DesignDad
DesignDad
About
- Username
- DesignDad
- Joined
- Visits
- 0
- Last Active
- —
Comments
-
@sizzlingmario4 said:
https://forum.deadbydaylight.com/en/discussion/comment/2871635#Comment_2871635
I just realized, if Z-block no longer applies to healthy survivors, doesn't this mean it and Cat Block have the same effect? They both apply Hemorrhage for 60 seconds.
Such a good point that I went and double-checked my notes.
"Z" Block now applies Hemorrhage for 90 seconds. I need to update the patch notes draft! Thanks for asking and making me double-check!
-
@OniWantsYourMacaroni said:
https://forum.deadbydaylight.com/en/discussion/comment/2871635#Comment_2871635
Wait,does that mean that Sloppy will lose the mangled effect?
@GoodBoyKaru said:
https://forum.deadbydaylight.com/en/discussion/comment/2871635#Comment_2871635
Can I just check Sloppy Butcher keeps the Mangled status effect, and just becomes a very good anti-heal perk, right?
Sloppy Butcher still applies Mangled. Sorry for not being clear about that, I was only thinking about the things that changed and not what stayed the same!
-
I just finished writing our patch notes on this, what timing!
Sloppy Butcher is being changed to "Increases the rate at which healing progression is lost by 15/20/25%". Note that it's a % of a %, so instead of losing 7% per second, you'll lose 8.05/8.4/8.75% per second with these numbers.
As for the "Z" Block Add-on, it no longer causes survivors to bleed when healthy but still applies the new Hemorrhage as per normal.
As usual, these are the current values we're aiming for on the PTB, subject to change, etc.
-
@Valik said:
New system is good and gives the survivor something engaging to do instead of just stare at the screen and be upset.
The problem is the great skill checks.
These bad boys give something like 1% additional progress each. In just a few seconds, it's easy to pick up an extra 15% progress or so.
Having a friend or two tank hits can allow a wiggling survivor to gain 30% more progress over a number of seconds if they are good at the checks.
Why Behavior gave survivors GREATLY increased wiggle speed, but didn't increase the overall wiggle time at all is beyond me. If the skill checks were harder and the progress bar was increased by 20% or so, it would be a lot more balanced overall.
The great zones don't give a flat bonus, they increase the progression speed of the wiggle bar for a short duration. They are tuned such that if you hit every single great skill check during the wiggle, you will wiggle free 1 second faster (15 seconds instead of 16 seconds). If you miss any then it'll be less than a 1 second difference.
Good discussion in here, by the way. We're taking notes on the feedback from all parts of the community, so keep it coming!
-
@ScottJund said:
https://forum.deadbydaylight.com/en/discussion/comment/2777119#Comment_2777119
Simple question. You are going into great in depth responses so I want to see a simple answer. Why, assuming this system is somewhat functional as you claim, can a brand new survivor get killers with thousands of hours as their first game, in the middle of the day? This is not an outlier situation.
Simple answer: It can happen, but it is not intended.
As I mentioned in the Q&A: There's a major issue with the matchmaking system right now with how it handles backfilling lobbies. Specifically, it almost entirely disregards the MMR of the replacement player in favour of filling the empty spot as quickly as possible. This can and does result in inappropriate matchups, especially when it's the Killer who is replaced. This is not the fault of the MMR system itself, but rather the matchmaker not using the MMR value correctly.
To be clear: This is being actively worked on. No ETA on when it will be resolved, matchmaking systems are very complex.
There is another situation where this can occur, and it can be a touchy subject so please do not infer that I am talking about you or anyone specifically...
There are players who absolutely love playing DbD and who are not especially good at it from an MMR point of view. While there is a general correlation between time played and increased MMR, that is not a causal link. There are people with 1,000+ hours who happily live in the lower end of MMR. So long as these players are still "winning" about half the time against new players, then this isn't a problem. Rather, it's an intended feature of the system. They should be allowed to have fun, too!
All that in mind, I completely understand that this case is NOT what you mean when you ask "Can a brand new survivor get killers with thousands of hours as their first game?". The truthful answer to a literal interpretation of that question is "Yes", for the above reason.
However, I think a fair way to reword that to get closer to your intended meaning would be "Can a brand new survivor get killers who are extremely good as their first game?". And while the answer to this is still "Yes" right now, our intention is to eliminated these scenarios as much as possible. There are already a number of systems in place to prevent this: New players start at a very low MMR, currently I think it's around 900. The maximum search range cuts out a lot of the upper tier of players, currently I think it's +/- 400. This starts very low and increases up to 400 the longer you wait in the queue, so this also helps to reduce mismatches. (Note: These are dynamic values that we tune based on the current statistics, so don't presume they have always been this way or will always be this way.) These should cut out a lot of the mismatches, but unfortunately it is dramatically undermined by the Backfill issues I mentioned previously.
I hope that helps clear some things up!
-
Some good points in here! Having to do that 4 times is a bit much, especially against survivors who know how to play against Pyramid Head. It should be pretty easy to lower the number, we'll look into it! 🙂
-
@Mandy said:
https://forum.deadbydaylight.com/en/discussion/comment/2763315#Comment_2763315
...
of course Peanits knows who Otz is, that was kinda the joke of it.
Who the hell is "Peanits"?
-
First off, amazing post and follow ups. Thank you for taking the time out of your day to help break down these very complex design challenges into a more readable and understandable form!
PS – Love Trueskill, I remember reading papers about it many years ago. It solved some problems with matchmaking that are unique to videogames (ie building a fresh team for every game) in interesting ways.
@PlaysByShady said:
https://forum.deadbydaylight.com/en/discussion/comment/2768256#Comment_2768256
I hear what you're saying, and this boils down to a phenomenon known as "The Curse of Knowledge", in that people who are very familiar with a subject often forget the nuances they're implicitly aware of when explaining to people outside of the knowledge circle.
In other words, Patrick is a developer, and understands the background mechanics of what's going on. Because of the way the questions were posed, he only explained bits and pieces of the system at a time, rather than the whole, which made it harder for people without the background knowledge to understand what he meant. I'm a developer, and I've worked with this kind of stuff before, and I understood completely what he was saying. It all makes absolute sense. But unfortunately the general community didn't.
I also do believe there are some bad-faith actors who've deliberately stirred things up for sensationalism. His hockey analogy was absolutely correct... there's no need to count/measure shots on goal when you can just measure wins overall as an indicator of a person's skill. If they're winning, they're obviously scoring more, and if they're scoring more, they're obviously taking more shots. It's all inferred, so you can cut out the middle dynamics (he said this almost verbatim).
The strimmer I happened to see comment on this literally went on a rant of "hockey is a game of equal sides, people in different roles, blah blah blah" which has absolutely nothing to do with the analogy he was making. This is a straw-man!
What's a straw-man? It's where you oversimplify the point to such a degree that it bears no resemblance to the original, and you defeat the oversimplification. You might be handy with a rifle, and you can shoot straw-men in the field all day... but when you're up against real soldiers who are shooting back, that's a completely different point.
Reducing and dismissing the point of "hockey is a different" game is exactly the same... his point wasn't about hockey! His point was about not needing to focus on the individual inner-game mechanics when the outwardly result (i.e. win/loss) is a good enough indicator as to what happened. It's the same analogy regardless of sport... but too many people who didn't understand the analogy got fixated on the wrong point and made their rants based on that. And that's embarrassing.
It’s absolutely a challenge to communicated about this topic, as it is both extremely technical and complex. When doing a Q&A stream, I am trying to boil down years upon years of research and experience into a few sentences that are understandable, correct, and quick to deliver so that we can answer as many questions as possible. As much as I would love to talk to everyone for hours about Matchmaking, the mathematical nuance of Glicko-2, and how all of this needs to be modified to apply to an asymmetrical game… At some point I need to get back to working on the game. 😊
@RainehDaze said:
https://forum.deadbydaylight.com/en/discussion/comment/2768595#Comment_2768595
We, as humans, designed the maths. We know how the game works, we know what the game requires, and we're defining what a win is.
My point is that they decided to model MMR over a game as a series of 1v1 competitions and then adjust MMR based on the individual results there: i.e., modelling it as if the outcome per-Survivor is (mostly) a matter of the Survivor's actions on their lonesome, with the factor built in if someone else has died first.
The maths would not care and would still do the same abstraction of skill into a single number if BHVR had modelled it instead as a 1v4, with the Killer gaining MMR for 3 or 4 kills and losing MMR for 0 or 1, and the Survivors gaining an inverse amount on all four of them regardless of their individual outcomes.
It's a baffling design choice to model MMR in such a way when your game is not built around the idea of a single Survivor's skills being enough to secure their win condition, and it only necessarily needs one bit of bad luck for a single person to lose despite their overall team victory. It's why the sports analogy was flawed, because you don't suddenly demote goalkeepers in, say, professional football for letting in one goal right at the end if the team still won everything.
Furthermore, it means that if the MMR achieved an even result from people with even starting MMR, you're still changing everyone's MMR. If your system can get a perfectly accurate result and it will then make adjustments anyway, I'm questioning your update logic. Looking at it from the perspective of spending way too much time with neural net loss functions, 100% accuracy should mean a loss of 0 and thus no parameters change.
I’ll offer up a bit of clarification about this point: There is a “soft” cap for MMR as it is considered by the Matchmaking system. A player’s numerical MMR value is allowed to go as high (or as low) as the math says it should. However, when the Matchmaking system is looking at building matches from players in the queue it will effectively round down (or up) the extreme outliers to help them find a match. This was done in direct response to one of the MMR tests we ran last year: Some players were completely unable to find a match because they were such extreme outliers, no matches could actually be formed with them involved. The soft cap is a dynamic value that we can change at any time (and we have adjusted a few times already). Right now I think it’s around 2,000, so any players who have an MMR above that value are treated by the Matchmaker as though their MMR is 2,000 for the purposes of building a match. Post-match MMR adjustment is still based on their ACTUAL MMR, though.
It's worth noting that this isn’t done because there are tons of players above the cap, it’s actually the opposite! The vast majority of our players fall within our lower and upper soft caps. The caps exist to handle the rare, but still real, edge case of extreme outliers. They deserve to play the game, too!
@Tranquil_Blue said:
One additional point: they said there's a soft cap for MMR, so there are obviously many players with ratings way above the soft cap. Those players are definitely playing matches at the highest possible MMR, are they not? I agree that no one should be making claims about their own MMR (since it's hidden), but I'm saying that technically there is a highest possible MMR (for matchmaking purposes).
Sadly, in practice, it’s much more similar to the worst-case scenario. It still weighs MMR a bit when backfilling, but not in a meaningful manner. This is partially a hold-over from the Rank-Based Matchmaking days. But, as mentioned, it has been identified and is something we are actively working on addressing!
@PlaysByShady said:
https://forum.deadbydaylight.com/en/discussion/comment/2768256#Comment_2768256
I hear what you're saying, and this boils down to a phenomenon known as "The Curse of Knowledge", in that people who are very familiar with a subject often forget the nuances they're implicitly aware of when explaining to people outside of the knowledge circle.
In other words, Patrick is a developer, and understands the background mechanics of what's going on. Because of the way the questions were posed, he only explained bits and pieces of the system at a time, rather than the whole, which made it harder for people without the background knowledge to understand what he meant. I'm a developer, and I've worked with this kind of stuff before, and I understood completely what he was saying. It all makes absolute sense. But unfortunately the general community didn't.
I also do believe there are some bad-faith actors who've deliberately stirred things up for sensationalism. His hockey analogy was absolutely correct... there's no need to count/measure shots on goal when you can just measure wins overall as an indicator of a person's skill. If they're winning, they're obviously scoring more, and if they're scoring more, they're obviously taking more shots. It's all inferred, so you can cut out the middle dynamics (he said this almost verbatim).
The strimmer I happened to see comment on this literally went on a rant of "hockey is a game of equal sides, people in different roles, blah blah blah" which has absolutely nothing to do with the analogy he was making. This is a straw-man!
What's a straw-man? It's where you oversimplify the point to such a degree that it bears no resemblance to the original, and you defeat the oversimplification. You might be handy with a rifle, and you can shoot straw-men in the field all day... but when you're up against real soldiers who are shooting back, that's a completely different point.
Reducing and dismissing the point of "hockey is a different" game is exactly the same... his point wasn't about hockey! His point was about not needing to focus on the individual inner-game mechanics when the outwardly result (i.e. win/loss) is a good enough indicator as to what happened. It's the same analogy regardless of sport... but too many people who didn't understand the analogy got fixated on the wrong point and made their rants based on that. And that's embarrassing.
Improving the algorithm to take into account team results is also underway. Originally, we didn’t pursue this approach because Survivors can leave a game and start searching for their next one before the previous game had completed, meaning there were no results available for updating their MMR. This is one area where the solo-result system has an advantage, as soon as you are done in a game your MMR can be updated.
In reality, the MMR adjustment from a single game is small enough that you likely won’t notice a difference in opponent from match to match as MMR adjustments are meant to show impact over many games, not just 1 or 2. Our hope is that we can hybridize the system and use both team results and personal results to generate an even more accurate MMR that contains the benefits of both approaches!
As another note about the various posts talking about the 1v4 aspect of DbD as it pertains to the maths:
Our batch update logic (that’s what updating with multiple results simultaneously) is based on the Glicko algorithm. It (Glicko) was designed to handle situations where you’ve got multiple results in a short period of time where there’s unlikely to be much difference in the skill of the player between those games. To put it another way, think about someone going and playing in a one-day tournament: They win some and they lose some, but their skill level at the game was probably the same the entire time. In DbD, we use this for Killers to process the results of a single game: 4 matchups that all happened simultaneously are “batched” together to make use of this approach.
If you’re more curious about the maths, I encourage you to check out Mark Glickman’s website about it! He provides papers explaining the formula in detail. A quick search for “Glicko” on your search engine of choice should lead you there.
We have made many customizations to the formula to reflect the asymmetric nature of DbD, but the core mathematical concepts remain the same. 😊
-
I am consistently impressed by your thoroughness, accuracy, and speed with these writeups.
Jolly good show!
-
@Edilibs said:
https://forum.deadbydaylight.com/en/discussion/comment/2681187#Comment_2681187
I dont get what your saying..Depiping lowers your grade and you definetly go from Red 4 to the previous grade by losing ie depiping ie when your constantly losing. This MMR thing is confusing!
Hey friend! Maybe I can help clear things up:
Grades are not used by the matchmaking system. They used to be called "Ranks" and were used for matchmaking, but that is no longer the case. Grades serve as a monthly tracker for how much time you spend playing, based on the Emblem system (Emblems turn into pips, and pips turn into Grades). You can't lose a grade through de-pipping anymore, either! At the Grade Reset on the 13th of each month, everyone goes back down to Ash IV and starts the climb again, and they get a Bloodpoint reward for their previous Grade. Get your Grade higher, and you get a bigger reward!
MMR is an entirely separate value used for Matchmaking. It's hidden, meaning you don't get to see it. It has absolutely no connection to Grades: Your MMR doesn't directly influence your Grade, and your Grade doesn't directly influence your MMR. MMR doesn't reset each month, but it is constantly adjusting based on match results to try to find you balanced games. This is why it's called SBMM (Skill Based Match Making), it's trying to measure your skill and put you in matches with similarly skilled teammates / opponents.
We've been considering removing the Grade icons from the post-game screen due to this confusion. I was hoping we could leave them there to allow everyone to have a little way to show off / have some bragging rights, but I've seen lots of people confusing Grades with MMR (understandably so!). What do you think? Should we hide the Grade icons on the match summary?
Thanks for taking the time to ask your question! I hope this explanation has helped clear up some of your confusion so you can get back to enjoying the game. <3
-
@GoodBoyKaru said:
https://forum.deadbydaylight.com/en/discussion/comment/2670844#Comment_2670844
If you wouldn't mind me asking, how does disconnecting impact your MMR? I've had games- unfortunately, a lot of them- where players have curbstomped me (these being incredibly good players that I woulnd't have stood a chance against without playing my mains) however at the end all 4 have DC'd. Would this raise my MMR and lower theirs? Freeze everyone's?
How does it work?
Disclaimer: I'll avoid giving specific numbers here as an anti-exploit measure. If I say the exact values, one of you clever folks will figure out a way to ride the line and do something cute with it!
If a DC happens early enough in a match, then the result of that matchup is discarded. So if one survivor DCs right away, then that survivor's MMR is left alone, and the Killer's MMR isn't adjusted for that survivor either. It can still be adjusted based on the remaining three, depending on how the rest of the match goes (these are weighted so that if the survivors die it doesn't impact MMR as much since they were at a disadvantage).
Later in the match, though, the adjustment goes through. This is so that you can't just DC before getting hooked to "skip" the adjustment. And likewise, if the Killer DCs when the gates open, that's treated like all 4 survivors escaped in terms of MMR adjustment. Rage quitting is the most common reason to disconnect in these scenarios, so that's what we're targeting with this logic.
Part of this also to make smurfing more time consuming: You can't just join games and immediately DC to drop your MMR. You gotta hang around for a while, which makes intentionally tanking your MMR much more of a slog and generally time consuming. Forcing people to hang out in a match also makes it more likely that they'll actually end up playing it normally, too! And then you add onto all of this the DC penalties, which increase with each disconnection, and suddenly those who disconnect are spending a lot of time for very little influence over the system.
Thanks for asking a great question! DC logic is tricky business, and we're always working on new ways to improve it to handle more extreme cases, like the one you mentioned.
-
@StarLost said:
https://forum.deadbydaylight.com/en/discussion/comment/2670556#Comment_2670556
Thanks!
I think I may be somewhat unique in this one - but I've definitely noticed an overall improvement in match quality. I am sympathetic to the complaints of high MMR players who seem to be struggling at the moment though.
A few questions, while I have you here (Dad to Dad :) )
- I understand the 'find you a match more quickly' but most of my crazier matchups have almost instant queues. Is this due to a 4man sitting in queue for ages and me just drawing the short straw?
- I've seen a fair few players (in their words) 'managing' their MMR, generally by suiciding at the end of the match. Are there any plans to help prevent/discourage smurfing?
- What would be the definition of a 'good match'? Because at least 1 match in 5 throws me up against either rather new players (don't worry, I generally ease off them once I notice) or against monsters who...sadly seldom ease off on me.
- I, personally, would not mind my wait times increasing by 50% to ensure closer matches. It's no fun to just roll over people or have to 'dial it back' and it's no fun to get pulverized by a nightmare squad. Isn't waiting in a longer queue and knowing that it's going to be an even-ish game better than waiting in a long queue and having no idea what I'm going up against?
- Are there any plants to make MMR/enemy MMR/gains and losses visible in postgame? This would allow us to find answers to so many hotly debated issues regarding MMR.
- Are there any plans to 'inflate' MMR when people play as a group (something of an industry standard) - for instance, applying a temporary, proportional multiplier to SWF groups - say 1.1x for a 2man, 1.3 for a 3man and 1.6 for a full SWF. This would go a long way to offsetting the enormous advantages coordination and information sharing provide, without 'punishing' folks for playing with their mates.
- Lastly, are there periods where the SBMM system is just entirely offline? I'll have patches characterized by near instant queues and absolutely bonkers matchups that really do make me wonder...
That all said, have a fantastic day and I appreciate the response.
PS: Buff Pig! (couldn't resist).
I'll answer what I can so we can both go back to building decks / falling asleep while watching The Big Game. I'm mostly focused on answering questions about the existing system, and not going into too much details about potential future changes.
- For those insta-queues that feel unbalanced, you're correct! Imagine the situation I was describing before where you are waiting and waiting for someone to show up in the queue that's within your range, but from the other side. A group of survivors is waiting (and to be clear, they might not be a full SWF, just that they are grouped together by the MM system), and then you pop into the queue: An opponent who is near the edge but within the boundary for their search. "Wonderful!" the MM system thinks, "I can finally give them a match. It might not be the most balanced, but hey, they're playing after a long wait!".
- We actually had a system in place to prevent smurfing / intentionally dropping your MMR before, and it ended up causing more problems than it helped. That's not to say that it's off the table forever, but rather that it's a really tricky problem to solve in a way that works across the board. it's not at the top of the list right now, as there are other issues that we can tackle with bigger payoff in terms of matchmaking improvements. Prioritization!
- I could talk for hours about defining a "good match", but I'll do my best to summarize: What the MMR system does is predict the outcome of a match before it starts. More specifically, it predicts the odds of the outcomes. With a large enough sample size, it's actually really good at this! If you look at all the games where the system says "PlayerA has a 60% chance of escaping this match", PlayerA does escape those matches about 60% of the time. Generally speaking (and I'm definitely simplifying here, so don't take this too literally) we operate under the assumption that a 50% escape chance per player is indicative of a balanced match. This holds true in most games outside of DbD, be it Chess or [Insert Popular FPS Franchise Here]. If both players are equally as skilled, then the odds of either one winning are about equal. So if you can make a match where the prediction is that the odds are equal, then you've likely got a balanced match!
- Some people are willing to wait 30+ minutes for a perfect match, and others don't want to wait more than 5 no matter what. So we have to find a decent compromise, which is extremely difficult. Remember that the MM system isn't just trying to find you good matches, it's trying to find good matches for every single player!
- I'm not going to go into long-term plans here but I will say that we have consciously decided not to show explicit MMR values for now, as doing this generally result in a lot of toxicity for other online games.
- I'm very familiar with the technique of managing group MMR, and it's something that we have also considered. For a while, I was thinking about maintaining a completely independent MMR for solo queue and group queue, which is something I have seen other games do. That said, several of them have moved away from that technique as it didn't actually provide much benefit for matchmaking, but did result in a fair bit of technical overhead. Overall, we're still open to doing something like this once we feel like we've got a good solution and that it's going to be a meaningful improvement, but there are bigger fish to fry for the moment. As it stands, we're taking the average MMR of the group and using that for matchmaking, which works fairly well at a high level. We'll likely revisit this after we've made other improvements elsewhere.
- SBMM never goes offline. It sees you when you're killing, it knows when you escape. It knows when games are bad or good, so play good for goodness sake! But for real, the MM system learns by doing. When you get matched in something that feels crazy unbalanced, the system is paying attention and adjusting both your MMR and your opponent's MMR accordingly. There are situations where the system can be very unsure about where a player should fit, and so it tosses them into games to calibrate their MMR. When you do happen to end up in a game against these folks, keep in mind that your MMR is likely not being adjusted that much by the result.
I hope that helps clear a few things up. Overall, the biggest challenge of building a matchmaking system is that it has to produce thousands and thousands of matches every day, so we have to take the big picture AND the zoomed in view into account. For every discussion we have about a graph showing all games in the last month, we also dig into specific examples to see how they are panning out, too.
Thanks for taking the time to ask questions and listen! <3
-
@StarLost said:
https://forum.deadbydaylight.com/en/discussion/comment/2669806#Comment_2669806
Yeah it's a tough one.
This one was a 'wat' moment though, because there is literally no way that I was the closest killer to their MMR at peak.
https://forum.deadbydaylight.com/en/discussion/comment/2669808#Comment_2669808
I've got some suspicions regarding the SBMM.
- The 'floors' and 'ceilings' do not apply when someone dodges a lobby, and it just grabs whoever is available.
- The system doesn't update correctly during some periods (I know if I go on a losing streak at night, I'll generally have a winning streak the next evening.
- People are still finding a way to smurf.
- There's a damn hack that can fiddle with your MMR.
- It has some sort of daily 'maintenance' period, where it's totally offline.
Would really appreciate some more transparency from BHVR on this stuff.
As a general rule, the matchmaking system is doing the best it can with what it has to work with. By that, I mean that the limiting factor in off-peak hours is typically the players in the queue. The system is trying to match you with somebody within a pretty tight range of skill at first, but there are times where there simply aren't any other players in the queue within that range for you to match with at the moment. When this happens, the system starts expanding the acceptable ranges over time.
The basic trade off is "I can find you a match more quickly, but it might not be as balanced". We can keep the range super tight to try to enforce extremely balanced matches, but those result in pretty crazy wait times. We could also start with the range extremely open to shorten the wait times, but now the matches are all extremely unbalanced. That said, we adjust this system on the fly from time to time and take a look at how it impacts both wait times and match outcomes overall.
So most cases of "It's not updating!" or "It seems to stop working at certain times of day" are actually cases of "There just weren't any good matches in the queue for you right now". The system has no downtime, and we have yet to see any cases of improper updates (and we do check!).
With all this in mind, we are far from calling the matchmaking system perfect or done. We're still monitoring it for bugs, exploits, hiccups, and also planning general improvements. The community is a dynamic group, the game is constantly being updated, and our matchmaking is no different! One of your suspicions above is absolutely spot on, and is something we are actively looking to address. I'm reviewing a design proposal for it later today!
Love y'all talking about matchmaking so much. <3