You make good points however the problem is that PBSR doesn’t make up for the differences. Say Michael Jordan and Lebron James play with a bunch of high schoolers. Let’s say Jordan wins every time. Sure, we can all see that James was better than the high schoolers that he lost to, but he still lost to them so he’s going to be at a lower rank.
Now imagine if every single game has someone perfectly able to shut you down but everyone else on both teams isn’t anywhere near that good. So James will keep losing and Jordan will keep winning. Even though James is clearly better than the majority of the team he’s losing to.
That’s what it feels like to get stuck playing against the same smurf every match or a rotation of two similarly skilled players. Especially since neither Jordan nor James should be playing against high schoolers in the first place.
This is a very good illustration of exactly what I am talking about especially because it describes the kind of skill discrepancy that is rife in online games.
How could PBSR possibly compensate for the effect that MMR/handicapping has on each players skill rating, especially when players can only gain SR by winning and can only drop SR by losing? And why on earth should we trust PBSR to make these calculations, all of which depend on completely arbitrary performance metrics? Match results should be the only factor of ranking. Rank should be the only factor of matchmaking.
In a random-around-rank system with pbsr (SRD), teams will be thoroughly remixed.
This mixing is key as to why no one goes on repeat farm or gets stuck in a non-transitive loop, wintrading the same smurfs back and forth.
You will percolate through these random lobbies (selected around your current SR, not rigged around everyone’s mmr). This will get you to your rank optimally (where you’ll just go 50% winrate with small runs of winning and losing, and some acceptable amount of streaks and stomps that are natural for competition).
If you over-achieve each time, you’re definitely going to bias yourself upwards because you belong higher plain and simple. This is how people overcome mmr rigging and climb, except you’ll be able to do so more efficienly and again, without rigging and all the terrible psychology that goes with denying you agency and affectance over the match.
I don’t like pbsr, but pbsr 2.0 could be vastly simplified (less sensitive to exploits, more responsive). You tack it on to post-match rewards (that are already scaled by team spread), and limit it to some back-to-basics data for the match (no history). You crop anyone who massively over/unders and give them this additional small +/-. You use no-nonsense indicators like on-fire (interactions/sec?), k/d per resource. NOT stuff like dmg or healing per 10 (because we want match case by case). You rank the people in that match and offset the top 1-2 with more gain (or less loss) and vice-versa. You are not referencing rank averages or data histories which is key. It’s nothing complicated that ends up breaking, just a “best/worst performer” kind of reward to scale your agency and affectance above the random backdrop so you get to your deserved rank label sooner.
It can help when adequate mixing isn’t available like dead ladder or sparse/offhours times.
But these won’t be issues for quite some time after OW2. If we get a playerbase back (population) resets (compression+recalibration of the ladder ranks), paywalls to stop alt infestations, and lose the mmr rigging - we should be fine without pbsr. You could apply this ‘careful’ pbsr 2.0 to mid-tier ranks and then stop it for the tail ends (it should be removed from bronze-silver just as much as for masters-gm).
It’s all about the randomization. Just to reaffirm, these players would rarely encounter eachother, and even sometimes be on the same team. The kids would be remixed in with other kids most of the time. These guys would rank up and out sort of like how it’s done now, just with much less backpressure from the rigging. They would converge to their true rank labels without these fake smurf matches in bronze because the mmr pulls you to the side with/against the same trolls, refusing to let you back into a natural mixture of the rank.
PBSR will never be good enough, because it is always going to be artificial, arbitrary, and relativistic. When we talk about players overperforming and underperforming, we have to remember that it’s all relative to their peers in the match or at a given SR level. If you trained and paid a real life human being to spectate every match and make these types of judgements I would not trust them to do it, nor do I trust PBSR. But props to you for your open-mindedness
MMR and mixed groups are what necessitate long wait times in the first place. If Competitive Play was solo queue only and purely SR-based, matchmaking could be instantaneous.
50% win rate does mean by the system you are “ranked” accordingly; OTOH, the question remains if this system yo-yo’s you too much, and does it put forth fair challenges in the process. To be frank, the main problem with the system is it’s by account, not by player and thus is simply inaccurate alone for that reason as it’s well known hard-core players often have between 5-20 alt accounts, stacking tiers they play in with artificially high skill representation. Until the system actually starts to do things like implement SMS authentication per account coupled with SR decay and a nominal fee for ranked to dissuade alts, it will never be a true representation of where a player belongs no matter what the system is.
I have seen it for years now, loss streaks are often longer and more frequent than win streaks; that is simply unnatural and it can’t all be due to “tilting”.
I am highly suspicious of any ranking system that refuses to tell you how you are being ranked, and don’t get me started on that systems of adaptive acceleration they put into to help streamers on their “Road to GM” runs that allows alts to advance faster whilst simultaneously over lowering people in a few matches due to circumstances like leavers and Smurfs beyond their control…
There are a lot of very suspicious things with the way the matchmaker works on this point. In large part its due to not enough players to start; OTOH, all too often I get a good team, win, then my next team is against harder opponents and its as if the system took everyone who was getting better, slapped them on one team against people who have a longer history of “doing well” and they get stomped. It suggests some strange voodoo in how the system decides to test you…
THIS is exactly what I’m talking about. See how we all experience this phenomenon and describe it the same way. When I played Overwatch I had a large friends list and usually organized my own groups. I also vetted my friends list, so all of the players I networked with were quite good players and we grouped to play regularly. We were practically always were faced with other challenging groups, often ranked higher than us.
I remember it getting distinctly worse over time with each group as we played matches over time. This experience is consistent with the statement of Principal Overwatch Designer Scott Mercer, who said that Match Making Rating forms profiles on groups as well as individual players, and that it informs the Matchmaker to counteract group synergy.
Let me ask you both another question. Do your wait time for matches typically decrease or increase when you play with a good group?
Well, tbh I solo queue only now - which is even worse I suspect… I have even tried not using Priority Passes, that seems to have no impact though… I don’t believe this is necessarily an issue of the groups themselves exclusively but I don’t doubt they do some weird finagling in their numbers with large stacks too…
One other major issue is the MMR for a team is effectively just a sum of the players; it doesn’t balance per role and with stacks in mind on top; this is major problem because effectively the strongest duos are tank pairs by a large margin. I can predict who will win at the start of a match just by seeing that a pair of tanks on the other team are paired vs. two solos on mine, easily they win 2/3 of the time outright… One can argue this is “huuuurrrr teamwork” but its not when the system doesn’t account for tanks being on average the strongest role; you cannot balance that by having two tanks that are both 150 SR less and assume the difference is made up with say a single higher SR DPS and one in Support… it just doesn’t balance out…
I found a pattern with quickplay at least. If you average an unusually high e/d ratio and elims per 10 min, the matchmaker will essentially force loss streaks. That’s why I lose around 7 out of 10 matches on a usual basis, even when I average above 23 elims per 10 min on soldier with 4.31 E/D ratio.
I looked at a public profile of another player who had a much higher win rate and noticed that the person’s stats were much, much worse - none of the dps characters had more than 20 elims per 10 min.
Given that both that player and I had a lot of games played, I am surmising that this is essentially the handicap in action.
TBH, it’s a degenerate system conceived by half-brained devs. I mean, I really am curious as to what the average IQ of the dev team is.
How stupid do they have to be to think that people will keep playing unfair matches? Of course, people quit. And parity is just as important in QP as it is in comp; it’s essentially the “face” of overwatch. New players will get instantly turned off by the moronic matching system.
considering they probably have a minimum of a bachelor degree i would say on average they are reasonably intelligent, and perhaps even an expert within their designated field, therefore, one can assume that their IQ would be about 85 - 115 which is the normal range.
This is a very well-measured anecdote! You are describing the quintessential algorithmic handicapping experience.
Most players give developers the benefit of the doubt, by virtue of their profession. We fall prey to the ‘Appeal to Authority’ fallacy, believing that their ability design computer programs makes them experts on game theory by extension. But we are seeing, in the case of Match Making Rating/algorithmic handicapping that this is not true. Activison/Blizzard staff clearly did not think through the implications of their Matchmaker invention. Or they have been blinded to those implications by their profit motives.
Anyway, I applaud your confidence in your own perceptions.
On the contrary, algorithmic handicapping was ostensibly invented to favor new and relatively unskilled players. As much as I consider MMR to be an abomination, I think it is actually very effective at stoking players’ addiction to the product and retaining new and mediocre players who would quit in the face of a real competitive experience, without an artificially levelled playing field.
More than the intelligence quotient (IQ) of the devs, I would be interested in knowing the emotional quotient (EQ) of Activision/Blizzard executives. I am betting that it’s low, based on their psychotic disregard for the gamers and the integrity of this “ranked competition” that they are algorithmically handicapping. The gaming pubic needs an ethics commission to protect us from sociopaths like Scott Mercer, Jeff Kaplan and the like.
im not sure you fully understand how much work those types of individuals do. you and i may take for granted the 40 hour work week, those types of people work that amount within a few days, the question we should be asking, is why would anyone WANT to be in that position?
There’s the onion. The original system would probably work perfectly if no one ever boosted, cheated, or threw matches to do bronze-to-dbag streams. But reality invaded and forced a bunch of changes to try to catch people who were obviously and blatantly doing something bad. Subsequently the people doing bad things learned to dodge those rules. And here we are.
You act as if the executives are the ones doing all the work at game design companies. Do you realize they’re in charge of teams of other people who actually do most of the work? All the execs do is collect credit for other people’s work, and take home the biggest bags.
Game company executives are paid to do whatever the hell the publisher wants, not what their development teams want, what players want, or even they want to do themselves.
i encourage you to look into the personality traits that are required for fortune 500 ceo positions, the requirements needed to fill that position. you clearly have shown your ignorance with respect to how top tier business works. they are always works, they never stop. where you and i may be able to go home and not worry about anything till the next day, those individuals are constantly taking calls, responding to emails, keeping tabs on int’l relations, etc. all the while it pulls from their home life and the relationship they have with the ones they love.
you think its easy being a ceo of top tier company? you wouldnt last a week
