It’s not that I think it’s obviously lying, it’s that it would be instructive. We’d see what is actually happening in those cases.
Like, could a GM player lose in Plat- sure. There are lots of ways that could happen. Maybe they have 2 or more leavers on their team for an obvious example. But that doesn’t mean the matchmaker is broken. That is a very specific claim to make.
Can you do one continuous stream of 5 games each, starting with the lower account?
My guy if they are where they belong then they shouldn’t be climbing lmao
Maybe in OW2 they’ll just give everyone 100 SR per day they play so everyone goes up.
Sounds like socialist gameplay
There are a lot of games where progression isn’t really skill based. OW obviously isn’t one of them, but there are still a TON of threads about how hard it is to rank up. I really do believe that at the core of these complaints is the not-unreasonable assumption that one should rank up.
Which is just wrong. There is no progression in OW. Never was, never will be, never meant to be. It’s a game design, not a matchmaker error.
I could see that being valid in a game that doesn’t have a “Player vs. Player” competitive mode.
A tank player who’s been gold for 3 years and isn’t really able to process the game mentally at the speed of Diamond players do, isn’t going to perform in diamond off of golden game sense.
I don’t think people remember that they’re playing against other people.
I never really played any other competitive games because I hate them. I hate them because I get put into a match with people that I can’t see, can’t hear, and can’t get to because I die instantly. Honestly, how anyone got started playing these games is beyond me, but maybe if I had been introduced rather than finding them on my own I would have seen the appeal. Ahem…to the point:
People like me, who come in, play a short period of time, then leave are ESSENTIAL to the game if it uses only a W/L ratio to rank players. In this case, it actually becomes true that you can have a >50% win rate for all the concurrent players, because you just killed me 20 times, I killed no one, and then I left to play Fallout 4 where my poor reaction time doesn’t matter.
I’m guessing that you don’t hate these kinds of games. I’m incidentally curious how one becomes good at games where you die instantly and then sit in a lobby for 30 minutes, but that’s really beside the point. The point is, there are downsides of SBMM for players who aren’t horrible and who aren’t great. Which, to be fair, are most players.
If you’re not horrible, you can progress in a non-SBMM system, too. In SBMM, though, you really don’t and OW is specifically designed for people to not progress without significant changes to skill.
I know how convergence algorithms work. My point is that OW’s doesn’t. If you had clans, with regularly constant teams, it would work. It cannot possibly work with independent events (new teams every game).
At any rate, my win rate is consistently below 50%, however for it to hover near 50% I get games that are slaughters, one way or the other.
Last night I got put into a game where both team averages were 500 above my current SR. We won easily and convincingly. SR bump was 27. So even though I can play easily and win easily in a game with 11 other players who have an SR 500 above me, my SR is still 500 lower. On the face of it, that seems like an error.
It’s already socialist gameplay. Socialism is rewarded by the game and individualism is punished.
If you have a different view to the party line, or if you take advantage of a game loophole caused by an unexpected consequence, the game is changed to match the party line much like all mistakes caused by central planning.
No one can be where they belong if people who are better than them are climbing. In a fixed population of players they would be descending. I think this is why Blizz probably turns the algorithm off when you get below 500 SR. At that point it’s just RNG.
I think this is part of the problem. On the face of it, that seems like an error, to you. But it shouldn’t. You are looking at one game out of however many. And you are looking at a game with 11 other people in it. I should expect some weird games to happen.
Hell, I’ve played in at least a couple of games with Masters and GM players that my team won. That doesn’t mean I should be Masters or GM. (I should not.)
One of the reasons I bring up the GMAT is that most of the conversations surrounding the OW algorithm are conversations that I have had repeatedly when discussing the GMAT with my students. And this is one of those conversations, if a student gets one of the hardest questions in the question bank and answers it correctly, what sort of rank/score should they expect:
Where should they be placed on the ladder?
And the answer is that they could be placed at the top of the ladder. Or they could be placed at the bottom of the ladder (though this might be unlikely.) More than likely though, they will be placed somewhere in the middle of the ladder with the majority of test takers.
The fact that you were in a game around 500 SR higher than your then current SR, that your team won the game, and that you were bumped up 27 SR is all very reasonable. This goes back to some of the earlier conversation about how much the matchmaker is allowed to suppose that your skill deviates from your current rank.
A worse system could have immediately bumped you 500 SR, but there would be all sorts of knock-on effects from making that choice. (There would be wild swings in the ladder rankings for one thing- if you think people complain now…)
Instead, you got a more than average bump as the matchmaker learned that you could contribute to your team winning in a game that was 500 SR above your current rank. That 27 SR bump makes a sizeable contribution to a climb on the ladder. If you continue to win, the matchmaker continues to learn more about you and you continue to climb. But if you start losing at some point, the matchmaker also learns more about you- (that even though you won that one game, you lose more than you win above a certain SR.)
Either way, we want the ladder rankings to reflect this increase in knowledge.
But it’s all relative.
Why can’t people be where they belong if people who are better than them are climbing? That doesn’t even make sense unless you assume that the SR ratings are somehow a fixed measure of skill. They are not. Just as playing basketball at a professional level (or sprinting, or power lifting, or hell chess) means that one is playing at a higher level of skill currently than it did in say the 1950s, but you can still meaningfully compare current players to each other.
Bronze, Silver, Gold, Diamond, Masters, GM, etc are each relative descriptors of skill. They do not purport to be some fixed unchanging measure of skill. If someone plays at a GM level, but they do not improve and end up playing in a later season at a Masters level, that does not mean that they are no longer where they belong on the ladder.
It means that while they did play at a level that was good enough for GM during some number of seasons when compared to the other players on the ladder during those seasons, they are currently playing at a Masters level when compared to the current players on the ladder. In both cases, they were/are placed appropriately for their play at the time.
Also, I’m really curious what you mean by the algorithm being turned off below 500 SR. Are you just talking about the actual SR number being hidden below 500? Because that’s a very different thing than turning the algorithm off. Turning the algorithm off would prevent matches being made for players below 500 SR.
No, it’s unexpected. It’s half a rank of SR difference. Why would the matchmaker put me in a game with 11 other players of 500 SR greater than me unless it thought I was able to play at that rank? And if I am able to play at that rank, why am I not AT that rank?
The answer is simply that the algorithm doesn’t measure skill in any meaningful way at all. It is a system of reward and punishment, for behaviour not skill. And that’s OK with me. What’s not OK is pretending that it is some kind of skill ranking system, rather than a measure of a willingness to follow unspoken rules.
Because if they are climbing, then they are (a) Winning games and (b) Increasing their SR which means that the people they are winning against are (a) losing games and (b) decreasing their SR. While there are some neutral SR games, they are the very small minority. You are either going up or down, you can’t be doing both, so you cannot be “where you belong”, you are always moving one direction or another.
No, it means using randomly selected teams rather than some kind of matchmaking. People below 500 SR are either so bad that it’s completely irrelevant who you put on what team in any game, or they are there deliberately (yes, it’s easy to do), for the lulz. Either way, the outcome of any game is inconsequential.
Usually this happens for one of a couple of reasons. The first is that there are not enough players currently queueing to make a better match. That is, it would prefer to have all 12 players be within some fairly narrow range of skill, but if it doesn’t have those players it will look in a wider skill range. The second is someone in that match is grouping with someone of a different skill level. In that case, the matchmaker is essentially prevented form finding 12 players of the same skill level, because 2 or more players have chosen to create a match between players of different skill levels.
And, again, you should expect that outlier games exist. That’s normal. Unless the OW algorithm were doing something very weird in order to enforce particular results, players of lower skill level would sometimes win in matches against higher skilled players. That sort of thing happens. It just wouldn’t be the norm. If it were the norm for any given player, that player would rank up. But if it is not the norm for a given player, they will not rank up, they will simply sometimes have these outlier games where they win against better opponents. And in those cases, they will get rewarded with higher than normal SR gains.
That’s just a well designed system functioning as intended.
This is something you can easily verify for yourself. Watch some vods of games at a variety of ranks. Do Bronze players play as Plat or Diamond or Masters or GM or top 500 players do? When I watch a GM match, am I confused about what rank those players are? When I watch a Silver match, do I think to myself, this could be Diamond. Or this could be GM.
Why not, because the rankings do measure skill. If they were not measuring skill, what I would see instead would not be an increase in skill evinced by the players as they rank up the ladder. Instead, I would see GM play at all ranks and I would see Bronze play at all ranks. There would be some indistinguishable mish-mash of skill up and down the ladder. And, for all that there is different skill among players within any given rank, it is not true that people playing in Plat look like Bronze players or people playing in Diamond look like GM players.
You should not expect to keep some particular SR, no. That is, I am not a 2850 player, even if that is my current SR. I am a high-ish plat player, say. I should expect to hover somewhere around +/- 200 SR from some rank that indicates my current skill. If I rank down to 2650 or rank up to 3050 for a bit, that makes sense. And I’m probably roughly at the same skill when that happens. If I start to rank outside that range for a longer period of time, I am probably playing worse than I was when compared to the current player base. Or if I rank up to 3200 and achieve a new equilibrium in that range, I am now playing better (as compared to the current player base) than I was before.
They aren’t randomly selected teams though. I’ve never seen a <500 player in a match. Why is that? Has the random matchmaker simply never placed me in such a match in any of the matches I’ve played in over 2100 hours of playing this game? If <500 players are being matched against other <500 players, that is not random matchmaking.
Hmmmm, sounds like hand waving. When my Tank SR was 2000, my W/L was 47%. When my Tank SR was 1500, my W/L was 48%. When my Tank SR was 1000, my W/L was 45%. When my Tank SR was 750, my W/L was 43%. When I went below 500 SR, my W/L was 41%, until I climbed out (Not deliberately, that place is a hoot. It just became almost impossible to lose down there.)
If my “skill” is <500, how did I win 47% of my games against SR 2000 players? If my “skill” was <500, shouldn’t my W/L when I was playing in Gold be < 10%, only winning by pure luck? How does one win 47% of one’s games at a rank 1500 above where one “belongs”?
How did I win more games in Gold than I did in 750 bronze?
I’ll address the rest of your post tomorrow.
The more games you play, the tighter the variance… because you have an increased sample size.
He continues to say that if he wanted to do it right ( in 2017), he would MOVE THE VARIANCE ALONG WITH THE RATING. That makes sense as the sample size for the player within that NEW band (or variance) of MMR is close to zero. The cycle repeats again. the more game one plays at the new rating, the smaller the variance. There is no forcing of your rating within a band. You quote correctly but out of context.
I honestly don’t know without more information. Looking at vods would be helpful. But it’s really hard to say much of anything with this information. My best guess (and that’s all it is) given this limited information would be that your skill might be roughly in the silver range (assuming that your skill didn’t change at all over this time period- which is a pretty questionable assumption to make), so if some things break your way or you are playing particularly well you might range up to low gold. But if some things break against you or you are not playing your best, you might range down into bronze.
I should have been more clear. With a new account I place Tank in Gold, then grind my way down very slowly to ~700 bronze. With Support I place in Plat, then grind my way down very slowly to ~700 bronze. Occasionally, when the game crashes due to a memory leak, the algo burns my SR even lower.
I basically never climb. SR declines on loss games are always greater than SR increases on won games. That, combined with a consistent W/L of just under 50% means that my SR is always decreasing.
I get accusations of smurfing in lower bronze, which is hilarious, because I’m in the rank I belong in, right? I’m not alone either. Down in bronze there are a lot of Gold/Plat DPS who just don’t want to be in voice chat. They get accusations of smurfing too.
Yes you have, it’s just that they weren’t <500 SR when they were in your match. And if you have never played in Bronze, then you really don’t know what you are talking about.
It is within that cohort. In higher cohorts, the algo tries to match teams on multiple criteria. In sub-500, it just picks 12 players and throws them into a match. The difference between someone at 499 SR and 100 SR is barely noticeable. The difference between 2000 SR and 1600 SR can be game changing.