How the MMR works, and why what Patrick said makes absolute sense (from a dev POV)

Oops, you were so close, and yet you are still so far away. Please continue to go educate yourself on what MMR is, and come back to me with the proper answer. Or perhaps you actually do know the answer, but don't want to say it. Hm.

Good, so your definition of "team" is that it's a group of survivors working together until they become unnecessary to fulfill the win condition of escaping. That it is simply an "option" and not a "requirement" to play as a team during each match, and that players are free to NOT work together, and NOT escape together - basically that it's irrelevant to the wincon of the game. Great, then we are fully in agreement. If that is how you want to define the term "team", you should have said so from the beginning. I would personally would have just used the term "alliance" but it looks like it was just a terminology miscommunication. Glad we understand each other now.

No, there is only one win condition for the survivor - to escape. If you believe that there is another win condition as the game currently stands, feel free to provide official references that indicate it to be so. Or perhaps you are confused and can't differentiate between the game's actual win condition, and personal goals that individual players set for themselves in the game.

What did I just say about Patrick's quote? Until your "team results" actually becomes implemented in the game, they are just unsubstantiated theories and conjectures with no substance, and meaningless to the discussion of what the game is in its current state.

I dont think the current MMR reflects the skill of a player,like many said,you can loop the killer for 5 gens and die,while the other 3 escape while doing nothing but gens.

Also what's that about many ways to skin a cat,is english not your first language because that's a weird example.

Good job! MMRs are based on players' wincon of each match. In the case of DBD, as stated in the game's tutorial and its official website, the win condition for Survivors is simply defined as an escape. I've already explained in detail why Cote's avoidance on the topic of wincon is irrelevant and discarded. The emblem system on the other hand, as you previously stated, has nothing to do with the wincon, but rather only the accumulations of activities that players did within the match. That's why nobody would consider the Emblem system as being even remotely the same as MMR.

I claimed that there is only one win condition. Results? I'm sure there are plenty of results that can be calculated and tracked in a match - the number of generators I've completed, the number of chases, the number of flashlight stuns, the number of pallets I threw down, the number of tea-bags, the number of lockers I've hidden in... so what? None of that is the actual wincon of the game, and that includes any "team results" that they may be tracking. You're jumping to conclusions and equating the word "results" as a way to indicate how to win the game - because it fits your personal narrative, and can't face facts of what the game specifically entails as being the win condition in the tutorial and official webpage.

That seems to simply be your assertion to fit your narrative. You are more than welcome to provide direct, concrete evidence that Cote's quote is referring to the Emblem system though. I doubt you can though - just more assertion, inferences, hypothesis, and stretched imagination.

Change in the matchmaking is simply an indication that the old matchmaking system was not working, and therefore needed a new one - it has nothing to do with the win condition, nor does it imply any changes in the win condition.

Ok, let's go ahead and pursue this line of thought further. Cote indicated that win condition was based on whether the players felt they had a good game, whether they had fun. By that logic, the players win as long as they felt like they had "fun" or "had a good game" and lost the match if they didn't. If the players felt that they had "fun" in the match by simply jumping in lockers throughout the match... by his definition, they automatically win. If they felt like they had "fun" in the match because they rescued a survivor right in front of a camping killer and the sandbagged survivor died in a hilarious way... they automatically win. Does that sound like a fitting definition of "winning" to anyone who understands basic game design? And what if everyone in a match felt like they had "fun?" Do they ALL win? How about the opposite where nobody felt like they had "fun?" They ALL lose? How do you matchmake players in such situations? Do we match players who most frequently have fun with other players who feel the same way, and match players who often don't have fun with others like them? Or should the developers figure out the level of intensity of "fun" that players are having to do the matchmaking? How should the developers measure "fun?" And you say that the emblem system is related to all of this? How? The emblem system had accumulated points for doing generators... what if I don't have "fun" doing generators? What if I don't enjoy doing chases or bothering to go rescue someone?

The ridiculousness of it all is the reason why Cote's lip service in that quote can be discarded as nonsense. The fact that you are still entertaining that part of his quote when I've already given you the history on why they constantly avoid the topic... it's pretty sad.

No, the Emblem system simply used "performance" in a given match to rank players, and had nothing to do with the win condition. Again, you are free to provide any evidence that points to the contrary, but I doubt you can find any.

Didn't we already agree on the definition on "team"? Why are you going back to this nonsense? We've already agreed that in term of this game, "team" is defined as just players choosing to work together (or not) and escaping together (or not.) And therefore has nothing to do with the wincon.

Look, if the wincon of the game is actually team-based, you'd find something like this in the manual/tutorial/official website: "In order to win, at least 2 Survivors must escape through the exit gates. You lose if at least 3 Survivors from the trial are sacrificed to the Entity." But nowhere does it mention anything like that - and in fact, it indicates that the goal is simply for the player to escape. I understand that you and some of the others would personally (and apparently, desperately) prefer a team-based wincon, but as the game currently stands, it doesn't and it never has. Could it change in the future? Who knows. But not right now.

Again, not right now. Perhaps instead of getting ahead of yourself, perhaps you should wait until it's actually been done. If Patrick finds that adding "team results" variable just doesn't seem to reflect the MMR accurately and therefore abandons that idea, then what?

Again, the emblem system did not consider whether players won or lost, but so-called "performance" that players within the match. So no, "according to me" that particular matchmaking system was NOT based on wins.

Again, if you want to define "team" as players choosing to work together (or not) and escaping together (or not), then that's fine.

I don't much care for players' past misconception about the game.

In the current MMR system, there is never a situation where all 5 players win or lose together. That's all that matters in this game - there's nothing wrong with 2 players winning and 3 losing in a 5 player game.

Sonewhat like. (Devs decide points to give) Survivors=

Gain points: escape by exit and number of players, all gens completed, seconds for chase, secure unhooks, bodyblocks, flashlights and pallet saves, totem cleanse/boon, heal teammates.

Lose points: die, teammates die, you drop pallet, you go Down

Killers=

Gain points: kills, power hit, down survivors

Lose points: survivors escape, seconds for chases without pallet dropped or hit get, Miss power hit, seconds for hook camped

We'll just have to disagree on that then, because I really don't see (in theory and my lived experience) that the people at high ratings (the most skilled at consistently personally surviving) aren't also some of the most generally skilled players, including teamplay skills. The system of course is not about skill in and of itself, let alone teamplay skills, so an argument would increasingly become about whether it should be. I for one have no issue with the system being about seeking 50% personal success rates for everyone, even if that contrary to my conviction would not correlate with any skill or teamplay skill whatsoever. Well, I would likely not be as welcoming of the system in that case, but still more welcoming than of a system that actually tries to define, detect and weigh various different "skills" and pairs people based on them with no regard for success. The Emblem system gave me enough of a taste of what that's like.

I see your point, and I empathize with your worse-than-it-used-to-be playing experience, but I don't think your personal experience is very representative of the system. As I've said, I suspect (and you do too) that rather than any detrimentally-selfish players actually having climbed to your rating by consistently surviving, it's more likely you have just fallen due to consistently dying more often by taking risks (that may very well not consistently be beneficial for group survival either, who knows), or stuff like playing for BP. So rather than it being indicative of selfish players rising in rating, I think it's more so just... that you're being paired with simply worse players at lower ratings, not few of which may also just be making bad decisions and plays that can be seen as "selfish" but are really not deliberate decisions but just expected tendencies of worse, inexperienced, scared players that regularly don't even know much better.

With regards to all of this I really can't say anything else than I have been saying, which is that such risky/teamplay plays that actually result in consistently increasing the survival chances and rates of the group (lead to a greater-than-neutral return in how many survivors escape through exit gates) will over many matches also be reflected very accurately in the personal survival chances and rates of the individuals employing such play. Even in cases where the last gen is guaranteed to get finished if you keep working on it instead of helping another survivor that is on death hook (and here again we are in a very late stage of the game, so how detrimental can the selfishness really have been if we take this as a consistent occurrence which it'd have to be in order to be systemically relevant), an endgame with 1/2/3 survivors is objectively still much more dicey than one with 2/3/4, every time. Or if you go for risky basement saves, that is actually only a good teamplay if it actually results in you saving them, and it also leading to more people escaping than otherwise would have. If you die trying to rescue in that case, it was a bad teamplay and will result in more people dying than otherwise would have, and if you successfully save them but get hooked yourself for it, it was a bad teamplay if it doesn't lead to more survivors surviving than would otherwise have (so if you merely trade your life for theirs, that is not a good teamplay risk to take, it's not a correct play from any perspective in that case but your own preference of going for risky plays and other survivors surviving instead of yourself).

There's countless scenarios you can come up with where selfishly refusing to help might lead to one's own survival at the cost of a decreased group survival, but none of them are clear-cut and far from guaranteed, and in most of even these cases there's definitely room for argument that personal survival is still more likely in the case of helping, and will over many matches accordingly have the impact on the rating even when looking at such scenarios exclusively. Same for general skilled plays, where I think you are falling back into the mistake of looking at outliers. The person capable of running a killer for four gens won't always do so, won't always then get camped, won't always then not get saved, etc. Rather, since they are good at running, over many matches that will much more frequently positively affect their survival rate, in the endless different scenarios where that is of obvious benefit.

Regardless of all of that, even if you were right and rating would not correlate with certain skills such as teamplay skills (or any skills at all), even if plays that actually consistently lead to increased group survival rate could possibly consistently lead to decreased personal survival rate, it is still not the goal of the system to select for skills or team results. If a player dies more often than they survive, the assumption is that most players in that position would not be very satisfied, and should therefore be matched such that they have a greater chance of survival in the future. Even if that player deliberately sacrifices themselves for the survival of others, it stands to be reasoned that they don't actually want to die, so why not put them into a place where they can potentially rescue fellow survivors without themselves being as likely to die doing so? Of course, this could lead to such players ending up at ratings where they otherwise might not "belong", but if their chosen playstyle actually leads to them consistently dying, the system cannot and should not make some complicated exception, it has to assume that player is simply having trouble not dying and doesn't actually want to die. Yes, the group-based metric will "account" for teamplay skills more (they will have a more consistent impact on the rating, leading people to appropriate ratings faster than the current metric and stabilizing them around those ratings more), but there are still tons of scenarios where selfish play can lead to increased group survival and selfless play to decreased group survival, and there's plenty outlier scenarios that make as little intuitive sense to us as with the current metric, such as the Meghead dying in the first minute of a match but getting an increased rating if her "team" manages to do it without her.

The only reason I used tournaments is to say that even if the 3 remaining survivors are some of the best, most coordinated, most win-oriented players (and in most tournaments they actually do win by getting kills/escapes above all), the 1v3 scenario dramatically shifts the power balance in the killer's favour, even at very advanced stages of the game (including the endgame), and dramatically decreases the survival chances of everyone. Players that are anti-altruistic in principle from the get-go or just awful at teamplay, will much more often have fellow survivors die earlier than players that are altruistic and good at teamplay, and the dynamics of 1v3 scenarios are one highly compelling argument (of various others) for the idea that such anti-altruistic play results in decreased personal survival rate. At any level of play. I just can't see that low group survival rates in the matches of a player are anything but a great predictor for low personal survival rates of that player, aligning with each other ever more closely the more matches you look at.

But if those players with time are bound to end up at different ratings, they are less and less likely to be paired with each other in order for that scenario to arise to begin with. Particularly if that scenario does arise consistently in cases where they are paired (altruistic players dying, selfish surviving). Gradually, the system should lead to more alike players being paired more often, since they'll have comparable survival rates and therefore ratings if and insofar their playstyle (such as self-sacrificial altruism) actually has a consistent impact on their survival rates.

Very cool if you'll actually take stats. I'm not playing a lot at all lately, but hell, I'll record my stats too (we can just take screenshots of the tally screen and collect them for a while, then evaluating after 100 images or so).

This is definitely the big problem, though I'm not sure any matchmaking system could better deal with that, at least not without creating prohibitively long queue times (although I think at least at higher MMR brackets there's still merit in increasing queue times in favour of making closer pairings). But I do think that the group-based metric by virtue of leading players to the appropriate ratings more quickly and stabilizing them around those ratings more will obviously also lead to a decrease in pairings of players that would over time even with the current metric settle around different ratings, so you could very well see a tangible difference in matches even on the personal, "lived experience" level. We'll see soon enough, hopefully.

While I for one agree that the game is better as a team game and would even want BHVR to push more for that, there's a few things here.

For one thing, as you yourself point out, the game is not strictly a team game, people may disagree that pushing teamplay is desirable. From a designer's perspective, the concept of a multiplayer game where a group has the same objectives and conditions that have to be met in order for them to achieve their objectives but are still individuals with personal objectives and personal motivations that may choose to cooperate and coordinate but aren't strictly forced to in order to achieve objectives (personal and otherwise) can be very interesting. As opposed to a more strictly team-based game where people can and indeed must cooperate and coordinate to achieve team-based objectives that they either succeed or fail to achieve only as a team-entity (i. e. you only win if 3 survivors escape), the "group of multiple individuals" format allows for a different type of player agency, more meaningful personal decisions (or rather more personally meaningful decisions), and an overall interesting dynamic that one could very well would argue is more fitting for the horror gameplay type (everyone out for themselves but alliances may be made; outrun the other survivors not the killer, or throw something (yourself) at them) and creates rewarding gameplay interactions too (in a way of "game theory", where players have to decide for themselves much more whether they deem it worthwhile to help others, whether it is at a given point worth risking themselves for it). Who knows, maybe if BHVR were to fundamentally re-orient the game and outright tell players they only win as a team (e. g. tally screen giving them a fat "you failed" if less than 2 survivors escaped total) it would be less successful than it has been, because players like that tension of everyone potentially being out for themselves but implicitly and explicitly able to help each other.

For another, I really don't think endgame heroics are worth it from a team survival perspective either. Over many matches, I suspect they decrease group survival, with at-best-net-neutral survival outcomes being the most common result, and instances where more survivors die than otherwise would have outweighing comfortably such where more survivors survive than otherwise would have. Even from a "team" perspective, the only reason you would regularly go for such plays is still the gameplay it yields, the thrill and fun of it, not actually it being a viable risk/reward team survival play.

Beyond that, the system doesn't really encourage or discourage playstyles. The rating is invisible, many people don't even know about it and those that do are free to not care about it. The only effect the rating has is that people that died more often than they survived at previous ratings will meet opponents that failed to kill more often than they killed at those ratings. There's no downside to it conceptionally, so if you like going for (even unreasonably) risky and (even not-beneficial-to-the-group) self-sacrificial plays, you have no reason not to do so, the system doesn't discourage you from it or punishes you for it, on the contrary, if you actually consistently die more often than you survive due to playing in such ways, you should eventually arrive at a level where you can survive more often again, and surely such a player does not actually want to die, they just want to go for such plays. Can you imagine having fun going for risky and self-sacrificial plays in the tutorial match to try and protect your bot teammates? I've done it, it's actually pretty fun, and of course, even the lowest MMR levels of actual players are still much more fun and rewarding than that, so even if a risky, self-sacrificial playstyle would lead a player to eventually sit at the lowest of ratings, they could still have fun and success playing as they are.

I for one do agree that the game is better as a team game where everyone is on the same page about playing "for the team" and the team result (though that may be because I like competitive DbD - I do actually suspect most players in pubs enjoy the "chaotic", "individualistic" gameplay experience more), and I welcome the addition of the group-based survival metric. I just don't think the matchmaking system is really the point of interest to yield more such type of play, it should have to be actual gameplay changes, and actually designing the game around the "team" in general more (with clear and clearly-communicated in-game win conditions based on the group result for starters).

It's just not very meaningful because in order for those scenarios to have an appreciable impact on ratings, they would have to not only be frequent enough in numbers themselves, but also arise frequently to begin with, meaning such "selfish" players would first actually have to consistently lead games into endgame scenarios. Either that does not consistently happen because their selfishness prevents matches from getting there often to begin with, or their selfishness doesn't have a very negative impact on matches overall because only once it comes to the endgame do they lead to other survivors being less likely to survive.

Either way, as I argued earlier, I think it on average would also have a negative effect on a player's MMR if they went for such plays in a team-based environment. It's just not a "correct" play from the survival perspective, personal and group, and in tournaments you will also see players ditch the last survivor 99.9% of the time. The MMR has to be based on success in order to function as desired (and at all), which means in any case, personal or team success, these plays would not be "encouraged" in an MMR system, rather they would always be something players are free to go for if they personally like the gameplay they yield, the thrill and fun and adrenaline, if they don't mind dying due to this often or enjoy enough to make up for this the rare instances where they are actually successful and get more players out. Obviously I have no hard numbers for this, but I really am convinced that if you took group survival stats from equal players one of which goes for endgame rescue attempts consistently and the other does not, you'd see that the latter actually has higher group survival rates too.

First of all, I want to again highlight the fact that we have to distinguish between merely-altruistic plays and altruistic plays that actually lead to increased group survival. If we look at the former, then there is no reason to go for them from either perspective, because you are not helping the "team" doing so. You yourself are part of the team and risking your life for no numerically-greather-than-your-own-life return is a bad teamplay. If we look at the latter, I really have a hard time seeing that there are many situations in which what you say is true, and on the contrary, see many more situations in what I suspect is true. That is to say, if in such a mid to late game scenario there are plays you can go for that actually result in an increased group survival chance (and consistently do so, as in leading to an increased group survival rate over many matches), not going for those plays will regularly result in decreased personal survival chances (and rates over many matches) too. Doesn't mean you actually always die if you go for them, or always live if you don't, but I just don't think there are very clear consistent routes of play to personal survival from a non-open-gates-endgame perspective that consistently come at a detriment to group survival. Even if you're on the last gen, having someone else be taken out can be a huge problem, and in cases where it isn't, helping them instead of finishing the last gen immediately would regularly also not have resulted in huge problems.

These are obviously less clear arguments and they revolve more around the specifics and complexities of the gameplay, we may not agree on them (and dev stats on group survival vs. personal survival rate ratios would be cool to have for an a little more objective look at this), but then ultimately, I would have to come back to the standpoint that even if we do assume that there are a lot of clear, consistent routes to personal survival that come at clear, consistent detriments to group survival, the system does not have to care about that - as long as it works to instate around 50% personal and therefore also group survival rates on average overall (it does), it's working as "correctly" as it possibly can, since that is its only goal, giving people more even chances to succeed and actually more even success rates. The gameplay experiences then may be more undesirable for some people (at least until they settle at different ratings due to playing in other ways and not succeeding as often as a result), but that is something that should have to be changed in the gameplay design rather than attempted to be worked around with a matchmaking system, set actual hard, gameplay incentives for people to go for and benefit from teamplay.

You could refer to the things @knell is stating for this. The only clearly-defined win condition for this game has always been to kill or survive. Just that. This is not at all controversial. When you ask people what a survivor has to do in DbD, they don't say "interact with the killer, go for unhooks, be in the killer's terror radius for a while, hit great skill checks, take hits, heal often, cleanse totems, try to get chase state a lot, try not to unhook other survivors if they are likely to be downed within 10 seconds after the unhook, and do a whole bunch of other stuff and get a so-called pip which allows you to get a better rank/grade"... they say "you have to survive". This win condition has always been very clear from the basic game description, as well as clear in most of anyone's mind that thinks about the game where survivors flee from a killer - they want to escape alive. (And almost nobody automatically extends this to "you have to survive but at least 1 other survivor has to also survive, and 2 or 3 others if you actually want to win".)

The Emblem/rank system was awful at making good matches happen, the disparity in skill and performances couldn't have been much more drastic. Average kill rates of 70+% in red ranks? Ludicrous really. A system that looks at actual success of players in terms of absolute (yes/no) and diametrically opposed win conditions (which MMR has to be) is just better at creating more even matches, and killing/surviving are definitely what most players in this game consider success. On top of that, most people definitely care the most about their own survival, more than about that of others. We could argue whether this is actually the case, I of course have no objective evidence of this, but I find it to be completely common sense. So having the system make matches happen where people specifically personally have more even chances and rates of success (personally surviving) is desirable to increase overall satisfaction levels. Mind you, I don't actually think the system would yield very different results with the group-based metric, and I would welcome even switching to it entirely, but the devs most likely have reasons why they aren't doing this, and will at most be switching to a hybrid metric of both personal and group survival. One reason for that I suspect is that they do not want scenarios (rare as they may be, little as they may affect the overall system's functioning) where individual players can have loss streaks of dying a lot but not being adjusted to a rating where they can survive more often because their "teams" are surviving a lot.

Well, I don't think the game is one where survivors are essentially competitors, and where turning on each other can consistently assure personal escapes. Teamplay is integral to the game precisely because it correlates closely with personal success. The question would more be about whether players should be free to decide for themselves whether they prefer altruistic even self-sacrificially altruistic gameplay, or whether they only care about it to the extent that it enables their own survival, or yet that they don't care about it at all and only want to be a lone wolf. The game does not dictate this, functionally teamplay is absolutely beneficial, but not absolutely necessary to "succeed" in terms of playing the game "correctly", with the personal escape goal or indeed any other goal. You may not mind dying in order for other players to survive, but someone else may well not mind others dying for their own survival, or not mind dying themselves by playing a lone wolf style, because they only want to survive "by themselves" as much as possible. And so on. Although I do think for the vast majority of players, the teamplay aspects of the game are obvious and something they naturally gravitate toward, and not only opportunistically (for personal and group survival) but even from a truly-altruistic place of motivation, such as trying to contest Bubba facecamps because they just feel bad for the person on hook and don't want them to have a bad experience, feel obligated to try and help, even if the chances of saving are slim.

Yeah, hopefully they'll share more on just what the "hybridization" of the metrics entails.

I remember you too, after all you are one of the main reasons why I ever even started looking at and contributing to these forums at all again. Cheers!