devtrackers.gg
almost 2 years
ago -
Play New World
-
Direct link
Transcript (by Youtube)
| 0s | remember the old door problem the |
|---|---|
| 2s | designer goes to an engineer and says |
| 4s | just I want to put a door in and it's |
| 6s | like oh is the door open and close can |
| 8s | you go through the door is it open from |
| 10s | the left or does it open from the right |
| 11s | does it lock does it like when you start |
| 14s | going through the questions it literally |
| 16s | opens up I think there's like something |
| 17s | like 250 questions that could come out |
| 20s | if I want to put a simple door in the |
| 21s | game |
| 22s | [Applause] |
| 27s | hi everybody Welcome to forward to the |
| 29s | tournament where we're talking about All |
| 30s | Things New World I'm Scott I'm here with |
| 33s | a different Scott today one of our |
| 34s | seasoned lead engineers and I'm here |
| 37s | with Dave V our creative director and |
| 40s | today we're going to talk about why some |
| 41s | features and some implementation is |
| 43s | harder than others |
| 45s | um you know in my years of making games |
| 47s | I I learned early on that you you know |
| 49s | when you look at a feature you're like |
| 50s | hey why don't you just do this |
| 52s | the word just almost never works out and |
| 55s | we're going to talk about that with |
| 56s | Scott so Scott I'm going to just throw |
| 57s | it just at you okay why don't we just |
| 59s | put 20 000 players on every server 30 |
| 62s | 000. magic number that would solve so |
| 65s | many problems for cues and other things |
| 66s | but why don't we just do that well that |
| 68s | would be really fun right but uh we we |
| 70s | do have some uh technical hurdles that |
| 72s | we need we need to worry about so one of |
| 74s | the biggest parts of that is how much |
| 76s | data we're sending down to everyone's uh |
| 79s | client their their game and the more |
| 81s | information you need to send down the |
| 83s | more bandwidth that's going to consume |
| 84s | and it's going to have an effect on |
| 86s | client performance like we already see |
| 87s | that sometimes where you know some of |
| 89s | the elite chest runs can uh get in the |
| 92s | laggy situations with lots of players |
| 93s | lots of effects things going on at the |
| 94s | same time so that's sort of like one |
| 96s | aspect of scale that we need to control |
| 99s | um and and sometimes that's a technical |
| 100s | Challenge and also sometimes it's a |
| 102s | gameplay before we get to the gameplay |
| 105s | let me ask on the technical side |
| 107s | other games can do some of this so is it |
| 109s | what makes new world different I think |
| 112s | our uh our combat model |
| 115s | um really is uh very like reliant on uh |
| 121s | you know sort of picture-perfect uh like |
| 124s | swings and misses and things like that |
| 125s | you know when the when the skeleton dips |
| 127s | his head and you fire over him |
| 130s | um like you miss right like it's it's |
| 132s | not tab targeting |
| 134s | um so |
| 135s | that level of fidelity and server |
| 138s | simulation uh then needs to be sent back |
| 140s | to the client so that everyone you know |
| 142s | is on an even playing field and seeing |
| 144s | the same thing yeah and this is a |
| 145s | trade-off we talked about earlier when |
| 147s | we chose this model and then and then |
| 148s | but also 30 000 people while it sounds |
| 150s | great |
| 151s | I don't know Dave does that sound fun I |
| 153s | mean I think there are parts of it |
| 154s | that'd be super fun right but then I do |
| 156s | think you know some of the design of our |
| 158s | game would not work well with that a |
| 160s | couple just immediate examples are uh |
| 162s | you know we don't instance our |
| 163s | gatherables right and it's already |
| 165s | sometimes tough to get more calcium node |
| 167s | can you imagine now if there's 30 000 |
| 168s | people going for that node that's going |
| 170s | to be a challenge uh the other part is |
| 172s | the territory game right like there's |
| 174s | only x amount of territories to go after |
| 176s | uh and that's fun when there's two |
| 177s | thousand you know two and a half |
| 179s | thousand people but if there's now |
| 180s | suddenly thirty thousand people vying |
| 181s | for those territories uh you know even |
| 183s | less people are gonna get to participate |
| 186s | you know why don't we just put some |
| 187s | mannequins in the house and just get |
| 189s | gear set storage in in like a day or two |
| 190s | like that seems pretty easy to me so uh |
| 194s | there's a lot that goes into that uh so |
| 197s | for example |
| 199s | um one of the things we have to do with |
| 201s | all player state is you know store it so |
| 203s | so you get the same data from session to |
| 206s | session where you log out log back in |
| 208s | um you know that's a that's a finite |
| 209s | pool uh that we can sort of dip into and |
| 213s | the more items the more you know player |
| 215s | state that that we're trying to save |
| 218s | um like we start bumping up against |
| 219s | those budgets uh and that means that we |
| 223s | need to spend more time optimizing |
| 225s | things we need to spend more time kind |
| 227s | of reducing the size of other things to |
| 229s | make space for this this new information |
| 230s | a lot of what I do is like sort of |
| 233s | backend you know data like messaging |
| 235s | kind of stuff and I always tend to |
| 238s | forget about the UI aspect of it and so |
| 241s | when when you're saying like oh let's |
| 243s | just put it on a mannequin it's like |
| 244s | well how would we actually do that and |
| 245s | like you know the mocks and the UI |
| 248s | design much less the implementation and |
| 250s | all that is something that takes a lot |
| 252s | of time in addition to like you know |
| 254s | technically functionally we could get it |
| 256s | working I also know that like whenever |
| 257s | you're manipulating the inventory |
| 259s | putting things in and out of it you have |
| 260s | to be extra careful because we know |
| 262s | we've made some mistakes there in the |
| 264s | past that have led to some dupes and |
| 265s | other issues so we're going to take our |
| 266s | time with that and and that's sort of |
| 268s | something we've added on to all of our |
| 270s | features uh or all you know all the work |
| 272s | that we do is like make uh doing extra |
| 274s | you know threat model analysis uh |
| 277s | hardening of our code more uh testing |
| 280s | you know we now that we're we've been a |
| 282s | live game for a year |
| 284s | um we've definitely evolved from our you |
| 286s | know pre-launch like you know we want to |
| 288s | get as many features in we want to get |
| 289s | as much fun stuff in as possible |
| 292s | um and you know in order to hit those |
| 294s | things like you know we took shortcuts |
| 296s | in some places or calculated risks and |
| 297s | you know |
| 299s | let me be sure let me just be rest |
| 300s | assured we still want to get fun stuff |
| 302s | in yeah really fast but we're not not at |
| 305s | that yeah not at the risk of breaking |
| 307s | things that brings up an interesting |
| 308s | conversation with gear set storage right |
| 310s | like I think everyone wants to switch |
| 311s | their gear that's going to be awesome |
| 312s | but I think people might also want to |
| 314s | switch their attributes or what I'm |
| 316s | mastering with it right so that makes us |
| 318s | make a an interesting decision like do |
| 320s | we do the feature just with uh gear |
| 323s | first so we can get it out quicker or do |
| 325s | we hold it to do the whole thing |
| 326s | together and that's a that's a fun you |
| 328s | know discussion and decision we have to |
| 330s | make yeah and even when we talk about |
| 332s | doing the gear do we do all of it like |
| 334s | if persistence and storage is the big |
| 336s | problem maybe we can save some time and |
| 338s | some energy by not doing parts of the |
| 340s | gear that people may not be as |
| 341s | interested in switching all the time the |
| 343s | the persistent side and also like I |
| 345s | mentioned earlier like sending that more |
| 347s | data down to the client so you know if |
| 348s | you have hundreds of people on your |
| 350s | screen and everyone's changing their |
| 351s | gear constantly like that's a ton of |
| 353s | stuff data we're sending down and uh you |
| 356s | know the pipes can get clogged |
| 359s | what's another one what's another just |
| 360s | example Dave uh sometimes weapons is an |
| 363s | example right like why don't we just do |
| 364s | another weapon and I think uh you know |
| 367s | scale comes super into important when |
| 368s | you're when you're making weapons right |
| 370s | every time you have an effect or you |
| 371s | launch something it creates what called |
| 373s | gde so it spawns things and if all |
| 376s | people are doing this and weapons spawn |
| 378s | multiple things it impacts uh things |
| 381s | that can make we have we have throwing |
| 382s | hatchets why can't we have throwing |
| 384s | daggers |
| 385s | but but they also not just that there's |
| 388s | also balance issues that come into play |
| 389s | and and when you add a new weapon it |
| 391s | needs new animations there's a there's a |
| 393s | whole plethora of things that all have |
| 395s | to come together to get these done the |
| 396s | VFX and everything the rewards is |
| 398s | another one right like you know then we |
| 400s | want to have like with the gear sort |
| 401s | gear Greatsword when we introduce that |
| 403s | right like now we've got to go back and |
| 404s | add named items and different sets to |
| 406s | chase uh so it just it adds a lot of |
| 409s | work that people don't understand and |
| 410s | you don't just make one Greatsword |
| 412s | like as you can see there's so much |
| 413s | visual Variety in the great swords that |
| 415s | you see like just in rewards in |
| 417s | Brimstone before you even start chasing |
| 418s | all of the end game variety so yeah |
| 420s | there's there's a lot I mean |
| 423s | you know one of the things I always uh |
| 425s | tell my friends who aren't in business |
| 427s | when they say stuff's easy I always just |
| 428s | remember the old door problem |
| 430s | I always tell them to search the door |
| 432s | problem on the internet and it just kind |
| 433s | of says you know a designer goes to an |
| 435s | engineer and says just I want to put a |
| 437s | door in and it's like oh is the door |
| 439s | open and close can you go through the |
| 441s | door is it open from the left or does it |
| 443s | open from the right does it lock does it |
| 445s | you know like when you start going |
| 446s | through the questions it literally opens |
| 448s | up I think there's like something like |
| 450s | 250 questions that could come out if I |
| 452s | want to put a simple door in the game |
| 454s | and some of these things are just a |
| 456s | little bit more complicated than that |
| 457s | which which you know is why we're lucky |
| 459s | that people like you |
| 460s | that's where you were going to pull the |
| 461s | old Scott well you know what's sometimes |
| 463s | easy and quick aren't the same thing |
| 465s | right it's easy to count to a million |
| 467s | yeah yeah the most important metric is |
| 469s | not the difficulty it's time right now |
| 471s | sometimes things are super easy but they |
| 473s | take a long time especially |
| 475s | like when you want to make sure that it |
| 477s | gets out and it is at high quality and |
| 479s | feels right for the players and a lot of |
| 480s | the the foundational Tech that enables |
| 483s | us to move more quickly is are things |
| 486s | that take a long time and and you know |
| 488s | don't always have that immediate |
| 490s | customer impact and so you know we have |
| 492s | to prioritize you know getting things |
| 494s | out because people want to see cool new |
| 496s | stuff and we want to build cool new |
| 497s | stuff versus you know like building up |
| 500s | that Foundation yeah I mean we're |
| 501s | working on like right now we're working |
| 504s | on features that aren't going to go in |
| 506s | the game for probably a year because |
| 508s | we're laying that technical foundation |
| 509s | so when they come in |
| 511s | we don't break performance or |
| 513s | persistence or anything like that and |
| 515s | it's it's hard to understand unless you |
| 518s | are like in it every day asking all the |
| 520s | questions because because I like a lot |
| 522s | of the players I'm like no just get this |
| 523s | in the game I want it now I want it |
| 525s | today I'm like I'm every bit as uh |
| 527s | anxious to get stuff in as everyone else |
| 529s | because I play the game constantly so |
| 531s | and like you know that's one of the |
| 533s | great parts of prototyping is like you |
| 535s | know there we could do stuff very |
| 536s | quickly but like in a sort of controlled |
| 539s | thing that to get feedback on is it fun |
| 541s | is this the way we want things to work |
| 543s | but it's not shippable code it's not |
| 545s | it's not scalable you know so there's |
| 548s | also that trade-off all right well this |
| 550s | brings us to our community question |
| 552s | and that is you know with all this in |
| 555s | mind what one feature would you like us |
| 556s | to add to New World now keep in mind I |
| 558s | can't promise we'll do it but uh I can |
| 560s | promise we'll try so let us know what |
| 563s | you think in the comments below give us |
| 565s | a subscription if you like us because |
| 566s | that'll tell us to keep making the show |
| 568s | otherwise thanks and we'll see you next |
| 570s | time |
| 573s | [Music] |
| 574s | [Applause] |
| 580s | [Music] |
| 583s | [Applause] |