Behavior Admits MMR Is Not An Effective System – Dead by Daylight



Read more Dead by Daylight ➜ https://deadbydaylight.mgn.gg

http://www.twitch.tv/ScottJund
https://discord.gg/TheBreaderhood

source

46 thoughts on “Behavior Admits MMR Is Not An Effective System – Dead by Daylight”

  1. Dude my games are all over the place! The other day there were 3 red ranks and me a bronze vs an ash nurse! Needless to say she got annihilated! And then the exact opposite happend we were all bronze ranks vs a red rank nurse she destroyed our ass no lube!

    Reply
  2. I can’t even rank up anymore. I’m a rank 10 player, but with the new mmr I’m stuck in ash 2…i wish i was joking. My first game after taking a two week break from the new mmr release ME an rank Ash 2 player (rank 18) was on a team with a 1, 4, 6 and the killer was rank 4. What am I doing going against all these good players and it happens constantly to the point where any time I get progress and I am about to rank up, I get 2 horrible games that take all that progress away. It’s so upsetting and frustrating cause with all of this happening I’m given no chance to improve in the game. I started 2 months ago so I’m pretty new and was doing pretty well, learning how to counter killers, just being a better survivor in general but then this whole thing messed it up to wear I’m a mid player going against rank 1’s all the time. It sucks cause I can’t even play for fun anymore since I’m literally fighting to crawl out of the slums it feels like.

    Reply
  3. So it's not surprising solo survivors will not go high with MMR and probable get stuck with potato teammates. That is my experience so far with this new MMR. Scott may think it is like the previous one, but I think its worse. The skill disparity of players in one match (survivors and killers) is too much compared to the previous one.

    Reply
  4. Yeah high mmr killers have ridiculous queues, luckily I don't deal with it but you can't stay at high mmr as survivor cause you're gonna have multiple games where you just get tunnelled out at 5 gens

    Reply
  5. They litterally already have a system for tracking all that shit, it was the BP and emblem system… why did they need to mess with a system that worked better than it did now? It wasn't good but it was a hell of a lot better than this.

    Reply
  6. I do blame them for not doing it, because they claimed they were doing it. For two years. All they had to do was nothing and the game would be in a better state right now. Now we're stuck with an even more flawed matchmaking system than we had before.

    Reply
  7. Yeah, it would absolutely take a lot of time, effort, and manpower to create a nuanced, functional mmr system for dbd. It might, for example, take a small team of devs 1-2 years of work *looks directly at camera*

    Reply
  8. Scott, with all respect (and I do mean that, I respect you tremendously and I love your content), you're misinterpreting what he means in that clip. What Patrick is saying isn't that the system is ineffective, he's saying that the system would be ineffective if it attempted to factor in all that nuance. And he's right, because MMR is NOT intended to be a progression system or a badge of honor. MMR's sole purpose is to decide what combination of players would make for a "fair" match.

    Let's take your example of the no-Addons Clown who finds himself faced against a team with 4 DS Unbreakables. You're saying that when the Clown only gets 1 kill, he should lose less MMR and the DS Unbreakable team should gain less as well. But…no you shouldn't. That's equivalent to saying you want more no-Addon Clowns to meet more 4-man DS Unbreakable teams, but that's exactly the scenario a matchmaking system is trying to avoid. That scenario is patently unfair, the Clown literally has no chance against them. The only time you want that Clown to meet a team like that on a regular basis is if that Clown is so ridiculously godlike at the game that he can beat teams like that frequently.

    Or let's take our ur-example: you run the killer for the whole match while your teammates do the gens, you're everywhere doing everything, but then the killer kills you at the end and hooks you. It feels unfair that you've lost MMR here, and I agree that feels bad, but let's stop for a moment and broaden our perspective. Yes, you've lost let's say 100 MMR. Let's say the killer loses 100 MMR for each survivor who escapes and gains 100 for each he kills. In this scenario, then, the killer has killed one survivor (you) but three survivors have escaped. So the killer has gained 100 MMR from killing you, but lost 300 MMR from not killing your friends. So the net effect of this match is that you lost 100 MMR, but the killer lost 200. Relative to this killer, you've actually gained 100 MMR, so even if this was what happened every time you met someone like this, you would eventually meet stronger and stronger killers because your MMR is actually increasing relative to the MMRs of those killers. In practice, however, you'd actually outstrip such killers vastly more quickly because there's only a 1 in 4 chance that you're the unlucky one who gets singled out. So if this same scenario happens 4 times in a row, on average you would gain, relative to those killers, a difference of 1300 MMR. I know that sounds counterintuitive, but statistically it does actually math out.

    Maybe the problem is the word "skill." By calling it "Skill-Based Matchmaking," it gives the impression that the goal is to match players of equal "skill," but this is misleading. What they really mean is that they want to match people of similar *power*, with power being a function of both intrinsic skill and extrinsic loadout strength. Average that across multiple games, and you get a value that approximates what a player's habits are on a particular Killer. A Killer who runs Pinky Finger Clown all the time should have a higher Elo than a Clown of equal skill but usually runs no add-ons.

    Does that mean I think they implemented this MMR system well? No. They did screw up royally in two major ways. First is the MMR cap, they set it WAY too low. Now, in the long term, this might even out. If you consider that in a given match, a killer can gain or lose up to 4x as much MMR as Survivors, you'd expect the best and most sweaty killers to shoot to the top way faster than the best and most sweaty survivors. Eventually, those survivors will chew their way up there, but, until they do, match quality is going to be pretty shite. I'd explain why, but this is already a giant wall of text, so I'll stick with saying "because math."

    The second problem is more insidious, and I'm not sure this one is even solvable. The problem is that Killer and Survivor aren't balanced. See, you're right that DBD is unique compared to other games, but it's not the RNG or the Add-Ons or the perks. Those average out enough to be a non-issue. The problem is that you have two populations: Killers and Survivors. This means that the MMR distribution for the two sides can be WAY different. If Killers are, on average, stronger than Survivors, then all the Killers are going to have higher MMR on average than the Survivors. Worse, you can wind up with populations that aren't centered bell curves, but scrunched up and transposed left or right. This FUCKS matchmaking, because the matchmaker now has absolutely has nothing to go on and it just starts throwing people into really random lobbies. In fact, it can even encourage the game to match the worst Killers with the best Survivors, depending on how the matchmaker is designed. Tofu mentioned in the podcast about how MMR can't work if it isn't zero-sum, and this is why. Because for each individual population, MMR isn't a zero-sum game, you can wind up with one or the other inflating their MMR dramatically. The only way they can fix this is by somehow balancing the game so that Survivors and Killers are almost perfectly balanced with each other across all skill levels, which…yeah. I don't see that ever happening. And THAT is what's really wrong with DbD using MMR.

    Reply
  9. For survivor, they also said that you get a higher MMR rating if you are the last to die. Their rating not only uses escapes as a point of reference to determine your MMR "score" for the match.

    Reply
  10. Wow… this is the second time Patrick has used the "Math is too hard" excuse. I believe the last time was when they were doing hit validation. Sometimes I think it's all a ruse so they can just buff survivors to hell and not answer for it. I mean, behavior (no pun intended) like this is common when people run out of ideas, but don't want to let go of their success. They fudge it as long as they can, and cater to whichever thing generates the most money, i.e. survivors. More generally, they're trying to succeed by avoiding failure, which never works out; they don't have faith in themselves. Hence all the non answers throughout the entire Q&A. DBD is on it's way out.

    Reply
  11. Escapes would in fact be tied to how good you are, if they considered the whole team and not just you.

    If 3 survivors in your games consistently escape, you're good.

    I don't care if you loop for 7 hours and do 10 generators. If all 4 of you die, you lost.
    I also don't care if you hid in a corner for 8 hours, watched your friends die, and got a lucky hatch escape. 3 of your team died, you lost.
    Flip that, you make some sort of ballsy play against a Bubbaface and trade in such a way that you die but 3 of your buddies escape? Nice play, you guys won.

    The root problem of this game is that it places you on a team but doesn't score you as a team. That will always, ALWAYS, cause frustration unless you only ever play with a 4 man SWF.

    Reply
  12. Before reset I was Iridescent 3, I don't have time to play a ton so I'm slowly climbing back 1 game at a time. I'm at Gold 3 right now. I played a match and destroyed 4 Ash 1 survivors, cool, ranked up a little bit, but my next match was against 4 Iridescent 1 survivors each with 1,000+ hours (I have about 200). I got all 4 of them to death hook, the game took a long time, I out mind gamed all 4 of them a few times, and I ended up getting 1 kill…. and I lost a rank point. I think the game also needs to take into consideration the skill gap between the opposing teams, because for how well I played against 4 Iridescent 1 survivors at gold 3, I think I should've at least not lost a rank point for that match.

    Reply
  13. I have a concern with a ratio'd system based on builds. With perks like DS and BT and Unbreakable, they change the trial in ways that help all other survivors in some way shape or form. Someone using DS could make a chase last 5 minutes longer and give survivors all the time in the world to do objectives, but only that single survivor will suffer from a lower progress ratio. Now if it was calculated to change the MMR ratio of your teamates, then you could just run a crutch build with tremendously degenerative affects towards all other survivors and prevent them from being scored fairly regardless of THEIR build. I think just adding a skill cap is the best option. That way, it is possible to get a maximum amount of "skill"(for lack of a better term) per trial. That way, using godlike perks simply makes it easier to reach that ceiling and maybe go just a little further if you do exceptional. Im not sure if im making much sense or if I'm just describing the last ranking system but i feel like having a semi-flexible ceiling per trial is an alright way to judge the broken placement of the game.

    Reply
  14. If the new mmr system is the exact same as the old one, how come my games have increased in QoL dramatically? I get way less sweaty killers and the survivor skill varies by a lot. Which is fine because Im not looking for a competitive experience, just a fun match.

    Reply
  15. Another reason that I can't wait for VHS. These developers do not give a single F about their game, nor do they know how to balance it accordingly. The Killers should be the powerful side, because they are the big Evil horror that survivors are trying to escape from. I said it once, and I'll say it again, F BHVR and their Rank 20 Devs. Can't wait for the day these A holes get dethroned when a better and more balanced game comes out.

    Reply
  16. Hate how this guy is so cocky and demeaning to DbD’s audience while explaining how he didn’t do his job for a whole year. Like bro you’ve got a product that people use in their free time to unwind. We’ve got every right to complain if the product sucks

    Reply
  17. i dont care about mmr at all, im just playing whatever game gives me. if killer is camping hook, well… what can i do against it? im playing same games as before, with same killers, nothing changed. it never will. dbd has lack of in-game rules and because of that calculating skill in any form is impossible.

    Reply
  18. For the 100% trial chase, what if a bug appeared not recording the chase? MMR would fail and then multiply that by other MMR unhooks, gen time, healing time, etc failing with a bug then MMR fails more frequently and is obvious there are bugs in the system. Escape and time played are two easy values to record to avoid people complaining about the bugs. Now imagine a game breaking bug where the killer couldn't hear a slugged survivor, how long has that been in the game?

    Reply
  19. The emblem system tracks so much more than SBMM does:

    Lightbringer: Generator repair progress, generators completed, exit gates opened, dull/hex totems cleansed, being chased while other survivors are repairing/cleansing.
    Unbroken: Time alive, being put into the dying state.
    Benevolent: Being hit, healing, healing another survivor, being hooked, unhooking someone, safe unhooks, protection hits, helping another survivor wiggle free (i.e; pallet/flashlight saves), hook sabotages – all with modifiers for basement and after powering the exit gates.
    Evader: Being in the killer's terror radius, chase duration, escaping a chase (with modifiers for chase duration), getting hit, stunning a killer.

    Gatekeeper: Time generators remain unrepaired, exit gates being opened.
    Devout: Sacrifices, Moris, bleedouts, disconnects, hooks, whether or not the killer hooks all 4 survivors.
    Malicious: Hits, hook states, survivors healed, survivors escaping your grasp.
    Chaser: Finding a survivor and initiating chase, losing the survivor, chase duration, hitting a survivor, proximity to a hooked survivor.

    Surely they could have taken some of this data, assigned appropriate numerical values to each, and calculated MMR from that? Hell, they could have kept using the emblem system but uncapped and it would have been better than this kills vs escapes atrocity.

    Reply
  20. I imagine if they implemented Scott's admittedly logical system there would be SOOOO MUCH debate about each and every number or coefficient or whatever. Numbers would be debated a lot I imagine, just like everything is already.

    Reply
  21. Imo old system still was better. As someone who is pretty decent at the game if I say it myself, I barely play it anymore because of mmr. At least the old system rewarded you a bit for repairing gens, doing chases and helping others. I've had so many games where we had 1/2 teammates that did absolutely nothing the entire game, let the other survivors die on second hook because they were too afraid to unhook them, found hatch and escaped. Before mmr, this was punished pretty heavily, especially in higher ranks, you would at max neutral pip but probably depip because your only badge is iri unbroken. But now, they reward those players like: hey, you did absolutely nothing, but that doesn't matter because you escaped! Your teammates on the other hand were carrying your * off, looped the killer for 4 minutes each, did 4 gens and unhooked and helped eachother constantly (probably if you did 1 gen yourself, looped the killer for about a minute and helped some survivors you would all have escaped, but that didn't happen) and they all died, so you know what, your teammates are bad at the game and deserve a loss of mmr while you my friend, will get a higher mmr!

    Reply
  22. genuine question for Scott: if BHVR hypothetically offered you 150% or maybe double w/e you're making now as a content creator to be their lead game designer and promised full creative control, would you take it?

    Reply
  23. The biggest red flag in the world was that their "mmr" system had a max cap.
    Shows they fundamentally don't understand how the concept is supposed to work.

    Reply
  24. Just before you read, i speak french, english only as a second language, so i'm sorry if my comment is hard to read or understand.

    I'll be the one to disagree with you on some of the things you said. First its unrealistic to even try what you are suggesting, from an investisment in time, money and workforce it would take sooo much just to get things to where its acceptable. Second, this MMR system sure is far from perfect, but its not the same as the old one and here is why. First, they can actually get accurate data on how much of the player base is at a certain level. before everyone at rank 1 ired would have counted as the same strenght, now its ONLY the case when setting up matchs for the sake of not having to wait for hours on the end of the spectrum. we also dont know how much of the old rank 1 ire even are at the current "ceiling" of MMR, if its lower, then it get its job done and well at that, making the top end playing even more against each other than before (witch would make sense seeing how increadible killers get mostly great suvivors against them, while they get far less bad ones than it was the case in the old system) and in doing so, achieve what they want. It might not be what a lot of people wants, but what the devs want is clearly said later, its making it so that you get hard games so that you can feel values from your good games. And it achieve that, but it make perks imbalance even more visible, in turn making people judge the system as a bad system, while the system does the job the devs wants it to do perfectly.

    To resume my point, i think what the community expected from the system and what the devs wanted from it is simply not the same, and that's why most like yourself think its simply not doing its job.

    I'll also add this, even if the system is far from perfect on a categorisation points, overall good survivors will escape more often, if that's what they want to do. Of course that does not mean its what a good survivor wants, but that's what the current system check to know the skill level of the player and so like in any other ranking system, it will punish people for not doing what the game/devs consider to be the "valuable" objective, in DBD surv side: surviving, and in the killer side: killing. Its a straight foward system that fits in what the game ask of you when you play. If you dont like to play like how the system ask of you great! just do so and you'll lower yourself in MMR by doing so, while still doing what you like, nothing bad comes out of doing that.

    Reply
  25. "Your actions during the match have too many variables to calculate"

    Then why are the emblems still here? If the match has too many variables to calculate, why are emblems, the thing meant to calculate the variables during the match, still in the game if they think it's impossible to do properly?
    (1/2)

    Reply

Leave a Comment