Behaviour doesn't understand skill

I'm actually baffled how backwards a lot of people are taking this whole MMR thing. This is like the third or fourth entirely-too-long post I make about this topic, but I somehow find it difficult to express in simple, short terms just how backwards people are looking at this system.

I think it wasn't a good decision to call the matchmaking system "skill-based", but not because it doesn't factor in skills (it does), but because people are getting way too hung up on that word and losing out of sight completely what we are talking about here. We are talking about a matchmaking system. People seem to believe this is some kind of ultimate judgement system, that rates players and somehow they are then only worth as much (or as skilled) as the system says, as if a good rating would be a reward, or as if it granted some sort of reward, and as if it would be unfair if it didn't correctly rate someone based on all of their skills.

First of all, realize that the system is only meant and indeed only used to create matches, and specifically matches that are more even in terms of chances of success for all players involved. There is no reward for being highly-rated, your rating is not even shown, and the state of your rating does not signify anything but the pool of players you will (ideally) more often be paired with and against, namely players close to that rating. Even if the system did not rate players based on skill whatsoever (which it again absolutely does), why does anybody care whether it actually assesses skills? Sure, if there were an actual, notable reward for achieving a high rating, like cosmetics or Shards or whatever, I could understand a little better complaints about how the system would not fairly or accurately rate players, that they think it does not account for the types of skills they may have that may also be important... but there are no such rewards, the only thing that changes if your rating increases is the rating of the players you are supposed to be matched with.

I find it perplexing that people take issue with the idea that the system does not rate people fairly based on skills, not only because it ultimately doesn't have to (and despite the fact that it ultimately does), but also because often when I hear other suggestions for matchmaking systems, they are stuff like "just throw everyone in one giant pool", or "use emblems", which are both basically describing the same thing, the old matchmaking system, and only go to show that people often don't actually care about consistently having opponents roughly around their own skill level, they if anything seem to want a completely willy-nilly matchmaking again. And so even if we do assume that the current matchmaking system does not actually match people based on skills, why is that then an issue for so many?

More than that, the most common complaint I see about the current matchmaking system is players saying they have a hard time killing survivors. Yet the system is literally meant to make players average out to around 2 kills. So either they are saying they want to have a system in which they can average more than 2 kills (which would be silly of course), or they are actually fine with what the current system is designed around, only they think it does not actually fulfil its purpose since they do not average around 2 kills (which they however could already very well be).

Either way, let's tackle the arguments of Scott here.

"Killing/surviving does not really determine skill."

In no game in existence MMR is simply about "skill" itself. I will refrain from touching on the hockey analogy because it is not important, but the gist of what Patrick said is completely true. Namely, MMR systems are concerned with winning. Not with winning in the most difficult, fancy ways, just winning by any means possible as long as those means consistently lead to winning. In any proper MMR system, people only advance if they consistently win. Consistently winning requires skill. The skills required to win correlate with winning. The MMR therefore factors in those skills. That doesn't mean the MMR system is about skill, it's about winning, and skills only factor into this if and insofar they respectively correlate with winning. It's an important distinction to make between skills that correlate with winning (which is: to contribute and lead to winning), and skills that are merely fancy, difficult things one can do, but that do not necessarily increase one's likeliness to win, and in fact can even do the opposite. You could separate them terminologically by speaking about survival/killing skills as the one that matter for MMR, as they pertain skills that correlate with winning. As such, you would specify "skill-based" MM as "survival/killing skill-based" MM, or "skill-correlated" MM.

Sure, there are a lot of things in these games people can be skilled at, but if they are not things that lead to wins, they cannot and indeed should not be factored into MMR. You can be the best moonwalker in the game, but if you can't use that skill to actually increase your survival rate, it has nothing to do with MMR, and it would be absurd to argue it should. Basing matchmaking on skills just on the basis of how difficult they are rather than on the basis of whether those skills translate to increased chances of winning is completely backwards and misguided understanding of what MMR and its purpose is. What would high-MMR gameplay look like in a system that looked at difficult stuff for difficulty's sake? You would have 360 youtube montage survivors spinning around in one corner of the map, and a skybilly going for trick montages in the other. Endlessly of course, because nobody would ever do gens which is unskillful and purely, primitively survival-oriented, and Billy would never kill anyone but rather try to win an infinity-approaching amount of chases with tricksaws in a single game.

I hate to break it to people, but hooks are not actually the win condition in this game, and in its current state absolutely cannot be. The win conditions are killing and surviving, and you would have to redesign the game in major ways that shift its entire character to change that. Yes, someone might be very skilled at chasing survivors and therefore gets many hooks, but if they cannot convert those hooks into kills or get enough hooks that this naturally happens, they are not skilled at winning, which is what matters for MMR. For MMR only actually winning is important because its entire purpose is only to match players such that their chances of winning are more even, because that leads to healthier competition and more satisfactory playing experiences where players' win/loss ratios are more balanced, i. e. where they are more likely to succeed at least around half the time, which is to kill/survive, which in turn is also what most people actually mostly care about.

Now mind you, a player that is very skilled at chasing survivors is absolutely likely to often kill survivors, and so if they for some reason consistently end up only getting hooks and not kills, they are not converting their chase skills into wins, thus they are lacking other skills, such as macrostrategic skills (or they are going out of their way to not kill survivors, meaning they are actively hampering their own performance for difficulty's sake, which is again something MMR cannot and should not account for, just like it should not rate people higher that play perkless for some reason, or with a steering wheel, or put other such artificial limitations on themselves).

Getting many hooks is certainly indicative that a player is skilled at chases, but if they are not killing survivors that way, they are evidently not skilled enough at them for it to make up for other skills which allow other players to kill survivors with less hooks. Camping, tunnelling and slugging exist, they allow kills to happen with less hooks, and they absolutely require skill to make use of against good opponents that know how to play against them, which you in an MMR system would be expected to go against. As long as it is possible to kill without hooking survivors 3 times each, hooks cannot possibly be a win condition, and therefore also not possibly what MMR concentrates on. Base MMR on hooks and low MMR will be a realm of 99% kill rates where bad and new players get demolished by campers, tunneller, and sluggers constantly. Base MMR on hooks without making survivor MMR also be based on hooks and both sides can win simultaneously, rendering MMR useless as tons of players will be inevitably gaining MMR and ending up at around the same ratings. Base survivor MMR on hooks and they can kill themselves and leave each other on hooks and guarantee "wins" that way, meaning you would have to get rid of hook timers, of the ability to kill oneself on hook, and remedy situations where all survivors are hooked or slugged, allowing them to unhook themselves and recover. Basically, you have to remove camping, slugging and an entire sphere of strategy from the game. Rework the game like that to make hooks the win condition and you alienate the character of the game as well as a lot of players because killing and surviving absolutely is what the game revolves around and what most people care about.

Besides, why stop at hooks? What about the tons of skills that merely looking at "hooks" does not directly account for? Whether you won your chase with a fancy long-range hatchet shot rather than a point-blank one, or with a mindgame on a safe pallet rather than an unsafe pallet? Whether you used good chase perks, or chase perks at all? Add-ons that helped you? The map you were on? The killer you played? How many red stain mindgames you went for and succeeded with? How well you zoned? Shouldn't these things as well as the thousand other things that require skill be measured and accounted for by the MMR system? Aren't these things to hooks what hooks are to kills? Wouldn't we have to acknowledge that the system would become impossibly convoluted and impossible to design altogether, if we wanted to define, detect, evaluate and weigh the countless different skills that can go into the gameplay? Wouldn't we come to the realization that we can and in fact must simply look at the results that these things lead to and that we effectively, over large enough sample sizes, account for all those things by doing so?

The higher MMR levels are not supposed to be some holy realm where the most fun, fancy, difficult, noble, interactive, exciting, etc. gameplay happens. It is not something players are even meant to want to achieve. It is not some ladder to climb that people should fight their best to make it to the top of. It's a system entirely in the background, and only meant to match players with more like-minded and like-abled players when it comes to winning, without them even thinking about it, and at the top levels of that it is simply and precisely about pairing players that care the most about winning and are the best at doing so. That doesn't mean they need to be the best at getting 12 hooks, or the best at getting cross-map hatchet snipes or whatnot, it means they are the best at killing or surviving consistently, with everything that entails, the most effective and efficient, absolute "sweatiest", most "tryhard" loadouts and playstyles, filthy busted-setup-abusing, DHDS genrushers, mean 4-slowdown Nurse/Blight camper-tunneller-sluggers. And yes, consistently winning (killing/surviving) in that environment where you face off against other players that are increasingly also better at those same things, does require a lot of skill. Players in tournaments are also absolutely some of the most skilled in the game, and they too play for killing and surviving more than hooks, and it is actually not easy to kill by camping/tunnelling/slugging/just-doing-gens in that environment, a wide variety of high-level skills are at play, and chase skills specifically are definitely highly important for success there as well, the most successful tournament players are likewise some of the best 1V1 players that consistently get some of the highest chase times in 1V1 competitions. And believe it or not, plenty of people even have fun and find it exciting playing to win by any means possible in highly competitive contexts, that competition and playing-to-win and winning is desirable in itself to them, camping-et-al included.

If you personally do not enjoy how the game is played when everyone tries their hardest to win (to kill or survive), and if you consider other types of goals in the game to be more desirable or noble or even skillful (such as getting many hooks, even if it costs you kills you could otherwise get), then you are absolutely, completely free to play in those ways, MMR will not "punish" you for that one bit, on the contrary, you will even benefit from the MMR system, since you will be separated from those sweaty tryhards further above in the ratings that only play to win, as you will of course not be winning as consistently as they are. Instead, you will more often be paired against like-minded players that also do not care as much about killing and surviving, and you will have more desirable game experiences as a result, and games where you can actually average around 2 kills by going for many hooks (something that is indeed not possible at higher levels of play, the game at that level is not balanced around hooks but kills).

"Consistently" is another keyword to counter an argument of Scott with here: he points to instances where lower-skilled players in a team may survive without having contributed much to that win, or where higher-skilled players die despite having contributed a lot to the round, or yet where a higher-skilled killer gets 9 hooks but only 1 kill, whereas a lesser-skilled killer gets 2 kills with 2 hooks... And the answer to this is completely obvious and apparently lost on so many: over large enough sample sizes, the higher-skilled player that contributes a lot to the rounds will survive much more often than that lower-skilled player that contributes little; the higher-skilled killer that ends chases so fast that they can get 9 hooks in that round will over large enough sample sizes kill much more survivors than that 2-hook 2-kill player that must rely on cheap tactics (primitive facecamping) that get more and more impractical to be successful with the better your opponents are. MMR is about consistently winning, and it's completely obvious that better players will win more consistently, the correlation between skills and winning is blatant, and the scenarios he and other people point to are edge-case exceptions to a rule that would over large enough numbers of scenarios always prove true and lead to the player rating rather accurately accounting for those skill differences. Sure, if that 9-hook killer goes out of their way to always make sure they don't hook survivors they've already hooked, they might not surpass the hardcamper in kill rates, but if that killer wants to win they will of course do precisely the opposite. If they on the other hand aren't going out of their way not to kill but in fact simply cannot consistently kill despite getting many hooks, they are simply lacking crucial other skills that also play a role in this game, since this game is not at all only about chases.

The lower-skilled players that only do gens, hide, facecamp, suck at chases, etc. will hit an MMR barrier much earlier than those higher-skilled players, since they will at best stagnate playing in those ways over many matches. But that they gain MMR in rare exception matches where they are hard-carried by better teammates or stupid decisions of opponents is not actually a problem either: you actually want those players to gain MMR such that they face better opponents that are better at killing/surviving and as such punish those playstyles and that lack of skill. You don't want the locker hero or the facecamper Bubba to get easier and easier opponents every time they succeed, opponents against which they will be even more successful playing that way; rather, if they are successful playing that way, you want them to face tougher opposition such that and until they aren't anymore.

I'm not saying the game could not be extensively reworked to actually become about hooks vs. preventing hooks rather than about killing vs. surviving, and I even agree that it being changed to at least shift somewhat more from being about kills to being about hooks is a desirable direction, but the reality of the game as it is simply is not that, and without major redesigning it cannot be. Simply tying MMR to amount of hooks without doing anything else would be absurd. And anyway, I do think camping/tunnelling/slugging have a place in the game, they are integral aspects of the "macrogame", the strategical side, they involve decision-making and play-making and therefore different layers and degrees of mental and mechanical skills of their own, and they on top of that add a "danger" factor to the game by being capable of in a short duration turning matches around and ending matches altogether, which makes the game more exciting, thrilling, differentiates the trial-to-trial gameplay experiences much more, and is a big part of what makes the game a horror-type game on top of that (including that it is about killing and surviving specifically, which are clear and thematic horror-type conditions, as opposed to... "achieving X hooking actions" and "preventing X hooking actions from happening" - again, people at large will never feel as if they won a game of Dead by Daylight if they all died, or if they had all survivors escape, no matter how much you tell them the hooks actually matter more).

I do obviously agree with the last section of Scott's argument here though, namely that the matchmaking system actually has to match players with similar ratings if it is to make any sense at all to calculate ratings to begin with. And it currently doesn't really do that all that reliably. It does so reliably enough to still be a much better system than the rank-based system in my experience and that of the people I know and follow, but it's definitely far from perfect in that regard. It could also very well be improved by looking at group survival rather than per-survivor survival, but as per my earlier explanations, it doesn't actually necessarily need that for ratings to accurately-enough represent player skill over large enough sample sizes - survival/killing skills of individual players very closely correlate with individual survival rates/per-survivor kill rates over many matches. But perhaps the rating adjustments would be a little more reasonable if based on the average rating of groups, rather than that of the individual survivors in them, so that could be an actual improvement.

Finally I want to note that looking at hook stages (not hook events) would account for kills and hooks simultaneously, and it would not require any changes to the game either. So yeah, I guess that's my suggestion for a compromise between the MMR system we currently have and that fits the reality of the game, and the MMR system some people wish to have that however does not fit the game's reality: Use hook stages to support the kill-based rating adjustments.

The matchmaker prioritizing queue times over rating similarities is definitely an issue. Matchmaking systems are difficult to get to work well when accounting for both queue time and making reasonable matches happen, most of all of course because they have to work with the actual players queueing to play in respective regions and at respective times. But I do believe the current system does have leeway to increase queue times in order to look for more closely-rated matches.

My survivor queue times are not terribly long at all, and in fact during certain times consistently shorter than my killer queue times. In the early morning to late noon hours, my survivor queue times tend to be shorter than my killer queue times, during much of the day my killer and survivor queues are fairly similar, and toward late evening and night my survivor queues start to get longer than my killer queues. This has been a fairly consistent experience for me for years, and MMR has not drastically changed things. I was gathering evidence from streamers for another post, but might as well post an example here to sort-of substantiate my claims:

Otzdarva (EU):

Survivor queues: https://www.twitch.tv/videos/1264522221

Queue start: 24:34; queue pop: 24:44 - 10 seconds

Start: 53:41; pop: 54:29 - 48s

1:11:34; 1:12:55 - 1min21s

1:30:07; 1:30:28 - 21s

1:42:28; 1:42:38 - 10s

1:55:54; 1:56:04 - 10s

2:23:02; 2:24:13 - 1min11s

2:34:59; 2:35:41 - 42s

2:51:14; 2:51:25 - 11s

3:04:38; 3:04:48 - 10s

3:16:27; 3:16:37 - 10s

3:38:36; 3:38:46 - 10s

3:53:05; 3:54:58 - 1min53s

4:11:59; 4:14:23 - 2min24s

4:31:49; 4:34:32 - 2min43s

4:47:46; 4:50:50 - 3min4s

5:03:12; 5:07:05 - 4min7s

5:24:11; 5:27:57 - 3min46s

5:40:35; 5:44:57 - 4min22s

This VOD covers a recent playing session of around 8 hours, and only towards the end do the survivor queues move away from a 30-second average (10-second median) value. And even at their longest they never surpass 5 minutes, which I think in the past has been the cut-off point beyond which the matchmaker all but throws rating (in the past of course ranking) considerations out of the window.

Another reason why I think queue times still leave room to weigh ratings more heavily is that the crossplay off queue in my region always featured a significantly higher density of higher-skilled players ending up in lobbies, and yet queue times were not unreasonable. MMR has made things worse because now various of the people that used to play on cross off here are playing cross on (since they find they get well-enough-balanced matches there often enough now, as opposed to before MMR; or they mostly practice in customs because there's lots of tournaments going on), but still, the cross off pool showed me that it is possible to get multiple high-level players into a match frequently without prohibitively long queue times - there are seemingly enough of such players queueing at roughly the same times in the same regions, the matchmaker just doesn't tend to pair them as frequently because it gives them other pairings beforehand rather than waiting just a bit more to look for those better pairings.

I don't agree that MMR has nothing to do with skill, some of the reasons for which are pointed out in my essay. The main skill aspect of the game people usually bring up that would matter the most are chase skills, and it is a completely obvious fact that survivor players that are better at leading killers on longer chases survive more often than survivor players that are worse at it, and that killer players in turn that are better at catching survivors quickly in chases kill more often than killer players that are worse at it. The correlation couldn't be much more close. Sure there are edge-case matches where this is not the case, but they are exceptions to the rule that will hold true consistently over large numbers of matches.

My personal experience and that of various people I know/play with/watch is not that MMR has made the game less fun. I don't think it's perfect, there are improvemements I would make (harsher ratings, higher cap, stricter matchmaker, look at group-based survival instead, take hook stages into account), but I absolutely much prefer it to the willy-nilly rank-based MM of the past. But perhaps it really is an idea to introduce a "casual" willy-nilly/Emblem/BP-based queue alongside the "competitive" MMR queue. People usually argue that queue times would suffer, but if we reasonably suppose that a roughly equal amount of players of both roles would leave one queue to play in the other, the role ratios of people queueing would not change and as such neither would the queue times. At least something to test perhaps.

Problem is that skill isn't the only part into matchmaking.

If you projectile spam in a fighting game you also aren't being skillfull but if you can win with that you should still go up in rank. So you can face people who can deal with that.

Same with dbd. If a bubba can get 3-4 kills with 1 hooking and camping them he needs to go up in mmr so he starts facing survivors that can punish that. If a claudette can hide all match and escape she needs to face killers that don't lose a 3v1.

People are tunnelvisioning on the skill part and forget that mmr is a matchmaking tool first and formost. Perfectly calculating skill is not what mmr is trying to do

Person disagrees.

OMG, HE DISMANTLED YOU.

Hey Pulsar. sorry to hear that, but I have to question you on your story.

Its clear that the game was not really made for survivors of your caliber, so why dont you just play when more survivors like you are playing?

lol dont ask me, I dont have an account that can find em.