that’s a topic for another discussion.) This isn’t really what the thread is about, though. If it is, the OP should cut down the size of the post and get right to the point, which most of us would agree with.
As for the first bit of your post, this is actually what this thread is about and FOR ONCE someone has finally understood! Thank you!
You’re going to get a Table of Contents here to make it easier to follow:
Rank Deflation
Useless/frustrating matches
Inaccuracy at the margins
Summary
(1) So my first concern is that it would lead to rank deflation. I mean that everyone below a 50% win rate would leave. Since OVERALL there mathematically MUST be a 50% win rate (every game has one winner and one loser) then the higher ranked people will eventually start getting harder matches (more often) and their win rate will lower. There will be, of course, some low ranked Very Serious Players that are willing to put up with being stomped over and over to learn. I don’t think you would share this concern, but I would and the developers seem to as well so I bring it up. I think, based on your above post only, that you’d be ok with comp mode being comprised of only good and/or Very Serious Players.
(2) The “WinRate Ranking” will actually lead to more of the problems that people dislike about the current MM. Let’s assume for the moment that everyone is forced to play comp to avoid the first concern. The average player will have a 50% win rate, right? If half are better than you and half are worse, you will win half the time.
Consider how they will get to that win rate though. Some games they will have Bronze teammates. Some games they will have GM teammates. They are average because they will have no effect on the game. Let me explain further.
There are 2 types of players that are the bane of competitive ranking existence, Smurfs and Throwers. They mess with the system because it turns our productive matches into unproductive ones. We don’t actually know which team is better if there is a Smurf or Thrower on either team.
But if you allow all the ranks to mix it up in games, from the perspective of the average player, Smurfs and Throwers will be the rule rather than the exception. Sure, they’re no longer technically Smurfs (they could be technically Throwers, I guess, but there will be no presumption of skill in your match), but the end result, that there is a person much better than you on the enemy team while you are stuck with a bunch of overripe potatoes is still the same. You have zero chance to win this game. Some games will be impossible to lose. It will all be up to completely random chance.
We know this will be frustrating because it’s one of the criticisms of the MM right now! However, right now the frustration exists due to a combination of bad actors, inconsistency in play, and accuracy of the MMR. In the alternative version, this will be part of the design.
The “productive” games, the ones in which you have an effect on the outcome and are able to show your skill, will be rare less frequent. What will they look like? They’ll be games without much higher or much lower ranked players. They’ll look like matches that the current MM attempts to make. The further away you are from average, the more rare these games will be.
(3) Even more to the point, if you are a rare good player, Diamond or above, you will ALMOST NEVER meet your equal in competition. You will have no way of directly comparing yourself to your competition. And you can’t start matching people based on win rates, because matching people of equal skill will result in a 50% win rate, so we’re back to square one. I’m not sure what the OP means as “unproductive” but I think playing games over and over without actually having competition is unproductive. I think you would agree. The person with the slightly higher win-rate may have just gotten lucky with more Bronze opponents, you’ll never know.
(4) Don’t get me wrong, “WinRate Ranking” is actually a perfectly legitimate way of ranking. It could be done like this. You’d have to force people to play it to get any kind of accuracy out of the system and you would never get any accuracy at the extremes of skill. For most of the population the games would feel completely random. This would be true by design rather than a byproduct of the limits of what any MM can do and the interference of bad actors.
The ONLY benefit of such a system is that it’s more intuitively easier to understand, but the current system can be explained. For an explanation of how “Skill Ranking” is actually productive, i.e. how it is supposed to work, go to this post here:
I haven’t seen Mercer say that anywhere, and I don’t think it’s true. The purpose of Match Making Rating is not to predetermine the win/lose result of a match for individual players – that’s impossible. Its purpose is to give every player a 50% chance of winning. It does this by distributing the most skilled players evenly across teams.
My argument is that this form of handicapping is wrong, because it corrupts the win/lose results of games as a measure of player skill. Yet that is what the Skill Rating system is based on: winning and losing. It is the double standard that makes Competitive Play a complete waste of time. By hiding Match Making Rating, Blizzard is being patently dishonest with Overwatch players. We assume that our Skill Rating is based on our ability to win matches in an impartial system, but that is not true.
Maybe don’t attribute moral failing to a technical method of measurement that you simply fail to understand.
Whether MMR is hidden or not doesn’t really matter. Whether skills are spread evenly across teams doesn’t really matter.
There are other ways of measuring skill than counting up win ratios against random opponents, which would be the result of removing the Rating they use for Match Making. Much better ways. Ways that have been used for decades in a variety of sports and games.
MMR is one of those ways and it works much better than your alternative, if you even have one.
But that is exactly what the Match Making Rating system is designed to ensure, so you can’t say that. Blizzard has done a lot of careful engineering to make sure player skill is spread evenly, on a per-match basis. My argument is that by doing this, MMR ruins the accuracy of the Skill Rating system which relies on win/lose results to rank players. By hiding MMR from players, Blizzard not only denies players recognition from their peers, it keeps them from recognizing their own personal handicap in competitive play. That kind of deception is a violation of players’ rights.
I mean in your last thread i linked you to the article, here we go again. As stated before read between the lines
Skill ratings go up when we win a game and play well during it. On top of that, Overwatch is constantly making predictions about how well we’ll play compared to an enemy team and to a chosen hero’s average player stats—and we receive more SR when we exceed expectations. (Contrary to popular belief, the amount of time a player spends “on fire” isn’t a direct influencer as much as it’s a reflection of how well they’re doing). Also, SR isn’t the only measure for skill. There’s also an invisible stat called matchmaking rating (MMR), which only Overwatch developers see. Essentially, SR reflects MMR, but with rose-colored glasses. So, if you’re playing on a super-powerful team against a not-very-good team, and you win, you’ll gain at least the minimum amount of SR, but not necessarily much MMR. That way, you’ll feel rewarded for winning, even though the match wasn’t fair.
