I was straight up throwing and lost 5 games in QP, and the immediate 6th is a free win even though I was throwing. So yeah there must be some kind of underlying match manipulation. Devs are definitely not telling truth of the matchmaker
2 Likes
Thatâs fairly obvious. No one wants to play a game in their chill time and get slapped about.
But, if youâre matching on âskillâ (or and MMR number), and every is in the same ball park⌠Youâre still going to get clapped from time to time.
They looked into that and found it was just general player stuff that caused that. Nothing that was doing.
Three things to note here:
- Sometimes lobbies are imbalanced for whatever reason- there could be one or more smurfs, one or more people inebriated or high, one or more people throwing, etc. So you winning 1 match out of 6 while throwing could have any number of causes.
- If you lost 5 matches in a row, you should have been placed in a lower rated lobby. It can be difficult to lose in a lower rated lobby even while throwing.
- You know you were throwing. How do you know someone on the other team wasnât throwing?
So your conclusion that there must have been some sort of underlying match manipulation doesnât actually follow. It might be interesting to consider, though, why that was the only explanation you could see for the small dataset you offered.
3 Likes
A lot of thoughts here! lets set aside some definitive facts.
- A matchmaker is a system, set to certain algorithms. This match maker has an algorithm because its not ârandomâ.
- Any proof thats outside player(s) outcomes of wins / losses would need so much to setup that a streamer or something could do and maybe have, thinking Jay3s putting a diamond in a GM lobby. - but even that seems like its not enough. So therefore I think any evidence outside of that is unattainable. Even the Number 1 ranked tank in the EU thinks its rigged and his âevidenceâ was not enough - so rank of the person telling the story doesnt matter.
So just as I subscribe to âyes I will lose to better playersâ, I think folks who say its not rigged HAVE to admit, there IS an algorithm - definitive steps for putting a match together.
Now, what I dont understand and what has yet been explained to me
- 50/50 over multiple ranks and multiple seasons. - how?
- I think this goes along the line of, how can I have terrible matches during a stretch of SR where I had amazing winstreaks (over multiple days) - only to hit a âbottomâ and have amazing games again.
- Back to #1 - Gavin and others consistently say they can set the difficulty of a persons experience. - How? If there is an algorithm, albeit rigged or âfairâ, working as intended or not, how can the developers express âSetting the difficultyâ?
Would love to hear your thoughts.
2 Likes
I love these âforced 50/50â conversations. They bring me genuine joy.
If you believe forced 50/50 is a thing, and you donât believe youâre so far gone that you canât see reason, hereâs what you can do. Go to a free and open source chess website called âLichessâ (I canât post the link, but itâs a â.orgâ), click âWatchâ, click âLichess TVâ, then find a player who has had an active profile for more than a year. You can click on their profile and look at the rating changes and entire game history. Guess what? Theyâll have climbed to a 50% win rate⌠in a game with NO TEAMMATES. If you look at their bullet, blitz, rapid ratings youâll notice massive wins and loss streaks. The shorter the time control, the more chaotic the games, and the more streaks there are.
Still believe there could be forced 50/50 matchmaking, even in a completely transparent system where you can see your opponentsâ ratings and game histories? Think the algorithm is pairing you with difficult opponents to force a loss to break your win streak? You can even set the rating range that the matchmaking pulls players from. You can see when someone has fewer than 25 games and is doing their âplacementsâ. You can see when your opponent randomly tried an opening they donât normally play or is on a loss streak and tilting. Master players also do speedruns (the equivalent of unranked to GMs) all the time. You can even see this or when a player is âsmurfingâ, because you can check back and see theyâre 300 points higher rated than you now or are throwing games by resigning after 2-3 moves.
If the game histories of the players you find on that site look like your game history in OW, except with more data public, thatâs your proof. If you ask a chess player if the algorithm is arbitrarily determining their rating, they are going to look at you like youâve completely lost it, because the exact same thing happens during in-person tournaments, where you get paired and reach a rating that gives you a 50% chance of winning/losing.
5 Likes
Okay. So, thereâs a lot to get into here, but the most important bit to understand up front is that if you and I were equally skilled at any given contest and we played against one another the expected outcome would be that you would win roughly 50% of our matches and I would win the other roughly 50%. Over a long enough period of time. It could easily happen, though, that there might be a stretch of matches where you won 3 or 4 or 5 in a row, or lost several in a row. Over dozens or hundreds of matches, though, we would each coalesce around a 50% winrate.
Now, letâs say that your skill improves. Your winrate might start to creep up a bit. But, suppose you then went looking for better competition, so you could continue to improve your skills. If you found a new opponent equal to your improved skill level, your winrate would start to drop back down to something like 50%.
In this way, one should expect to have both win and loss streaks over a shorter period of time, but still have a roughly 50% win rate over longer stretches like seasons- even if oneâs skill at the game changes. Because the matchmaker is trying to make matches between evenly skilled teams, most players should have a roughly 50% win rate, regardless of their skill at the game.
That can still look much different and much more chaotic when one takes smaller snapshots of 5-10 matches. This sort of chaos is normal in a system as complex as OW. The other thing to note here is that players at the very top of the ladder and at the very bottom of the ladder will have different experiences as the matchmaker cannot find equally skilled opponents.
One more point, the divisions between ranks are somewhat arbitrary and can be calibrated differently at various times. This means that the devs can make it easier or harder to reach a given skill division. Itâs important to note, however, that this does not change the actual ladder rankings. That is, if the devs decided tomorrow to make it much easier to rank to GM, they could do that. They could adjust the SR requirements such that everyone who is now Gold or above found themselves in GM over the course of the next season. But youâd still have the same ladder rankings- the people who are now Gold would find themselves at the very bottom of GM, and the people who are now Plat would be above them, and the people who are now Diamond would be above them, and so on. This sort of shift has happened periodically over the life of OW1 and OW2 (though, obviously, never to that degree) and it happens during rank resets. But you arenât going to suddenly get current Gold players above current Diamond players or anything like that. Itâs not like people are being held back by these choices- itâs just that the people above any given player on the ladder win matches that player would lose- thatâs why they end up higher on the ladder.
I think that addresses most of what you were asking. Anything else (explaining why individual matches can feel screwy like thereâs no way to win or no way to lose when the matchmaker is trying itâs best to make matches between evenly skilled teams) starts to involve things like player psychology and player behavior (smurfing, throwing, drinking, smoking, illness, tilting, etc) and the fact that OW is freaking complex and players can run into bad matchups, bad maps, bad comps, etc even if they are theoretically equally skilled.
4 Likes
This is all great. I might even take up the chess suggestion.
One question remains.
As a software developer, and even a little masters in algorithmic analysis, I have to credential myself even though it might mean nothing.
I always regards algorithms as two types, some naturally occurring like factorial growth in stems of leaves, and others engineered like the various types of array sorting.
Was the ow teams algorithm in a way engineered to meet a natural 50/50 or was it discovered and feels nice?
To me, they could have chosen an algorithm that consistently produced a 60/40, and thus in that alternate world the forum would be filled with discussions of forced 60/40. But they âsettledâ on 50/50?
Are there multiple 50/50s algorithms? Of course! And so this is the rub, itâs their game and they have a right to set their targets to whatever. My qualm isnât 50/50.
So I asked, is the algorithm they chose, was it a decision to stay with said algorithm or was it a naturally occurring?
And be careful, I think if you say natural, that could explain game to game occurrences and keep the overall picture. But natural doesnât explain the precision that so many players are so near 50/50.
If you say engineered, that certainly doesnât explain the game to game occurrences of bad matches.
So just like losers âqueueâ I think this whole thing was a play on words. Itâs not a âforcedâ 50/50, but encouraged 50/50.
An âencouragedâ 50/50 leaves room for game to game mischief but still captures career 50/50. I think there were plenty of âalgorithmsâ they could have chosen, and multitude of other decisions, so I do think there was some engineering.
So thanks for the discussion- Iâm now leaving with an âencouragedâ 50/50, not forced, not natural either.
2 Likes
i mean, my stats page tells a different story.
i was stuck high silver/low-gold for a while. practiced a lot. then my stats spike majorly. i was straight up dominating my matches.
then i climbed.
now my stats look like they used to and i am struggling to maintain rank.
long story short, git gud. when you are gud you will climb.
5 Likes
The 50/50 winrate is a consequence of matching evenly skilled teams. So the algorithm itself does not actually attempt to get anyone to a 50% win rate. It simply attempts to match even teams against one another. In OW1 this was done by averaging the skill ratings of the two teams (this led to some issues that one might anticipate- is it really finding evenly matched teams if team 1 is all Silver 3 players and team 2 has 1 Gold 5 player and a bunch of Silver 1 players). In OW2, however, the matchmaker attempts to find evenly skilled dps players on both teams and evenly skilled support players on both teams and evenly skilled tank players on both teams. To the extent that it is generally able to do so most players on the ladder will coalesce around a roughly 50% win rate. The least skilled players on the ladder, however, would expect to have a lower win rate and the most skilled players would expect to have a higher win rate (as the matchmaker simply cannot find enough players at those skill levels). Everyone should be in one of 3 states (roughly speaking):
- Players who are currently underrated by the matchmaker will have a win rate somewhat above 50% until the matchmaker understands their skill level better at which point they will start to hover around 50% (unless and until their skill changes).
- Players who are currently overrated by the matchmaker will have a winrate somewhat under 50% until the matchmaker better understands their skill level at which point their winrate will start to hover around 50% (unless and until their skill level changes).
- Players who are currently correctly rated by the matchmaker will have a winrate around 50% (unless and until their skill level changes).
But, again, the long term winrate of 50% isnât actually something the algorithm seeks out. Itâs just what happens when evenly matched opponents/teams play against one another across a large enough sample size.
3 Likes
You wonât, canât and shouldnât⌠But Iâd be curious to see these credentials. Nothing Iâve seen from your takes suggests this is your career path.
It feels very much like youâre trying to make your bad take seem better by pretending to have some background.
2 Likes
No they couldnât. This does not make any sense.
Have you ever heard the term âzero-sum gameâ?
3 Likes
So youâre doubling down, meaning itâs a lite algo that does a ânaturalâ tournament style. Meanwhile the devs themselves said thereâs so many knobs and tunements. And trust team avg vs role avg was something I recommended posted way before it was implemented.
A foil is easy to seek out because letâs say an algorithm pairs people with <50% win rate against people with >100 ping. And >50% with <100 ping opponents.
Iâm sure that could create and hover people around 50% win rate by giving artificial ping handicaps all the while matching people with similar sr / Mmr.
Point is, on paper you can ignore ping and map win rate and yadda yadda, voice comms - and just focus on a âtournament styleâ and say the algorithm is ânaturallyâ finding a position, but⌠if you start to factor in other things we donât know what they do, we donât know the weights they assign to variables and factors, ânaturallyâ keep people at 50% other then eventually they get to it.
(And they said they look at ping and other factors)
2 Likes
Seems pretty obvious how it ended up as 50/50. There are 10 players in each game and half of them will win and half of them will lose. Hence a 50% win rate. Would be pretty difficult to engineer a 60/40 split without a lot of players leaving OW forever after a loss.
Only at the very top and bottom of the distribution curve does it stray away from 50% win rate.
2 Likes
Major university in America. Undergrad, took the GRE got a good enough score to stay at the university to pursue a Masters - its rare, but kinda typical. I dont really have anything to prove other than I get a pay check every 2 weeks. So the right people in my life believe me.
I think this is a mesh of one team loses one team wins.
Overall, as a software engineer, it can be done. What would look differently maybe the distributions at any given time, maybe even worse match quality.
But to re-iterate, the âalgorithmâ they have I think nudges 50/50. Like someone already wrote, it helps with streaks. Like didnt they add something in to help prevent loss streaks? So again, âhowâ did they do that if the match maker doesnt look at actual streaks? Or does it now? Wrap the hows into a black box and just know 95% of players have close to 50/50 - so the âhowâ is working.
What I am trying to unravel is the âwhyâ is it working. Is it engineered or is it a ânaturalâ. Perhaps the way our geographics layout and ping networks are unnatural and this thing is filtering out the natural order of things? - but thats still engineering. So we cant separate one and ignore the other. Mean while people want to quickly defend and say EOMM doesnt exist when, perhaps the ânaturalâ 50/50 IS the EOMM itself, its what we expect, however in our ping network, group of fiends, the actual outcome would be 80% win rate for the kid that has the T1 line in his neighborhood.
Here is a story. I was the best street fighter player. Something clicked with me as a young kid. Also what helped was my highschool had marvel vs capcom or one of those Vs games. NO ONE GOT ME OFF THE STICKS. I had a 100% win rate. At the local arcade, I had a 100% win rate. And this was at a top 5 cities in America. I was playing more than 5 people.
THIS matchmaker would see this, and scour the earth to make sure I play âharder opponentsâ (or pair me with worse teammates), to give me 50/50 win rate - because THIS match maker wants me engaged. It would fear that I would stop playing if I had 100% winrate. See what I am saying? the 50/50 IS the engagement.
But to be fair, it cant predict the minutia of every game and every player, so it ânudgesâ 50/50.
Guys I found peace with this.
You can try to discredit my university or my back ground by what I type on an online forum dedicated to a video game, your just wasting energy.
1 Like
You can assume that there are all kinds of things happening that you have no evidence for. That is trivial. But once you start doing that, you have to ask yourself a few questions:
- Why am I assuming all of these things? That is, what is my goal here, and how does having that as my goal affect my critical thinking in this case?
- Why would Blizzard add all of these systems (and to be honest, Iâm not even sure what systems you are assuming they are adding at this point- you think they are selecting for ping in order to cause players to lose matches?) That is, why would Blizzard add systems that will make their game feel worse to play, thereby increasing churn, which decreases revenue? Again, when people who buy into the EOMM conspiracies discuss their theories, what they describe are the opposite of EOMM. Why would Blizzard do that?
- How has Blizzard managed to get everyone involved in this to not only lie about it all the time, but to lie about it consistently, and to prevent everyone involved from ever talking about what they are actually doing, as well as preventing everyone who investigates or talks about Blizzard from finding out?
Your theory is radically implausible on a number of fronts. Itâs also a self-serving theory. You, yourself, should be highly, highly skeptical of it. Indulging in radically implausible, self-serving theories creates terrible habits of thought which erode critical thinking.
Not only that, this particular theory only exists to convince people that they have less agency in the game than they actually do. That is, the entire reason that people have invented this conspiracy and continue to indulge it is so that they can tell themselves that it is impossible to meaningfully improve at the game- that the game is specifically hindering them from climbing the ladder and that it is specifically making their matches worse. But consider the implications of that for a minute. If I believe that the game is causing me to lose matches and I have less agency in my match outcomes, I am stymied from actually assessing my own contributions to my matches. And that means I improve slower, I have worse match outcomes, I contribute less in other players matches than I could have. This decreases not only my own satisfaction with the game, but also my skill and my impact on othersâ games. In short, indulging in this almost indescribably implausible, self-serving conspiracy makes the game worse for all of us. It exacerbates every issue that people think they are explaining by indulging it. (And it makes everyone who indulges it a worse critical thinker, which has implications outside the game as well.)
2 Likes
They just use ping to find low ping servers for players, thatâs all. We know they dont use ping as a factor for matchmaking itself.
Because they have said that matchmaking uses only mmr.
https://overwatch.blizzard.com/en-us/news/23910161/overwatch-2-developer-blog-explaining-matchmaker-goals-and-plans-part-2
Your current MMR is the only thing the matchmaker takes into consideration when forming your matches.
They also do not look into map winrates.
https://x.com/SrslyPaladin/status/1861868778277212377
Q: Do you take into account player performance on map types? As an arbitrary example, a Ball or Doomfist 1 trick will perform much better on push than on escort, which could contribute to some unexpected stomps?
Nope! we donât want to assume that players will play a certain hero. and we actually pick the players in the match before we pick the map
Youâre just confusing yourself for no reason.
Keep in mind, its impossible for them to list all the things they -do not do-. Because that would be an endless list. Iâm sure you understand that.
Anyone could come up with any kind of weird features that they have not explicitly denied.
So, what blizzard can realistically do, is say what it does. And they have.
The ping handicap altogether would be quite weird to implement because i have in every single game 30-40ms ping max.
Youâre free to elaborate how they would âhandicapâ me around that. Or anyone else with similar connections.
And if i play with vpn, i get a stable ~130ms ping to other region. Never randomly changing. Iâd expect that to happen if there was some handicapping happening.
Nor are the games any easier or harder with higher ping. Lobby ranges stay the same.
And that would also be something people would have picked up for sure over the last 9 years of playing, if higher ping affected somehow the lobbies you get or whatever.
It would not be a secret.
Yes, and then they removed it since it didnt actually do anything but prolong queue times. They also told what it did.
And you know what, they said when they implemented that change, and as it was removed.
It was not some hidden thing added to the game. Like what you keep suggesting they âdo other thingsâ, but have nothing to back up.
Now youâre just playing with words again. overwatch uses SBMM not eomm in their common meanings.
It just uses mmr like an elo like system should. as you win games, your mmr goes up which means you are paired with and against equally higher mmr players. And visa versa if you lose games.
It just seems you dont really understand how elo-like systems work.
Chess is the simplest example to grasp the concept as its a 1v1 game. Ow just does similar thing in a team setting.
It again does not try to find you âharder opponentsâ or âworse teammatesâ to keep you at a winrate. a 50% winrate of players is an emergent property of the system.
Games in the far past usually did not have any kind of matchmaking and players were just thrown together at random etc. Best players could easily pubstomp games, etc. Which was maybe enjoyable to them, especially if they were young.
Just like your streetfighter experience.
sbmm evens out the playing field as itâll find the skill level of players and put them against one another and theyâll end up in ~50% winrate. Because it is expected if you play against equally skilled opponent, youâd win and lose 50% of your games.
It really is pretty straightforward how it works.
If you are a grandmaster chess player, would you expect to be playing against other grandmaster chess players or not?
And if you do play against other grandmasters, youâre probably having closer to 50% winrate than if you were playing against novices.
Which system does make more sense to you? Put a grandmaster chess player against other grandmasters or just anyone at random?
How does it ânudgeâ 50/50, and why is it now ânudgingâ?
why do you keep changing the definition? What happened to ânaturalâ?
It would be helpful if you define what you mean with these words, as you seem to have your quite own peculiar concepts that you come up all the time.
1 Like
They actually took this system out, because they found that it did not help with streaks- all it did was increase queue times. But thatâs interesting too, because even when they tried to prevent players on loss streaks from losing- it didnât work. When players are on loss streaks, they tend to keep losing even if Blizzard doesnât want them too. Consider that for a moment.
Itâs not an engagement thing, though, itâs a competitive ranking thing. Consider, you StreetFighter king exist on a competitive ladder ranking all of North America- all 5 of those top cities. In each of those cities, there is an equivalent player with a 100% win rate within their city. If you are ranking those 5 players on the North American competitive ladder, who goes in rank 1, rank 2, rank 3, rank 4, and rank 5 if you never play each other? If you each only play the players who are significantly worse than you, such that you each maintain a 100% win rate, how can the competitive ladder ranking have any actual validity?
If you want to rank the entire ladder of competitive players, you honest-to-God just need the best to play the best and the worst to play the worst and everything in between. Thatâs how you accomplish accurate competitive rankings. And that results in the vast majority of players having a roughly 50% win rate.
Thereâs no reason to assume hidden systems and an impossibly vast conspiracy to hide those systems.
1 Like
Assuming true⌠makes your take on the game even more baffling.
You should quite literally know better.
2 Likes
except having a paycheck to maintain a janky react app centering divs, consuming APIs and whatnot does not give you credentials to talk authoritatively on ranking systems
2 Likes
No I think my masters in algorithmic analysis allows me to ask questions about it. I donât think anyone here can speak with authority unless they coded it.
The engagement factor is the promise of competitive ranking. You canât âcodeâ engagement, the pattern of wins and losses is the psychological effect to engage. Along with sound ques, music and skins.
The promise of the white paper is if people win too much or lose too much. Well this was their approach to solve that.
So them making this âtournamentâ style to the best of their abilities using wins losses is the psychological aspect of the match maker to engage people to play their games.
Everybody is too distracted by numbers if I put 50/50 55/45 or 60/40 - based on team balancer and multitudes of other factors they can âtryâ whatever win rate they want to go for.
Guys I made peace with this. No need to throw shots about my real life education lol, itâs kinda sad.
As I reflect back and think of the terrible stomps that are still happening (not to me Iâve moved on to Mr for this week), I think they need to find out whatâs wrong and how duds and suds are at the wrong rank - they may need to look into recent play time.