Visit the Kill Switch Master List for more information on these and other current known issues: https://forums.bhvr.com/dead-by-daylight/kb/articles/299-kill-switch-master-list
We encourage you to be as honest as possible in letting us know how you feel about the game. The information and answers provided are anonymous, not shared with any third-party, and will not be used for purposes other than survey analysis.
Access the survey HERE!
Question about the MMR Accuracy in the new update
Just wondering whether the new Team-Based Ratings have been tested or will be tested for its accuracy? It is my understanding that for most types of MMRs (ELO, Glicko, TrueSkill), when you add in secondary factors unrelated to the win condition of a game, it becomes less accurate over time. (Accuracy in this context means the ability for the game to predict the matches' outcome.) The only advantage of adding extra elements (usually some sort of 'performance' based actions of the players) to determine MMRs is that it usually allows the players to get to their ratings range more quickly at the cost of some accuracy.
What has your MMR stats shown in regards to its accuracy so far (if it has already been tested to some degree)? Are you looking to make MMR more accurate than it is currently? Or are you adding in the Team-Based Ratings because it is what the community is asking for (as well as the issue with 'sacrificing self for others' that is in direct opposition to the current win condition) even if the MMR's accuracy suffers in some level? Have BHVR ever discussed the possibility of changing the wincon of the game?
Comments
-
While these are secondary factors, they're still related to the win condition of the game, being escapes.
The problem before was that survivor MMR was only influenced by their own escape rate, but that's now how survivors play. Survivors generally play as a team, and the actions of one impact on the results of the others. So personal escape rate wasn't a true representation of the survivor win condition.
MMR is always going to have accuracy issues, but survivor MMR simply wasn't representative of the true survivor win condition. So team-based factors are less about improving accuracy, and more about representing a true win condition.
Dying over and over again because your team mates screwed you over, or because you sacrificed yourself for you team mates, despite you personally playing well and contributing to the teams success, wasn't just inaccurate, it wasn't representative of the players skill, at all, it was in fact inversely correlated with the skill of the player, and that's the whole problem with survivor MMR.
0 -
But escaping would generally correlate with playing well and vice versa. Someone dying over and over again probably means they aren't playing well on some level, or the matchmaker isn't doing them any favors (which as far as we know isn't changing). Somebody *could* die because they sacrificed themselves for their team, but is that really happening enough to warrant the change? Is altruism for the sake of altruism now considered skillful? Because it shouldn't be. That's how games are thrown.
If survivors make the correct decisions at the correct times, they will escape more often than not. I don't see why the system needs to change.
1 -
As it is currently stated in both the game tutorial and in their official webpage, for survivors, 'Escaping the trial' is their win condition. Nowhere does it say anything like, 'Have 2 or more survivors escape the trial to win.' Would it be a better game if that was the case (like in most many vs one asymmetrical games)? Possibly - but that isn't what it is currently. Which is why I asked, "Have BHVR ever discussed the possibility of changing the wincon of the game?" I made it a point to emphasize that the MMR doesn't determine the game's win condition - in fact, the win condition determines how the MMR is calculated. Like I stated, for MMR to accurately reflect the players' ratings, they usually just go the values reflected by the wincon. I'm just curious if BHVR found a way around it to demonstrate better accuracy somehow, or whether they are just willing to loosen their accuracy to appease the playerbase, as well as their contradictory end-game design.
0 -
"Escaping the trial" is vague, that's the issue. Is it personal? Is it a team based goal? BHVR doesn't have a clear stance on that, and yet the MMR system which attempts to quantify this win condition, does impose a clear stance, despite the almost intentional nebulous interpretation of the win condition.
Either way, your contributions as a team member affect the outcomes for other survivors. You play well, you help the team, you increase the chances of some or all of you escaping.
You can repair 3 gens and then get killed. The 3 gens you repaired might enable other survivors to escape.
You can run the killer for 5 minutes allowing other survivors to repair gens, who might then escape while you get camped.
You might make a last minute altruistic play and trade your life for another survivors, who escapes.
None of these things should be discounted. (they currently are)
Conversely, you can escape the trial as the last remaining survivor, despite the fact that your three team mates died at 5 gens. That should also not be discounted. (as it currently is)
To discount any of those things completely invalidates MMR as it applies to the win condition.
The way they've proposed implementing team-based MMR changes isn't a drastic shift away from what already happens, it's merely an additional multiplier.
They've stated that an escape will still result in an increase in MMR, and a death will still result in a decrease in MMR, but the amount at which your MMR increases or decreases will scale depending on the fate of other survivors.
So if you sacrifice yourself to allow three other survivors to escape, your MMR still goes down, but not by nearly as much, which would allow you to claw back some more MMR the next time you do escape.
If you're routinely playing well enough to aid your team mates, so that sometimes you escape, and sometimes your team mates escape instead, you will average a positive MMR. Whereas currently, you could average a neutral or negative MMR despite playing well and influencing a positive win condition for the team.
This isn't sacrificing accuracy to appease the playerbase. This is improving the MMR's interpretation of the win condition, to more closely align with the intended win condition.
0 -
I'm not sure what's vague or nebulous about "Escaping the trial." Why would you assume it's a team-based goal? As it stands right now, it's clearly written that survivors' objective is for them to escape the trial, nothing more and nothing less (that is, unless it's written somewhere that the majority of the survivors must escape to win.) Again, is the current objective for survivors a good win condition for this game? Maybe, maybe not, but it is what it is. Look, I'm not advocating for one game goal or the other - I'm saying that the way the current win condition is spelled out by the game, this change would normally make MMRs less accurate than it would otherwise. And even if it does make it less accurate, may be that's a sacrifice BHVR is willing to make - I'm just wondering if that's the case. But if isn't, and that's a problem, then one way to solve it is for them to change the game objective for survivors to clearly be team-based.
I think the discrepancies in our understanding is that you have this notion of "intended win condition" or "true representation of the survivor win condition." I don't - I'm simply going by what BHVR has written in their tutorials and official website.
But all of this don't really particularly matter in this thread. I've actually had this same discussion multiple times in the past. I was just wondering statistically, whether BHVR was able to find some way to make the MMR more accurate despite adding secondary variables, or they just came to the conclusion that its accuracy isn't as important as some other factors.
0 -
The problem is the devs have never actually clearly defined the win condition for the survivor side. Sometimes they act like it's every person for themselves, sometimes they act like the survivors are supposed to win or lose as a group. It's an underlying base issue behind a few problems in the game like some people sandbagging or ragequitting because "they" lost even though the rest of the survivors might be ok and could all escape, or people who think "pips" are the goal or bloodpoints or what-have-you.
As far as MMR goes, though, they're sort of shifting the win metric from "individuals escaping" to "how many survivors escaped". Personally I think it would maybe be even better if it literally was "If 3-4 survivors escape all the survivor MMRs go up, and if only 0-1 escape they go down, with no change on 2 escapes". That would be even more crystal clear as a win/loss condition for the survivors as a whole then the "you still lost for MMR purposes but not by as much if others escape" type of thing. (Basically semi-cooperative game design is usually bad, DbD unfortunately isn't an exception.)
1 -
I'm not sure where you're getting the idea that "accuracy" is contingent on a faulty win condition.
0 -
Nah as a killer main, I don't see anyway that escapes is a better representation of skill than something like a uncapped point system for survivor's. You could have a high mmr doing absolutely nothing but playing for gates after hatch is closed every single game. You could have a a low mmr running the killer for 3 gens just to be face camped. Way too many variables on survivor side to simply say escapes = skill. This game is not something like league where if you are good enough you can solo carry, the killer will eventually get you and down you unless you are miles ahead of the killer and have good rng. You may get hooked right as the gates are open and the other survivor's just leave without any attempt at a save.
0 -
Like I said, to me, it's pretty clear what the current game objective is - "Escape the Trial." That is what is mentioned in the tutorial/game manual and their official website. But it's obviously confusing a lot of players, so like you said, BHVR should probably clarify what the game objective is, given that win condition is what MMRs are usually based on.
But it's also important to remember that the win conditions define what metrics and variables are used for MMR - not the other way around (MMRs don't define what the wincon is). Like I mentioned in the very first post, developers in some games would add secondary variables (usually players' performance levels) to their "MMR" so that their players can be transferred much more quickly to their general range of MMR, but that usually comes at the expense of MMR's accuracy level. These game developers adding secondary variables don't change the actual win conditions of their game - which is why it introduces slight errors in the algorithm. But for them, it might be worth it, which is what I'm wondering if it's the same case with BHVR. (Or if they had found some way to maintain or improve their accuracy level despite adding secondary variables to their MMR - and if they did, I'd like to know how they accomplished that.)
0 -
Like I said, you have this notion of "intended win condition" or "true representation of the survivor win condition" and now, "faulty win condition." And like I stated, I don't.
0 -
So you're saying over the course of 100 games, the survivor who plays for hatch and does nothing all game will escape just as often or more often than a survivor who runs tiles efficiently, has good game sense, and knows when to save vs when not to save? That's disingenuous at best. How about over the course of 1000 games? Hiding for hatch can work at times, but it's not going to work on a statistically significant level relative to a genuinely good player. And a genuinely good player isn't going to be dying so often in endgame that the entire system is thrown out of balance.
0 -
In my experience as a killer main, it's a lot of rng on who escapes. One survivor can be an absolute juicer but that really doesn't raise their chances of escaping unless the other are equal to them or close. I decide who lives and who dies at best, at worst I decide who gets hatch. Yes the person who does nothing and plays for edc gate will gain mmr more consistently than the juicer who plays for chases, absolutely.
1
