mitch1971
Member
Suck Big Melon Mammaries.First, what the hell is SBMM?
Suck Big Melon Mammaries.First, what the hell is SBMM?
My quick take from just glossing over the paper is that, at the very least, good matchmaking is not exactly computationally easy.I think there's a lot missing from what they have presented here. But it is nice to see them at least pretend to be honest.
... we can observe the percent difference in the number of players returning after14 days between the treatment and control groups. With deprioritized skill, returning playerrate was down significantly for 90% of players. The 10% of highest skilled players came backin increased numbers, but in aggregate, we see meaningfully fewer players coming back tothe game...
So, the take away is that favouring skill using SBMM allows for a higher player retention in aggregate. Deprioritising skill causes more mid-match quits, which they translate as "less fun", and triggers lower player retention in aggregate. Perhaps an interesting example of online bubbles versus the average casual with a 30 million sample size?... we observe that the quit rate significantly increases across 80% of players, and only the top 10% see a meaningful decrease in quit rates. We have historically found that quit rates have a strong negative correlation with self-reported “fun” gathered through player surveys...
Was my complaint earlier this SBMM swings too wildlyA fascinating read.
So, the take away is that favouring skill using SBMM allows for a higher player retention in aggregate. Deprioritising skill causes more mid-match quits, which they translate as "less fun", and triggers lower player retention in aggregate. Perhaps an interesting example of online bubbles versus the average casual with a 30 million sample size?
In any case, one item not addressed - at least that I could find - is addressing the wild swings that occur as the system evaluates players straddling skill brackets. Two wins throws you into an eSports championship, and you need to be curb stomped five or six times before the game finally lowers its evaluation of you and you can have fun again for a round or two. I'm curious to see their plans to address that anecdote, at it's by far the most repeated online.
There aren't many games with a greater track record of long term health than CoD. Kinda crazy to suggest they're prioritizing short term gain over long term health when they sell 20+ million copies every year.SBMM is about retention of the fickle masses to satisfy the algorithm and maximise monetisation in the short-term.
A fascinating read.
In any case, one item not addressed - at least that I could find - is addressing the wild swings that occur as the system evaluates players straddling skill brackets. Two wins throws you into an eSports championship, and you need to be curb stomped five or six times before the game finally lowers its evaluation of you and you can have fun again for a round or two. I'm curious to see their plans to address that anecdote, at it's by far the most repeated online.
This is the part that i think they have wrong(on page 3):Was my complaint earlier this SBMM swings too wildly
I have a few decent games and I am seeing iridescent cards on the oppo team
In no realm of any dimension should I be playing against top 250s in the world unless they have been reverse boosting which is also a glaring weakness of this system
Common sense. The only people who wanted SBMM turned off are weird high skill players who like dunking on little children.
They should at least give you the option to vote for a rematch or to stick with your winning team mates.I‘m ok with SBMM but I prefer the randomness of the lobbies in old COD games. You never knew what to expect from the other team and after a match, you weren’t put in another lobby.
Same lobbies were cool for two reasons. After a bunch of matches you know whos good or not and playstyles if you keep track. Also, the new match loads faster as the game doesn't have to find new players again.I‘m ok with SBMM but I prefer the randomness of the lobbies in old COD games. You never knew what to expect from the other team and after a match, you weren’t put in another lobby.
That's a fair point. It would seem the systems are quick to promote, but slow to demote. I suspect the idea is to bubble the best players into their top 10% quickly and then keep them there so they don't brutalise the rest of us into not wanting to play. The trade off is a few good games and the system thinks your ready for the world cup. It's worth noting they're clearly tuning the system to focus on player retention, and didn't present all of their data points - they might have tried slowing promotions or speeding demotions, only to find it counter-intuitively caused steeper player drop offs.This is the part that i think they have wrong(on page 3):
"4. It should be capable of adapting to a player’s ever-changing performance quickly."
I think It changes too quickly(if I'm reading it right), one good stomp and it thinks I'm an esports god. I needs to give you more matches in the bracket you currently are imo.
They don't acknowledge our complaints anywhere as far as i can tell which is strange as its a common complaint from the community.
Which is great in theory but this SBMM swings too wildly
You have a couple of decent games you know you are getting thrown into super sweat lobbies to intentionally get knocked down again
Rinse and repeat and you have CODs SBMM
I think Rocket Leagues SBMM system works much better gradually moving you in better or worse lobbies
I‘m ok with SBMM but I prefer the randomness of the lobbies in old COD games. You never knew what to expect from the other team and after a match, you weren’t put in another lobby.
(NO its not because im old!)
You never heard of S&M?First, what the hell is SBMM?
Ah that’s interesting. Yes, I’m talking about MW1+2 (the original ones). Didn’t know they had SBMM, and back then I didn’t have a clue what that actually means. I started noticing the term once people started to complain about it (which was way way later). Your explanation makes sense.By older COD you mean Call of Duty 4: Modern Warfare and above yeah? (cuz I think CoD2 had server browser and quickmatch)
Cuz those (COD4 and above) had SBMM the only real change is that lobbies stuck together, but when people quit the SBMM will evaluate the lobby before populating it.
Also the number of high skill players was lower in general, I remember in CoD4 and MW2 I could break into the top 50 world rankings easy work, literally sold multiple top 50 XBL accounts......now I dont think I would rank in the top 50,000. (NO its not because im old!)
I really wonder why they got rid of lobbies?
The new paper, a 25-page document titled "The Role of Skill In Matchmaking," is as impressive as it is jarring. On one hand, it's a little funny that Activision is treating the topic with tact and seriousness, posing such questions as "What is Skill?" and mapping "kills per minute" and "skill buckets" on graphs, when we all know that naysayers will probably just say "SBMM still trash lmao" and move on. But it's also a wild document for a videogame: I don't think we've ever seen a major studio go this in-depth on the intricate criteria and details that its algorithm uses to judge us.
MM in COD can also be just as busted as I just quit a game we were getting steam rolled and matches like this happen daily in CODNah, Activision had to do this in Overwatch too! I sware for like a week the game I were in had no kind of matchmaking. Playing in a lobby with someone way above the skill level of everyone else is no fun for anyone in the lobby. Its no fun in an easy win, and no lesson to be learned in an overwhelming force destroying your face. Playing in games like that did make me play less overwatch.
I believe they have some Kind of variable that is switched on which lags or slowS you downI've had this too. Sometimes I get shot from straight ahead on a corner and the person isn't even there at all. I feel as though they go in and out of a corner quickly where they get to see me based on their ping to host only and before my game updates with their new position they're back around the corner and not visible to me. I think it's too dependent on ping.
MM in COD can also be just as busted as I just quit a game we were getting steam rolled and matches like this happen daily in COD
This version of SBMM does not fix that as I have said before I am a very average player, maybe slightly above because the average player has a below .9 KD and last I check I was a little over 1 but I see great players in my lobbies often
Possibly they reverse boosted to get there to get clips for their social medias but even if thats the fact thats another problem with this system
I dont know when COD added SBMM, but those old 360 era games musta had zero SBMM. At least to me it seemed like zero. And people loved it back then.The results? Over 90% of players with toned down SBMM played less Call of Duty after the change, with the top 10% of skilled players not as affected. The paper goes on to confirm what has, to me, always been the most obvious reason to have SBMM: random matchmaking primarily benefits really good players, and disproportionately sucks for average-skilled players.