7 months
ago -
Play New World
-
Direct link
Transcript (by Youtube)
0s | Players who have dyed all of their equipment yellow. |
---|---|
3s | If design wanted to expose that as an attribute to our matchmaker, then we could do that. |
8s | We're providing the flexibility for Game Design to be able to make cool experiences for the players. |
20s | Welcome to Forged in Aeternum, |
21s | where we talk about all things New World. |
23s | Today we're going to be talking about Service Improvements. |
26s | My name is Kyle Richter. |
28s | I am the Software Development Manager over at the Azoth Backend Team. |
31s | We have Matthew Woo, who is an Engineer on the Services Team. |
35s | Kenny Doyle, who is also an Engineer on the Services Team. |
38s | And as you all have seen before, Quynn Carter, QA Engineer for the Backend Teams. |
44s | So last time guys we went through a rundown of where Services came from and how we were approaching Services moving forward. |
55s | And we did touch on matchmaking a little bit, but I figured we could really kind of dive in deeper here, kind of see where Cross World OPR landed and the data that we got from that and start moving into, okay, well, how are we taking what we learn from that and expanding it and building it into what we're building for the future. |
76s | So, Matt, you you were pretty instrumental in the initial implementation of Cross World OPR. |
82s | What was the target there and how did we build that Cross World OPR matchmaking system? |
91s | And then we can kind of take it from there and go into how we've approach V2? |
97s | Sure. |
98s | Yeah, that's a good question. |
101s | Yeah. |
101s | Our first version for Cross World matchmaking was to get something in front of the players and see if this is a beneficial feature for you guys. |
113s | And OPR is a very fun activity to play together, and we know that it's difficult to find a group of 40 players to play with inside a single world and wanted people to be able to experience that more. |
131s | And so turning that into a Cross World feature. |
134s | And when I was designing it, we were designing it to, um, to launch us forward into more capabilities for Cross World with that in mind. |
148s | And so I think we've seen success with Cross World matchmaking and just continuing to move forward with that. |
156s | Absolutely. |
157s | And in our initial pass on matchmaking, we were just trying to get some of the infrastructure in place, right. |
163s | And build something that we could test against. |
165s | Mm hmm. |
166s | And so OPR is a good game mode for that, right? |
169s | Because it allows us to just fill teams kind of evenly. |
173s | Yeah, but that kind of leads us into the next side is once we've got that kind of baseline in place, now it becomes, okay, well, how do we start to form better matches, right. |
182s | Or more complex matches? |
185s | And Kenny, I figured you could talk a little bit about where we're kind of going moving forward with our version two of matchmaking. |
191s | Yeah, absolutely. |
195s | Well, one thing, it's been great to see the player feedback around OPR, so you know that people are able to find matches so much faster now. |
204s | So I think it's really exciting to think about how we're going to take that forward. |
208s | And yeah, with the matchmaking service, expand it out. |
213s | And yeah, so going forward, looking at bringing a rules engine into our matchmaking service. |
220s | So moving beyond the kind of OPR team filler technology and expanding out based on, you know, player attributes, group attributes, using those to get players into matches quickly, but that are either say more relevant to their interests or have a good team composition for taking on a particular challenge. |
248s | Yeah and I know we've been we spent quite a lot of time already building out this newer system. |
253s | We've got a number of rules that we're kind of experimenting with and just being able to do things like, you know, like we've talked about before, like role compositions and, and doing interesting things with that I think is, is a massive added benefit to players. |
270s | Quynn, we talked last time that like the process of testing the initial matchmaking system was already hard enough. |
278s | Right. |
279s | And and this system is quite a bit more complex. |
282s | So what are what are we looking at when we go into starting to test some of this stuff as we get closer to launch? |
289s | Well, I mean, like when we first went to the matchmaking testing on that, we first had to figure out, okay, where are the bounds? |
296s | Like, what can we test with? |
297s | Like, are we actually able to see the right number of players hitting the match? |
301s | Do we have the tools to fill up as expected and then follow up with some of the other features like the backfill; making sure that players are able to get into the match after it started and when it ends, can players get to the next match and so forth. |
313s | And then not to mention all that just tracking where they are in that process and at the beginning until we actually |
319s | understood exactly how they were landing. |
321s | It was a definite challenge understanding like where where they needed to be for us to track them, to know what we were looking for |
329s | for the success of the actual matchmaking system, much like we found a few problems in the early days of it. |
335s | So it's definitely a lot more improved. |
337s | Or at least now it's working, but definitely had trouble at that point. |
341s | Yeah, and one thing that we've been working on in the second iteration of matchmaking is to improve our automated testing, which Kenny has really helped with adding in into our unit tests, our integration tests. |
355s | And just to be able to run those every single time we're making changes |
359s | so that QA is not like, Oh, this thing's now broken all of a sudden, which we did have difficulty with. |
365s | Absolutely. |
366s | I mean, there were several playtests, right, where it just it just getting into a reasonable state to test through some of our our working paths was difficult. |
374s | But yeah, now we got the, you know, kind of sanity checks baked in and even adding some more of the esoteric test cases. |
383s | Yeah. |
384s | Just to make sure that we're in a good state and really comes down to not wasting Quynn's time. |
389s | Nobody wants to do that. |
390s | And we want to, you know, optimize around that as much as we can. |
394s | Absolutely. |
396s | So for people that may not be as familiar with some of the some of the the terms. |
404s | Right. |
405s | Like Rules Based System, what is, in terms of what players might think about, what is this really what kind of thing can we can we associate to players for understanding what a Rules Based Matchmaking System might might do? |
422s | Sure, Yeah. |
423s | So the main thing is it comes down to, let's say, configuration and providing our Game Designers with the tools that they need to be able to describe these different game modes with our rules engines. |
438s | So if we wanted a game mode with like, you know, varying team size, varying composition, could be other other attributes as well, if we wanted to say. |
452s | Not that not that we have this. |
454s | It's just an example, but it's like, oh, you know. |
458s | Players who have dyed all of their equipment yellow if they want to. |
461s | If design wanted to expose that as an attribute to our matchmaker and have us use that and form matches around all the these players who had dyed their their outfits, you know that we could do that. |
473s | We're providing the flexibility for Game Design to be able to make cool experiences for the players. |
479s | Now that's that's really interesting to to think about because like sometimes I know that players have brought up that it seems like sometimes it's it's slow for us to get around to making changes or it requires a new build. |
492s | And one of the major benefits here is that these things can kind of be modified on the fly, right? |
498s | And as Designers see what what is forming good matches and what isn't, right. |
504s | Now, if I can talk about something that Matt's been working on with standing up our rules control plane, it's exactly that so that Designers could be able to submit the rules changes kind of out of band or independently of a new build or a new service deployment and and more instantaneously have those new rules in place for matches. |
525s | And I think actually too, for QA, could be, you know, simpler in that near near soon to be reality to to have these like let's say you know maybe like a 2v2 OPR mode for a particular test and then be able to test it is legitimate matchmaking and go through the full legitimate flow. |
547s | But, you know, without having to use like a side band way of starting a match early, like to actually go through the full matchmaking experience. |
556s | Yeah, I think the big benefit is that, you know, by building a system like this, we get to change data, right? |
562s | As opposed to having to change code whenever these things want to be, want to be modified and that is that's a big area that we that we want to target moving forward as we extend and expand our our basis for how we build games, right, is to make sure that we're building systems that are extensible and don't require as much dedicated engineer time to to accomplish. |
585s | Yeah. |
585s | And I think there's there's so many people on the team who have worked on matchmaking systems in different games, and pretty much every single one of them is like, we're not going to get the rules right the first time. |
596s | And so we need that ability to iterate. |
599s | Yeah. |
600s | So Kyle has been good at pushing this. |
602s | Appreciate that. For sure. |
605s | We've we've had a couple of wins in some of these areas. |
608s | Quynn, it sounded like you were going to say something. |
609s | Well, I probably a question that might pop up with the community in general, and that's something that's kind of digging in my mind, especially using this describe is with the way that we're structuring the Role System and all that one aspect that people might either misconstrue or have in the forefront of the mind from other games would be this is going to be similar to scale based matchmaking of any sort. |
628s | And is that going to deride our ability to provide adequate gains for people to have both fun, but also the challenge? |
636s | But of course, I imagine that's more of a the gameplay side and not necessarily our server side. |
640s | Yeah, for us it's really about for us, it's really about just providing the technology that allows Game Teams to do these types of things, right. |
648s | And so that's, that's one of the benefits of these Rules Based Matchmaking Systems is that Game Engineers, Game Teams or Designers can take these tools and then use them in creative ways, right, to do different things. |
663s | That does actually bring up a point though, because our community |
667s | question last time was around what players would like to see as cross world experiences and a lot of players |
676s | I notice were bringing up Expeditions and Arenas which we do have on our roadmap and is currently targeted at Season Four. |
687s | But it does it does bring up an interesting idea of what types of things can we build. |
692s | So we're we're targeting Cross World Expeditions and Arenas coming up very soon. |
700s | But just having this technology here allows us to do some really, really neat things in the future. |
705s | And instead of having just a system built for exactly what we need, it does allow us to to expose some functionality that they can use in different ways. |
713s | Yeah, absolutely. |
715s | And like the attributes that I was discussing earlier, it's like who who's to say what Design chooses for those attributes and to feed into our service. |
725s | But, uh, don't blame Services then. |
727s | No, but we're definitely on that approach, right where it's like, yeah, we have that flexibility. |
733s | So if skill became a consideration that that's something that our service could take into account. |
740s | Yeah, absolutely. |
742s | So matchmaking is an interesting one. |
744s | We talked quite a bit about inventory service last time and, and some of the approaches that we're taking there to kind of, along the same lines, kind of expose these same types of extensible functionalities. |
756s | One of the other things that |
757s | I've noticed players really talking about a lot has been like queue times around releases and things like that. |
763s | Right. |
764s | And we we are also working on a set of, of, of improvements right now related to some of those things like players being stuck in really long queue times, allowing them to potentially like do some transfers for free or something like that to get them into a different state to distribute that, that that playerbase a little bit. |
788s | Um, I know none of you guys have worked specifically on that, but we've been involved in some conversations. |
795s | So what do you, what do you guys think about some of those areas that we're making improvements in. |
800s | Well, I think it's really cool, just like sort of the broad range of ideas that's being considered to really improve the player experience around these queue times. |
812s | And you know, if that's in terms of being able to open up more worlds as set of worlds starts to fill up or to be able to offer players the ability to transfer for free, that like, I think it's awesome that we're thinking of our players in that way and giving them kind of the ability to get to play our game fast. |
841s | Absolutely. |
842s | And I know we've talked about a number of ways to to accomplish that, but it can be a somewhat hard problem to solve. |
848s | Yeah, right. |
849s | Yeah. |
849s | Like you said, we've thought about a lot of different solutions, some that are bigger or some that are more feasible. |
857s | And it's interesting because, yeah, from the very beginning we've had ideas on how to solve this problem and just cool to see us trying different things. |
869s | Absolutely. |
871s | I think one of the challenges around that are like one of the interesting problems is just re rethinking how we think of like how active a particular world is and kind of going beyond just CCU. |
886s | Like how many players are actually like active and in that world at that time versus, you know, like historically how how popular has this world been? |
897s | And you know, does that factor into our thinking and like, you know, just because a world that has been very popular has a lower count of players at this time, maybe it still makes sense to open up worlds because we know, you know, peak hours do come back. |
914s | It's like a cycle here. |
916s | So anticipating around that. |
918s | Yeah, it's and we've also had several different attempts along the way of, or not attempts, but like marked improvements right. |
926s | Like we introduced not too long ago the ability for this like latent character idea. |
931s | Right. |
931s | Where if a player has been inactive for a certain period of time, we added some functionality that allows them to select a new world when they come back. |
941s | Right. |
942s | To again, help kind of distribute some of this. |
945s | And there are other other examples, too. |
947s | Last time we talked a little bit about Cross World Transfers and and some of these other mechanisms, we we have had some reports recently of, uh, latent character, uh, weirdness though, that we're, we're currently working on. |
964s | You want to talk a little bit about. |
965s | Yeah. |
965s | Oh going with a lot of live player reports? |
968s | Yeah. |
968s | A few players come into the fold as we run through the Game Service and they're hitting certain cases that obviously we can't really account for. |
976s | In some odd cases, like for example, the problem we had was that depending on the world state, like the world getting more popular over time or piece of our ebbs and flows of the player population, when we end up finding a point where the world has too many players and want to cut it off. |
993s | Unfortunately, there are some times where, hey, maybe a returning player comes in, here's that latent player. |
997s | They come in and they want to join back in. |
999s | They don't want to move again. |
1001s | That's the problem that we now came across, was like, Hey, we didn't necessarily think about this at the time. |
1005s | Yeah, partly shame on us. |
1007s | But again, that was one of those like common cases, where QA or Development as a whole, just to figure out, hey, what's all the cases we can figure out at this time without getting too, too lost in the sauce or again, trying to not lose out what we're trying to resolve. |
1022s | And we had an issue. |
1023s | We had added some world locking functionality like automatic world locking to get around some of those things that we talked about. |
1031s | But one of the downfalls to it is it's very like rigid, right? |
1035s | It hits a certain point and then it just automatically locks the world and kind of to what you are talking about. |
1040s | Kenny, some of the like starting to look at historical and active character trends starts to slow maybe some of those those locking and unlocking states. |
1053s | And so that's another example of how we're how we're improving things. This does bring us to our Community |
1058s | Question. All right. |
1060s | Yeah. |
1061s | So the last time we asked what were some of the things that you'd like to see coming up? |
1065s | And then you guys asked about things like Crossword Expeditions, Crossword Arenas, Wars, things like that. |
1071s | But a big question that comes into following with that would be: What were some of the pain points on finding groups for that content or without Cross world, as well as, what would you like to see as other means to help facilitate your grouping and social aspects for playing the game? |
1088s | So give us a note or a comment on that if you can |
1090s | Absolutely. |
1092s | And it's always, if you like the video, please subscribe. |
1096s | Helps with the algorithms and uh, we'll see you in Aeternum. |