over 1 year
ago -
PlayOverwatch
-
Direct link
Transcript (by Youtube)
0s | foreign |
---|---|
4s | [Music] |
14s | [Music] |
25s | [Music] |
39s | [Music] |
43s | foreign |
46s | [Music] |
71s | thank you |
73s | [Music] |
81s | [Music] |
91s | foreign |
95s | [Music] |
101s | [Music] |
122s | [Applause] |
131s | [Music] |
148s | [Music] |
154s | [Music] |
156s | foreign |
157s | [Music] |
191s | thank you |
199s | [Music] |
208s | [Music] |
214s | [Music] |
217s | foreign |
219s | [Music] |
239s | [Music] |
247s | [Music] |
259s | foreign |
263s | [Music] |
271s | [Music] |
284s | [Music] |
289s | foreign |
330s | thank you |
337s | [Music] |
360s | foreign |
387s | [Music] |
394s | foreign |
399s | [Music] |
428s | thank you |
434s | [Music] |
441s | thank you |
501s | thank you |
503s | [Music] |
522s | thank you |
561s | thank you |
563s | hey everyone welcome to the OverWatch 2 |
566s | Dev chat we have a stacked lineup of |
569s | developers and Leadership from Team Four |
571s | here to talk about us about overwatch's |
573s | first anime collaboration ever one punch |
576s | man |
577s | I'm Jody from the OverWatch team and I'm |
579s | joined by my epic host Duo cusoe and |
582s | game director Aaron would y'all like to |
584s | introduce yourselves |
586s | yeah sure uh thank you so much Jody I am |
589s | so thrilled to be here with you today |
591s | and pick the developer's brain about |
593s | this super cool collaboration which uh |
596s | that that one took me by surprise it's |
598s | not gonna lie when I started popping up |
600s | on my time and I was like excuse me what |
602s | now yeah so I can't wait to dig into |
604s | this |
605s | yeah I can't wait to also this is a |
608s | pretty exciting moment to talk about |
610s | it's like a pretty exciting moment for |
612s | the team to be able to work with one |
614s | punch man |
615s | that's awesome |
619s | the OverWatch one punch man |
621s | collaboration uh so you can jump into |
623s | the game right now and experience it and |
626s | you know I think our trailer really said |
628s | it best so in case you missed the |
631s | YouTube Premiere let's have a look at |
633s | the OverWatch one punch Man trailer |
644s | thank you |
652s | [Music] |
661s | ready for a knockouts |
680s | la la |
696s | now that was a serious punch |
699s | well that was great and I'm so excited |
702s | to kick off the discussion and to dive |
704s | in a little deeper uh to where this |
707s | collaboration originated so Erin can you |
710s | walk us through kind of the original |
712s | idea for this collaboration with one |
714s | punch yeah I'd love to I mean |
718s | so the the whole team the OverWatch team |
720s | we we love our game we we love the |
723s | universe of OverWatch and and telling |
725s | the stories and showing the |
727s | personalities of the the different |
728s | Heroes that we have and the locations |
730s | that you go to but the OverWatch team |
733s | doesn't just love OverWatch where where |
735s | Geeks at heart there's all sorts of |
737s | other fictional worlds that we've fallen |
740s | in love with and one punch man it's at |
742s | the top of the the list for us and so we |
746s | we want to be able to celebrate other |
749s | other universes and other characters in |
752s | other worlds and for us like one of the |
755s | best ways to do that is to be able to |
757s | bring them in to our own world and and |
761s | um play with the characters that we love |
764s | from from um from one punch man |
767s | um and the the idea is when we started |
769s | talking about doing other collaborations |
771s | with um with other fictional universes |
774s | and we kind of started zeroing on on |
777s | anime one punch man and was just at the |
779s | top of the list because it's it's |
782s | um |
782s | it's so endearing to watch and the |
785s | characters feel like they're they're |
788s | their their own version of sort of |
789s | Larger than Life real characters just |
791s | like we have in OverWatch so it's just |
793s | like a really great natural fit for us |
796s | I I'd love to know though where this |
799s | kind of shift in The Stance is coming |
801s | from because historically uh the over |
803s | team was more or less I don't want to |
805s | say against bringing in other characters |
807s | or other universes but there's |
808s | definitely been pushback on that why why |
810s | does something change |
812s | yeah there there has been a change it |
815s | hasn't been overnight it's been it's |
817s | been a gradual change for the team and |
820s | when when we first released OverWatch we |
822s | wanted to be very protective of our |
824s | heroes and our and our universe and we |
827s | wanted to keep |
829s | um we wanted to give it as much space as |
831s | possible so that people could really |
833s | understand what our world was like this |
836s | like amazing inspirational near future |
840s | Earth with all these really cool heroes |
842s | in it we wanted people to understand who |
844s | all those Heroes were but over time we |
847s | started loosening that even the Skins we |
849s | started making for those Heroes we |
851s | started pushing them more we eventually |
854s | started doing skins of other blizzard |
859s | um um universes we even did a Bastion |
861s | skin a Lego Bastion skin at one point so |
865s | we started opening up once we felt like |
868s | the world of OverWatch was established |
870s | enough and then if you if you look at |
873s | the gaming landscape there's there's a |
875s | lot of collaborations happening out |
877s | there and it was really interesting for |
880s | us to watch some of that happening |
883s | because it let us analyze what we felt |
887s | comfortable with and what we wouldn't |
888s | feel comfortable with that other games |
890s | were doing so there was this great |
892s | lesson there was there is collabs out |
895s | there that that I could get really |
896s | excited about in other games and we |
898s | started to look at that and say like hey |
900s | we want some of that in OverWatch also |
902s | we want people to be able to look at our |
904s | game and see other really cool things in |
907s | it and be able to relate to those and |
909s | celebrate celebrate those other |
911s | universes yeah I mean I think the |
914s | reaction from the community speaks |
916s | volumes of just how excited everyone was |
918s | to see one punch man popping off in |
921s | OverWatch uh so I'm wondering what that |
923s | means for one punchman now what does it |
925s | mean for one punishment to exist in the |
928s | OverWatch Universe yeah okay so I love |
931s | one punch man watch the series like |
933s | multiple times and so like first and |
936s | foremost we just want to like celebrate |
938s | something that we love inside of our |
941s | game you know it's just like it's just |
943s | like pretty amazing the characters in |
945s | that world that they've developed are |
947s | are pretty amazing |
949s | um and so so to do it we we have like |
952s | our own take on what one punch man looks |
955s | like inside of OverWatch and it's this |
957s | take these values that we have will |
960s | probably bring these forward into the |
962s | future as we start looking at other |
965s | collaborations that that we could be |
966s | doing but first and foremost is is |
969s | OverWatch it's a it's a pretty |
971s | competitive very fast-paced game and so |
974s | at any moment you need to be able to |
977s | look at look at |
979s | um the screen in front of you and be |
982s | able to tell in a split second what your |
985s | team is doing what the enemy team is |
987s | doing who they're composed of and all of |
990s | the the silver wet the animation the VFX |
994s | the audio effects of all of our heroes |
996s | are a huge part in players being able to |
1000s | translate that visual information so |
1003s | even though we are dressing up |
1006s | um doomfist let's say as saitama we want |
1009s | people playing the game to be able to |
1012s | instantly say that's doomfist over there |
1014s | so it it's still using doomfist voice |
1018s | actor it's still using |
1020s | um the same visual effects or mostly the |
1022s | same visual effects but everyone will be |
1024s | able to look at it and be like this is |
1026s | doomfist running around and on top of |
1029s | that we we still think of the OverWatch |
1032s | universe as a cohesive space it it has |
1035s | its its own history its own lower its |
1039s | own characters and we want to preserve |
1041s | that as much as we can so as we bring |
1044s | the one punch Man characters into |
1047s | OverWatch |
1048s | um we're doing it almost as as costumes |
1052s | that it really are just skins for the |
1055s | heroes that we have but at the end of |
1057s | the day we want we want to be able to |
1059s | represent and celebrate the characters |
1062s | from one punch man but not to replace |
1065s | our own |
1067s | yeah absolutely it's it's kind of like |
1069s | when we say like doomfist doomfist is a |
1072s | fan of one punch man he's a fan of |
1074s | saitama and US putting that in the game |
1076s | is a reflection of us and like the |
1079s | OverWatch team like we're really big |
1081s | fans of one punch as well |
1083s | yeah you never suspect that doomfist |
1086s | would be such a fan of such a quirky |
1088s | character yeah |
1091s | this is really exciting though and you |
1094s | know what what does this really mean for |
1096s | OverWatch and like for the possibility |
1099s | of baby future collaborations oh okay |
1103s | um so I mean when we when we launched |
1106s | OverWatch 2 and we put as much effort as |
1110s | we could into supporting the game we've |
1112s | talked a lot about how we've grown the |
1114s | team a lot and reorganized the team and |
1116s | we have this this focus of trying to |
1120s | always be evolving the game to best suit |
1123s | like the needs and desires of the people |
1125s | playing it we we just want to always be |
1128s | making more OverWatch for people and and |
1131s | for people to see what we're coming out |
1132s | with and to be excited for it um that's |
1136s | what this the one punch man |
1138s | collaboration is about and and we can |
1140s | see |
1141s | um online that so far it feels like |
1143s | people have had like a really positive |
1146s | reaction to it and they're really |
1148s | excited about what we're doing and we |
1149s | even see them put putting out ideas of |
1153s | other collaborations we can do and |
1154s | matching Our Heroes up to characters and |
1156s | other universes and so I think if |
1159s | something like this goes well it will |
1161s | just give us the confidence that this is |
1163s | what players want from us and this is |
1165s | the sort of thing that we ought to we |
1167s | ought to keep doing I'm not sure if |
1169s | we'll ever get to the point where this |
1170s | is like something we do every single |
1173s | season we still want |
1175s | um we still want to push the the |
1178s | narrative in the world of OverWatch |
1180s | forward all on its own but this this |
1182s | could become a regular part of OverWatch |
1186s | in the future |
1188s | I mean I wouldn't want that to happen |
1190s | every season yeah like it should be a |
1192s | special treat we're getting with it I |
1195s | like the way you put that there you go |
1196s | that's that's going to go into the |
1198s | values it's a special perfect |
1201s | it's a it's a really cool way to show |
1203s | that like we're fans of the things that |
1205s | our community are fans of and you know |
1208s | we like to geek out over things too and |
1211s | one punch is really special |
1213s | relatable we loved it |
1217s | uh I'm wondering Erin for you or or your |
1220s | team was there like a really big fan boy |
1222s | or |
1223s | fangirlman stared at Cola |
1226s | okay so |
1228s | um I love the show there's a lot of |
1230s | people on the team that love the show |
1231s | and as we started talking about what it |
1234s | would mean to translate the show into |
1237s | OverWatch we had multiple brainstorming |
1239s | sessions over it and so that's like one |
1242s | of the funnest parts of this is just |
1244s | talking about |
1246s | um what what different characters we |
1248s | could use you know like tornado kariko |
1251s | like all of that stuff is really fun but |
1254s | there are so many different things we |
1255s | talked about that didn't make it you |
1258s | know like hey are we gonna be able to |
1259s | use Garros or sea king or is there some |
1262s | weird like censored version of like Puri |
1265s | Puri Reinhardt out there you know it's |
1267s | just like crazy stuff that we gotta that |
1270s | we gotta kind of like talk about and |
1272s | have fun with but um I think that once |
1274s | we started actually implementing a lot |
1277s | of this stuff there was there's two |
1278s | moments the the first was when they |
1281s | started getting the lobby together of |
1283s | just that sort of end joke of training |
1287s | with Hunter push-ups 100 squats you know |
1289s | the 10 kilometer run like it's a pretty |
1292s | amazing Lobby we've never really had |
1293s | something like that before but my |
1296s | favorite thing was the first time I saw |
1298s | the Justice crash highlight intro it's |
1301s | like such an amazing highlight intro and |
1304s | the funny thing is earlier in season |
1306s | three we released hanzo's Cupid kiss |
1309s | highlight intro which I just thought we |
1311s | would never be able to top right like |
1314s | that that is like s tier highlight intro |
1316s | it's so cool and then Justice crash |
1319s | comes along and it's like oh my God |
1320s | which one of these is better than the |
1322s | other so yeah it's just been fun like |
1325s | all the way around working on it |
1328s | it's so good soldier is movement writer |
1330s | makes me so happy like those those two |
1332s | are actually a match made in heaven |
1335s | yeah I love it yeah |
1338s | um not to speak about the balance of |
1341s | soldier 76 or anything like that you |
1344s | know maybe they're not a match made in |
1346s | heaven on that front but yeah maybe not |
1348s | so for the next balance practice all |
1351s | abilities |
1355s | exactly his new uh alternate ability is |
1359s | like bloody nose or something like that |
1361s | so there's a guy who helps out on the |
1364s | sidelines Ren that's it I actually love |
1367s | Moomin writer he okay this is just my |
1369s | own take but he feels like the real hero |
1372s | of that show so like like saitama has |
1374s | like these massive amazing abilities a |
1378s | movement Rider doesn't really have any |
1379s | but he's still like first on the scene |
1382s | he's always responding he's like the |
1384s | person that's really putting their life |
1386s | on the line to help other people even in |
1388s | like the face of like certain |
1391s | uh defeat you know yeah no he's showing |
1395s | up with courage uh so I I respect that |
1398s | not necessarily with any abilities or |
1400s | skills which that's unfortunate but a |
1403s | forever he's answering the call the |
1406s | world could always use more hero it |
1408s | really is I love that |
1411s | I'm so happy that we get to share this |
1414s | with players now and you know it was |
1416s | really incredible watching everything uh |
1419s | go from like that idea the brainstorming |
1422s | sessions and like becoming a reality and |
1424s | finally making it to the live game so |
1426s | everyone |
1428s | we're super amped we're super amph this |
1430s | is finally in game this collaboration's |
1432s | finally live so yep I I am too like |
1435s | cannot wait for to see people's |
1437s | reactions to all the things that we have |
1439s | in the game it's really exciting time |
1441s | for us |
1442s | absolutely well thank you Aaron thank |
1445s | you so much for joining us and giving us |
1446s | more of an insight into the |
1448s | collaboration |
1449s | um yeah this is a legendary moment for |
1452s | gym for one punch so we we really |
1454s | appreciate you being here to answer our |
1455s | questions yeah of course thank you for |
1457s | having me it was great yeah awesome |
1459s | thanks Aaron |
1461s | bye |
1464s | Zoe and I are joined but two of my |
1466s | incredible teammates to dive a little |
1468s | deeper into the development process of |
1470s | the one punch man collaboration so I'd |
1473s | love for you guys to introduce to |
1475s | yourselves and kind of give us a little |
1476s | insight onto your role in on this team |
1479s | uh do you mind kicking us off Dion yeah |
1482s | hey I'm Dion Rogers I'm the art director |
1484s | for OverWatch 2. uh my name is rakan |
1486s | kamash I'm a senior character artist on |
1488s | the OverWatch team I'm responsible for |
1489s | creating characters and skins that you |
1491s | guys see in the game awesome |
1493s | I'm so excited to be able to pick both |
1496s | of your brains today about this |
1498s | particular and collaboration so I want |
1500s | to run back to the very beginning of it |
1502s | and I want to hear from you why did we |
1506s | end up with those specific characters |
1508s | why those specific characters not just |
1510s | from one punchman but also from the |
1512s | OverWatch universe |
1514s | yeah when we first started to think |
1516s | about okay we're ready to do a |
1518s | collaboration with an uh an anime one |
1521s | punch man was a pretty big one for the |
1524s | team everyone was super excited and so |
1527s | as we were announcing it that's why |
1528s | Recon was just like literally like as we |
1530s | were telling the team this is what we're |
1531s | gonna do and and it's a go Recon started |
1534s | blocking out it was like crazy he's just |
1537s | like he grabbed defense started he was |
1539s | posting little images on our little |
1541s | slack Channel and we're just like whoa |
1543s | that actually looks pretty cool right |
1546s | so Doom was really the starting point |
1547s | like saitama was our yes right |
1550s | and every time feel free to jump in if |
1553s | there yeah I mean going back to that |
1555s | meeting I remember you know sitting in |
1556s | like every day meeting it's all going |
1558s | good and then boom one punch man collab |
1560s | like hold up wait a minute what is this |
1562s | I did I saw the concept shout out to |
1565s | kitchen for the awesome concept I was |
1567s | like I need I need to get on this like I |
1569s | need to work on it so I hit up my |
1571s | producer I'm like I don't know who's |
1572s | working on this I don't know if it's |
1573s | already a sign I need it needs to be |
1575s | mine I'm a huge fan of you know one |
1577s | punch man and yeah I stopped working on |
1579s | the blackout like Dion said just took it |
1582s | from there so it made sense that |
1584s | obviously saitama would be doomfist you |
1586s | know he's our basically our one punch |
1588s | man in our universe and um if you notice |
1592s | the heroes |
1593s | so we decided you know we learned Aaron |
1596s | Keller our game director is a big fan of |
1598s | one point so he knew the characters |
1600s | really well in a lot of their lore |
1602s | and so as the character team and the |
1605s | concept artist started to dig in we |
1607s | decided that's how we started to develop |
1609s | which which characters made sense to be |
1611s | who and we'll dig a little bit more into |
1613s | doomfist in just a moment but literally |
1614s | because Recon started to block out he |
1616s | has to station |
1617s | but Janelle's being Genji just made |
1621s | sense from a silhouette perspective and |
1622s | the way the characters look in in their |
1624s | combination from the show and then uh |
1626s | Moomin Rider 76 it just it was just so |
1629s | fun he's a kind of a fun character for |
1631s | us already so with Moomin Ryder becoming |
1633s | 76 was a a pretty awesome clap |
1637s | yeah and walk us through some of the the |
1640s | concept art or through some of the early |
1642s | development stages like you mentioned |
1644s | that like Genji and Jenna's they share |
1646s | similar Silhouettes like what what what |
1649s | does that mean |
1651s | so gameplay wise you know OverWatch is a |
1654s | very fast-paced game so we're gonna dig |
1656s | into uh doomfist here |
1658s | so Tama has a cape that's a pretty |
1661s | iconic look for this hero in one punch |
1663s | man so we needed to add a cape to |
1666s | doomfist |
1667s | which is a big deal this is changes |
1669s | doomfist silhouette significantly and so |
1672s | when you play in OverWatch tactically |
1674s | you're looking for certain Silhouettes |
1675s | we do a lot of work behind the scenes |
1677s | when we're developing a hero to they |
1680s | have a certain shape a certain size that |
1681s | that's eventually becomes recognizable |
1683s | to the players you can see the the |
1685s | original doomfist that you know the one |
1687s | that we already have in the game and the |
1688s | first thing like Dion said we need to |
1690s | test out the cape so I did a quick block |
1693s | out of that model and then I submitted |
1694s | to our Tech team that you know what we |
1697s | want to test out how how's that is going |
1699s | to work with this movement because again |
1700s | if you think about doomfist he's not |
1702s | like your average character just he |
1704s | walks straight or you know he's always |
1706s | jumping doing crazy moves so we did this |
1708s | uh lockout test and it was successful |
1710s | and then you know we got the green light |
1712s | to move with the other |
1715s | Parts here you can see uh more uh we |
1717s | call this a blackout stage because it's |
1720s | basically just throwing Big Blocks and |
1722s | making sure we we represent the concept |
1724s | in early stage to test out for example |
1726s | the belt needs to be smaller bigger we |
1728s | need to move parts around |
1730s | and so this is him going from like |
1732s | normal Doom to yeah we're dressing we're |
1735s | putting him in the block out for it |
1737s | exactly yeah we start to test out in the |
1740s | game too as well right now we get this |
1742s | block out in game and we get a feel for |
1744s | how fighting this hero now yeah feels |
1748s | here you can see I'm actually in game |
1750s | but again there there isn't much details |
1752s | going on just a quick color and the |
1754s | model and after you know everybody's |
1757s | like he's good to go we moved to the we |
1759s | called the high poly stage this is where |
1761s | we put all the details like you can see |
1763s | all the the pants they got the the seams |
1766s | on the sides he got the armor pieces and |
1769s | uh after that we go back to uh something |
1772s | we call the low poly which is like the |
1774s | clean mesh that you guys see in the in |
1776s | the engine and here you can see him with |
1778s | you know all the textures and the |
1780s | lighting uh in the engine |
1782s | much of the uh characters which were |
1784s | selected from the one punch Universe |
1787s | were selected based on their Silhouettes |
1789s | matching up with one of the Heroes and |
1791s | how much of it was just like well it's a |
1793s | favorite of ours so we have to present |
1796s | I think a lot of it came from a favorite |
1798s | of ours so we we have a pretty varied |
1800s | cast of characters so surprisingly a lot |
1803s | of people just fit automatically but a |
1805s | lot of it is mostly came from the team |
1808s | just that's their favorite characters |
1809s | from the show |
1811s | um and you can see that the characters |
1813s | are kind of cosplaying the heroes from |
1816s | one punch man it's not saitama's |
1819s | literally in the game it's doomfist |
1820s | wearing a costume that is uh very |
1823s | reflective of saitama he's a big fan |
1825s | himself so that's kind of our approach |
1827s | for um whenever we do a collab like this |
1830s | is that it's our heroes enjoy the |
1832s | universe just as much as we do in real |
1834s | life and they're kind of cosplaying as |
1837s | the heroes from that Universe it's |
1838s | actually so funny because if you're not |
1840s | thinking about kiriko cosplaying as uh |
1842s | tatsumaki tatsumaki being such a ancient |
1845s | words kind of character and certainly |
1847s | like arrogant kind of edgy person and |
1849s | kirika seems so wholesome and sweet so I |
1852s | love that she wants to cosplay as is a |
1854s | dark one yeah that's right yeah |
1857s | I actually want to talk about |
1859s | um something that young just mentioned |
1860s | is you know our characters are |
1862s | cosplaying you know the the one punch |
1864s | man right now so uh one of the |
1866s | challenges that uh we can run into is uh |
1869s | in the original concept that uh we did |
1872s | he had a lot of muscles and then if you |
1874s | look at you know cytom and the enemy is |
1876s | like super simple you know just a yellow |
1878s | shirt red gloves that's it like he |
1879s | doesn't have much going on so the the |
1882s | tricky part was to make sure we add |
1884s | enough details so it still looks like |
1886s | doomfist but at the same time you know |
1888s | you can tell it's aside from us so uh |
1890s | one way you know to go around these |
1893s | um challenges was like for the the shirt |
1895s | for example it's just yellow so we do |
1897s | something in our watch where we add |
1898s | gradients it's like a cool way of adding |
1900s | details without making something too |
1902s | noisy and same thing pretty much with |
1904s | the gloves and the boots we just added a |
1907s | little bit of gradients because again we |
1908s | we're trying to make it match the |
1910s | concept but at the same time still make |
1912s | it look like it's in the OverWatch world |
1914s | awesome you kind of went into uh some of |
1918s | the challenges for development I'm |
1920s | curious like from a design perspective |
1921s | what was it like putting like these |
1923s | color palettes in game or you know |
1926s | creating creating the meshes that you're |
1928s | talking about |
1929s | I mean it's it it was really awesome |
1932s | like to to see that concept and then uh |
1934s | you know figure out how how are we gonna |
1936s | make it work in our world because you |
1937s | know yeah we we don't we want to keep |
1940s | you know the IP looking like it is at |
1942s | the same time we still want to represent |
1944s | OverWatch language in our characters so |
1946s | uh we we did a couple of tests we |
1948s | changed a couple small parts here and |
1950s | there just so it kind of flows better |
1952s | and you know it's not as noisy and at |
1954s | the same time you look at it like uh |
1956s | from this it looks simple just like uh |
1958s | cytom is and I know finally seeing this |
1961s | in game was a really significant moment |
1962s | for our team and like when we're sharing |
1965s | that information internally with one |
1967s | another like it's very common for a zoom |
1971s | call to devolve into a twitch chat very |
1973s | quickly with how excited everyone is |
1976s | especially when they start hearing like |
1978s | that theme music things start coming |
1980s | together like at the trailers and you |
1982s | know everyone's just completely stoked |
1984s | for this |
1986s | we had a lot of fun uh because again we |
1989s | realized there were so many fans on the |
1991s | team that you just never knew until we |
1993s | started to talk about this |
1995s | um even the lobby we worked on for |
1998s | the scene is the one punch challenge |
2002s | where he's doing the under squats the |
2004s | hundred so that's |
2005s | and you literally doomfist does the |
2007s | entire exercise if you watch the yeah |
2010s | Lobby he will do every all 100 of each |
2013s | step and then walk off and do the |
2015s | running too but he comes back at the |
2018s | time of day change it it just it was so |
2019s | fun because at first we had like a |
2021s | truncated version of it but we felt |
2023s | pretty strongly to just pull this off he |
2027s | can actually do the full exercise so |
2030s | it's a fun thing if you want to watch |
2032s | it's like five minutes long yeah I |
2034s | remember you saying the original was |
2036s | like 30 seconds and then you're right |
2037s | now |
2040s | people who are not watching this right |
2042s | now who are just sitting in that screen |
2045s | trying to figure out if it's the whole |
2047s | sequence so some people were 100 just |
2049s | like wait it out and see yeah that's |
2052s | what we hope and people get the joke |
2054s | basically but I think that shows just |
2056s | how strong some of the fans on the team |
2059s | where the animation team really pushed |
2060s | for it actually took a little bit of |
2062s | tech adjustment to allow a Lobby to have |
2064s | this level of uh kind of uh |
2067s | keyframing that lasted so long |
2070s | uh because usually our lobbies are |
2071s | pretty quick and and they're beautiful |
2074s | but quick yeah this one is just awesome |
2077s | forever |
2079s | watching doofus do this amazing exercise |
2082s | that was a pretty awesome challenge in |
2084s | real life right it's a lot of people |
2086s | used to do this as a |
2088s | a great goal of theirs to be able to |
2090s | achieve this episode so exercise so |
2092s | doomfist to be in one of our strongest |
2094s | Heroes he could totally do it not to |
2097s | mention everybody on the team that |
2098s | worked on this collab had to do that |
2100s | workout |
2101s | of course it's a rite of passage |
2106s | now obviously like from from a viewer's |
2109s | perspective like love the anime or the |
2111s | manga we see it in one of our favorite |
2113s | games that's huge people get so excited |
2115s | about collaborations like that what was |
2117s | it like for you guys in the team because |
2120s | being able to work on something which is |
2122s | still within your you know day-to-day |
2124s | business it's OverWatch it's what you |
2126s | work on yeah but then you get to add |
2128s | like something you're a fan of what was |
2131s | that like |
2133s | uh yeah I feel uh I think both Khan and |
2136s | I can answer this but we actually wanted |
2139s | to do this for a long time it's just |
2141s | finding the right uh group to collab |
2144s | with working with Bandai has been super |
2146s | fun and we've wanted to figure out what |
2149s | it meant for OverWatch to take this step |
2151s | it's a pretty big step for us as an IP |
2153s | we already have this huge IP with |
2155s | amazing heroes in it but we're fans of |
2157s | these other groups that have amazing |
2159s | Heroes too how do we integrate those |
2162s | into our game |
2164s | and like to me personally I remember |
2166s | when I saw the concept like Dion said I |
2168s | I had the blackout ready in no time how |
2171s | excited I am and I had other stuff like |
2174s | going on I had to finish and I told the |
2176s | truth I will make it work you don't |
2177s | worry about that yeah |
2179s | and everybody on the team was like super |
2182s | excited about seeing the concept seeing |
2183s | the whole thing you know just in the in |
2186s | the engine and the custom stuff that we |
2188s | did it it just shows like how everybody |
2190s | on the team were like super excited and |
2192s | just happy like we're doing this kind of |
2193s | collapse now yeah |
2196s | and finding one punch as the first one |
2198s | was awesome you know |
2200s | I like saying that like this was really |
2202s | the you know right |
2205s | partnership for us in a sense because |
2207s | like their their IP and rip really go |
2210s | together really well and you know seeing |
2213s | Our Heroes come to life as their |
2215s | characters uh or like representing their |
2218s | characters is really it's really special |
2220s | were there any other moments uh like I |
2224s | know we've mentioned doomfist and uh |
2226s | kind of the blockout for his uh his |
2229s | saitama skin but were there any other |
2231s | moments with say the other skins genos |
2234s | or a tornado kiriko that really stood |
2236s | out to you guys |
2237s | uh I mean I can talk about uh Janice |
2240s | because my friend Val he worked on that |
2242s | skin and we were both geeking out the |
2244s | whole thing so much just when we saw the |
2246s | concept of like I gotta work on this and |
2248s | so I gotta work on doomfist and we both |
2250s | you know while working we kept sharing |
2251s | updates because we were just so happy |
2254s | about it and excited this one is |
2255s | basically because just how much I get |
2257s | excited about these things I decided |
2259s | that you know what just showing it in |
2261s | the engine is not enough I want to take |
2263s | it to the next level and I just throw |
2264s | this like cool cinematic rendering of it |
2266s | yeah just to show you guys like how |
2268s | excited and again while working on this |
2270s | we're not sure if we're like able to |
2272s | share it or not but to us just how |
2275s | excited we get about these things like |
2277s | we we try to do extra stuff just to get |
2279s | you know the whole team but like super |
2281s | excited about it too and hyped yeah I'm |
2284s | pretty sure at least 75 of our team has |
2286s | that as our background yeah sure it's a |
2289s | really popular yeah |
2293s | channels is interesting because during |
2294s | the concept and phase this is where we |
2297s | learned where too much he at one point |
2300s | he looked too much like Janos we've lost |
2303s | um uh the OverWatch hero and so we |
2306s | peeled it back a little just to find The |
2308s | Sweet Spot uh before we model we do a |
2310s | lot of testing with the just the concept |
2312s | the 2D drawing |
2314s | to make sure the OverWatch hero remains |
2317s | while still celebrating you know the uh |
2319s | genos as a character so he was one that |
2322s | in some of the concept of faces actually |
2324s | was pushed a little too far and then we |
2326s | kind of peel them back |
2327s | yeah that's that's really interesting |
2330s | and I'm I'm kind of curious about kiriko |
2333s | and Moomin writer if there were any |
2335s | highlights from working on either of |
2337s | those Heroes |
2338s | um I know for Moomin writer I'm probably |
2340s | just as excited for the Highlight intro |
2343s | as I am the skin itself |
2346s | um it's it's pretty pretty awesome |
2349s | yeah Moomin Rider it just makes sense |
2352s | because we had this kind of uh |
2355s | Sentai skin for 76 already and then the |
2358s | character the show is kind of like that |
2361s | he's like a play on this category of uh |
2363s | entertainment so it just felt even more |
2366s | funny that in game we were literally |
2368s | taking you know kind of an evolution of |
2370s | a model we already had made him look |
2372s | like movement and then we even did the |
2373s | Highlight Reel with the cool anime lines |
2376s | going behind it as he rides it just |
2379s | started to you could just see it we the |
2382s | whole team was just having fun with this |
2383s | one every there was much more we wanted |
2386s | to do but obviously time and production |
2388s | costs we just had to really focus on the |
2391s | the heroes that uh made sense but |
2394s | hopefully we do more in the future |
2396s | I really love that love that have fun |
2398s | with this and and probably |
2400s | one of the highlights |
2401s | I'm curious |
2405s | and Fanboy moments from working on this |
2408s | partnership any like uh highlights for |
2411s | either of you |
2412s | I mean I could say I've watched it again |
2414s | like 60 times after like I just had to |
2417s | watch it again yeah yeah because while |
2419s | watching it I wanted to pick up I don't |
2422s | know some some small thing that I can |
2423s | put on on the skin that kind of you know |
2426s | yeah so I I watched it again again just |
2428s | wanted to pick up you know some small |
2430s | details here and there that I can add to |
2431s | the skin and hopefully like fans will |
2434s | pick them up because they're not super |
2435s | you know just out there showing but |
2437s | they're like hidden so yeah I'm |
2439s | interested to see everybody's got a spot |
2442s | yeah I I just thought the fact that we |
2445s | were doing it in general I again I |
2447s | watched this series already but they |
2449s | started watching it again and again and |
2451s | again |
2452s | so it just it felt like |
2455s | okay we did find the right |
2458s | show to collab with |
2460s | absolutely I know that I'm gonna be in |
2462s | game looking for those details too uh I |
2465s | was new to the series but when I found |
2467s | out we were doing this collaboration I |
2469s | immediately watched it and like the |
2471s | intro song is what grabbed me from then |
2474s | I was like okay yeah this is my vibe I'm |
2476s | really gonna like this so |
2478s | um yeah I'm I'm you know really excited |
2480s | this is a game and I'm I know that our |
2483s | our community is too well I'm gonna say |
2485s | this now and I feel sorry for everybody |
2487s | that's going to be in my game I'm really |
2489s | bad with doomfist but I will be playing |
2491s | doofus because of this skin so just a |
2493s | heads up I'm not too bad a doofus I'm |
2496s | really excited to pick this game up it's |
2499s | so good yeah awesome well this is |
2502s | overwatch's you know first major anime |
2504s | collaboration and players can jump into |
2507s | the game it's live right now uh so |
2510s | whether you want to pick up the |
2511s | Cosmetics try your hand at doomfist or |
2514s | you know count count the Reps and Doom's |
2517s | workout |
2518s | um jumping game right now and |
2521s | uh have a blast |
2524s | thank you Dion and thank you Recon for |
2526s | joining us and sharing your thoughts and |
2528s | experiences you know it's always a joy |
2530s | talking to both of you and yeah we |
2533s | really appreciate you being here bye |
2535s | thank you thank you yeah thank you guys |
2538s | thank you bye no we have |
2541s | a topic lined up for this next segment |
2543s | that's a bit of conversation starter in |
2545s | the community but it's also something |
2547s | incredibly significant to our team |
2550s | we're joined by some really incredible |
2552s | folks on Team Four who are going to walk |
2554s | us through some of the competitive and |
2556s | matchmaking updates that we're going to |
2557s | see here in the mid-cycle patch Scott |
2560s | would you please kick off intros and |
2562s | give us a little overview on your role |
2564s | on the team |
2565s | sure thing Jody um hi everyone I'm Scott |
2567s | Mercer uh I'm the lead meta designer on |
2570s | OverWatch 2 and that doesn't mean I'm |
2572s | responsible for the hero meta nope |
2574s | that's not me not at all different meta |
2576s | it's different meta completely uh what I |
2579s | do on the team is work with Morgan on |
2581s | matchmaking I also work on competitive |
2584s | systems player interaction and Trust I |
2586s | interact with Esports teams uh and yeah |
2589s | I'm really excited to talk uh with |
2591s | Morgan today about matchmaking |
2593s | um yeah hi I'm Morgan I'm a senior |
2595s | engineer on the team four |
2598s | um I also mostly focus on matchmaking |
2599s | and ranked so I actually work with Scott |
2601s | kind of yeah yeah kind of an uncanny |
2604s | amount of but yeah that's me |
2607s | awesome welcome on in you guys and |
2609s | thanks for joining us on this again |
2612s | um so Morgan uh to kick us off will you |
2615s | kind of give us a recap on the |
2618s | matchmaking updates that went in at the |
2620s | beginning of season three uh what were |
2622s | those updates and kind of what are some |
2624s | of the things that were uh seeing in |
2626s | them serving so far yeah so um season |
2630s | three was really big for matchmaking |
2632s | um it was kind of so |
2634s | um right after OverWatch 2 launched |
2636s | actually we were kind of thinking about |
2638s | like what we wanted to prioritize next |
2640s | and um kind of the overwhelming thing |
2642s | that we wanted to do was |
2645s | um what we've called the the roll Delta |
2647s | change which is minimizing the roll |
2649s | Delta between the two teams and so what |
2651s | that means is |
2652s | you know um before any of these changes |
2655s | went live you know we were pretty |
2657s | willing to make a match that could have |
2659s | like a diamond level tank against a gold |
2661s | level tank say and even if the rest of |
2664s | the match was fairly balanced you know |
2666s | in terms of like the ratings of of the |
2668s | two teams |
2669s | that matchup just didn't feel very fun |
2672s | and so |
2673s | um you know those matches had a 50 win |
2676s | rate uh but like it still didn't feel |
2679s | like a good match and so |
2681s | um we kind of wanted to like |
2683s | you know see how how well we could |
2686s | minimize the Delta between the rolls and |
2688s | so we wanted to do that not just for the |
2689s | tanks but for all the roles on the on |
2691s | each team |
2693s | um and that actually took a really long |
2695s | time it was it was definitely the |
2697s | largest kind of architectural change |
2699s | that we've made to the Matchmaker since |
2701s | roll Q even uh back in 2019 |
2704s | so um so we did that over a few stages |
2708s | uh we kind of got like an easy win and I |
2711s | think it was like the season two patch |
2712s | and then uh since that point we were |
2715s | kind of making iterative improvements |
2717s | and so season three is kind of the |
2718s | culmination of that work where |
2721s | um you know we've kind of like fully |
2722s | pivoted the the matchmaking architecture |
2724s | to be able to kind of like reason about |
2728s | um packing players into matches that in |
2731s | a way that minimizes this Delta and |
2733s | keeps it so that the teams have kind of |
2735s | like this balanced role |
2737s | um you know this property is kind of |
2739s | weird |
2740s | um it's not it's it's it's kind of hard |
2742s | to like get it for free |
2745s | um and so we actually kind of had to |
2747s | rethink a lot of the different |
2749s | components of the Matchmaker in order to |
2752s | to actually solve this problem well not |
2754s | just in the normal case but also kind of |
2756s | in the worst case be able to do this you |
2760s | know kind of in all cases so um so that |
2764s | was kind of the big thing that changed |
2766s | in season three |
2768s | um because of that uh you know there was |
2771s | a whole lot of um parts that we expected |
2774s | to get worse as a result of that |
2778s | um you know kind of like in in |
2780s | matchmaking you don't really get |
2781s | anything for free so by minimizing the |
2784s | roll Delta in these matches we've kind |
2786s | of |
2786s | um we've slightly impacted like other |
2790s | aspects of match quality that we're |
2792s | looking at |
2794s | um and some of that we kind of expected |
2795s | going into it because we just knew this |
2797s | was such a big change but at the same |
2799s | time you know there are some things that |
2802s | we were kind of surprised at like you |
2804s | know I was a little bit surprised at how |
2807s | our our expected win rate |
2810s | um you know normally we try to make |
2811s | matches that are 50 50. |
2813s | um our our average expected win rate got |
2816s | a little bit worse |
2818s | um by optimizing for role Delta and it's |
2822s | kind of unintuitive like you would think |
2823s | you know all of these these measures of |
2825s | of good matches would kind of like |
2827s | correlate together but they kind of |
2829s | don't um I mean you know if you make the |
2832s | one ideal match where everybody has the |
2834s | same exact rating you know then it's |
2836s | just a good match and we're willing to |
2837s | make it |
2838s | um but you know oftentimes we can't do |
2840s | that and so it's kind of like which |
2842s | thing do we give up on First and so |
2845s | um we've still been tuning the system a |
2847s | lot like ever since it launched I think |
2849s | we've we've probably made a tuning |
2851s | adjustment almost every day at this |
2853s | point yeah and we're still going you |
2855s | know like uh we're still learning |
2858s | um and a lot of of that has just been |
2860s | you know seeing what what goes wrong and |
2862s | seeing how we can kind of like tweak |
2863s | things to make things a little bit |
2865s | better so yeah that sounds like some |
2867s | huge infrastructure kind of challenges |
2869s | going on there right yeah so how was |
2872s | that perceived by the community like |
2874s | what was the how how did they react |
2876s | because you said it has a lot of side |
2878s | effects if you make such big changes |
2880s | like queue times are going to be |
2881s | affected What were what was the |
2883s | consensus uh and the feedback you were |
2885s | given after implementing such big |
2886s | changes that's a really good question |
2889s | um you know it's it's very hard to |
2892s | actually say with certainty like you |
2894s | know how much better things got you know |
2897s | we have measures that we look at that |
2899s | kind of ground us and like okay you know |
2901s | we |
2902s | um we did actually make you know most of |
2904s | our matches now have these very close |
2905s | role deltas and um and like you know |
2909s | we've seen a lot of community feedback I |
2911s | would say it's mostly been positive |
2914s | um I don't know it's a little it's a |
2916s | little hard to say for sure yeah it's |
2918s | when you think of all of OverWatch 2 |
2921s | matchmaking there's a lot of different |
2923s | people that are affected by this |
2925s | um you know we're going to talk a lot |
2926s | about competitive play but like this |
2928s | change actually mostly affects quick |
2931s | play or effects quick play to an even |
2933s | larger degree because their groups can |
2936s | be |
2937s | even more different in terms of like the |
2940s | groups you can form and competitive play |
2942s | there are restrictions on who you can |
2944s | group with you can't group with you |
2946s | can't have like a bronze player group |
2947s | with a grand master player you just |
2948s | can't and quick play |
2951s | all those things are possible yeah you |
2953s | get some really and you get some really |
2954s | you get some really crazy groups and the |
2956s | Matchmaker has to do the best it can |
2958s | with it and it's all this new role dealt |
2960s | attack |
2961s | that allows us to |
2963s | you know create what we think are like |
2965s | overall like better matches for those |
2968s | groups and then they're absolute groups |
2971s | in competitive play too so yeah we get |
2973s | some benefits there but like for the |
2975s | people that do play in groups with like |
2977s | different with different skilled players |
2979s | and quick play |
2981s | all the feedback we've been getting |
2982s | beginning it's been great yeah um in |
2985s | some cases the queue times have been a |
2986s | little bit like longer because that's |
2988s | really the main the main downside that |
2990s | we've heard so far is especially for |
2992s | large groups their Q times have gotten a |
2994s | little worse and that was a little bit |
2995s | unexpected I was hoping that that |
2997s | wouldn't get bad |
2999s | um it didn't show up in the simulations |
3001s | that we did but um we have some ideas |
3003s | for how to how to improve that going |
3005s | forward it'll probably just take another |
3006s | patch to really get it good |
3009s | um but yeah it's kind of funny you know |
3011s | we get a lot of feedback we see the |
3012s | metrics and then we also play the game a |
3014s | lot right like you know ultimately |
3016s | that's kind of the best way to ground |
3018s | yourself in like what this change has |
3020s | done and you know I personally have been |
3022s | able to feel this change you know it's |
3024s | really interesting |
3025s | um you know especially when you're |
3027s | playing tank but even with the other |
3028s | roles you can kind of tell now like it |
3031s | who your rival is on the enemy team you |
3034s | know because we're really trying to |
3035s | match you with somebody who's very close |
3036s | to your skill level and you know now in |
3039s | the average case that person you know |
3041s | like in ranked for example is is like |
3043s | within one one division of you |
3046s | um you know so they're very very closely |
3049s | matched with you and so you can really |
3050s | kind of feel you know a little bit more |
3053s | of like that rivalry where it's like you |
3055s | know are you winning that matchup or |
3056s | losing that match up and it's like it's |
3058s | not like you're just gonna get stomped |
3060s | either way like that's going to be a |
3062s | pretty close match and if you know |
3063s | there's there's kind of a feel to it |
3065s | that um I've thought is really cool yeah |
3068s | of course you know know there's like the |
3070s | other areas where like |
3071s | you know we're making slightly wider |
3074s | matches now in terms of the you know if |
3077s | you look at like the best player minus |
3079s | the worst player that's gotten a little |
3081s | bit wider also kind of as a result of |
3083s | this change |
3084s | um and so like you know that's actually |
3086s | something that we're going to be trying |
3088s | to improve we actually anticipated that |
3090s | that would get worse as a result of the |
3092s | Roll Delta change and so in the |
3095s | mid-cycle in in the 2.3.1 patch we have |
3098s | some uh changes going that will actually |
3101s | help shrink that back down hopefully |
3103s | even to better than it was before uh |
3106s | we'll have to wait and see the |
3108s | simulations were very promising but you |
3110s | know the simulations don't don't usually |
3112s | end up like real life so no for sure we |
3115s | don't know what the system like seeing |
3117s | how like our players actually interact |
3118s | with the system is the best way like for |
3122s | us to learn and take things away from it |
3124s | and I really like the way that you |
3126s | framed it as a big architectural change |
3128s | to the system because like you tweak one |
3131s | thing and then you find out that one |
3133s | thing like you have a goal for what that |
3136s | thing is supposed to change but then it |
3137s | inadvertently affects these other |
3139s | variables as well and |
3142s | um so it's it really is a process |
3144s | getting to a place where |
3146s | that feels good for the people that are |
3149s | are playing yeah it's been it's been |
3151s | really interesting too like you know |
3154s | competitive queue times used to be |
3156s | longer and now they're shorter |
3158s | um and like in retrospect like the |
3161s | system maybe wasn't actually working as |
3162s | well like the new change |
3165s | um |
3166s | you know it's like we're still making |
3167s | very good matches but we kind of like |
3169s | for free almost got rid of a lot of the |
3171s | queue time just because you know the new |
3173s | system is like a lot smarter and and |
3175s | it's kind of this like foundational |
3177s | piece that um you know gives us like a |
3180s | lot more ability to build off of in the |
3182s | future too which is going to be really |
3183s | cool |
3185s | so what about players who queue into |
3188s | matches and they see this wide skill |
3190s | disparity like they're queuing in in |
3192s | their diamond and they see a golden or |
3195s | or even more so than that they're |
3196s | queuing in at GM and see like a plot |
3199s | player in their match like what's |
3200s | happening there |
3202s | yeah that's actually funny so after the |
3204s | the season three changes you know this |
3207s | is it was a really big change and so |
3209s | there were a lot of unintended |
3210s | consequences and one of them was we for |
3213s | a short period were making matches that |
3214s | were a lot like had a much wider skill |
3217s | disparity somebody uh sent uh us a match |
3220s | that was like a GM match with a gold |
3223s | level player it might have been two of |
3225s | your colleagues on the OverWatch League |
3227s | cast Zoe that sent us a match and they |
3230s | were like |
3231s | we were we were matched with this like |
3233s | with this gold player it's like what's |
3235s | up so we actually went and looked at it |
3238s | yeah we spent like a couple hours |
3240s | talking there was actually a lot of |
3241s | things where we talked about which is |
3242s | why again looked like pointed feedback |
3244s | about some of the things that's like so |
3245s | useful to us but like yeah we looked at |
3247s | it and it's like yeah there there was a |
3249s | gold |
3250s | um the other things about the match is |
3252s | like they didn't realize it's like oh |
3254s | the the role the role does stuff was |
3256s | actually working really well it had a |
3258s | gold on the other team as well no |
3263s | see that uh so that was like at the same |
3267s | time it's like oh this is bad but it was |
3268s | kind of good too it's like it's working |
3270s | yeah it was working in a way we didn't |
3273s | quite expect one of the three things |
3275s | about matchmakers is you know the |
3276s | matchmaker is like we put this |
3278s | constraint and it's like yeah you know |
3280s | it's thing was like all right push this |
3281s | down and like something comes up you |
3284s | know it's like you can't really push |
3285s | everything down all at the same time |
3287s | because you don't have enough hands |
3289s | so yeah it was uh it's definitely like |
3291s | something that uh we happen but we we |
3294s | know that we we know the players don't |
3295s | think that's that's a high quality match |
3298s | uh even if we make one |
3300s | asking yeah because it happened to me a |
3303s | bunch of times |
3305s | player I ended up with GM people and |
3307s | what I took from it was the game |
3309s | complementing my skill because I was |
3311s | unaware I was good to get back I was |
3312s | like oh you really think I'm that good |
3314s | so I should be playing with the Empire |
3315s | as well thank you so much yeah I I've |
3318s | seen feedback like on various like |
3320s | social platforms where it's like yeah |
3322s | there are a lot of people are like that |
3323s | or like I was super excited to play with |
3325s | these players yeah I never get to play |
3326s | with these players it's like you know |
3328s | because |
3329s | playing with players that are more |
3331s | skilled than you is actually an |
3333s | interesting learning opportunity yeah |
3335s | it's like how do these players play yeah |
3338s | yeah it makes me play better but |
3340s | obviously that's not how it yeah |
3344s | obviously it's like we're gonna like |
3346s | actually try to prevent that sort of |
3348s | thing from ever happening again |
3351s | I want to sneak through in a blue moon |
3352s | but like it's you know but it's yeah |
3355s | we're able to make some tuning changes |
3358s | like once we saw the problem because you |
3360s | know we add so much configurability into |
3362s | every change that we make so once we saw |
3364s | that problem we were able to make it |
3366s | better I mean it's we can never like |
3369s | have like a hard cons you know like some |
3371s | players will be like oh I only ever want |
3373s | to make you know I only ever want to |
3374s | play like these perfect matches or |
3376s | something like that and like you know we |
3377s | ultimately can't do that we have to kind |
3379s | of like eventually make a match right |
3381s | you know we can't just have somebody |
3382s | stay in queue for four hours |
3385s | um and so you know this was just one of |
3387s | the things where it was like you know we |
3388s | kind of hadn't tuned the system just |
3390s | right and so we were able to I think |
3392s | mostly fix this problem at least improve |
3394s | it |
3395s | um but it's still something we would |
3396s | like to improve even more like I think |
3398s | you know ideally as fun as it is for the |
3401s | gold player like you know we probably |
3402s | don't ultimately ever really want to |
3405s | make matches that have gold players and |
3406s | GM players together yeah |
3409s | um and yeah that sort of gets back to |
3410s | the point of like |
3412s | High Jam players |
3414s | they're they're always going to be a |
3416s | challenge for the Matchmaker just |
3417s | because there's not that many of you |
3419s | yeah sorry uh that was really just too |
3422s | good yeah it's so like trying to find |
3426s | like a perfect configuration of people |
3428s | at any one time you know it's just |
3431s | difficult so that's why I mean well |
3434s | again well we've had queue times come |
3436s | down we've able to make some adjustments |
3437s | there that I think have been really |
3438s | positive in that regard but it's always |
3440s | going to be more difficult than your |
3443s | gold or platinum game is where we |
3444s | literally have |
3446s | you know it's so many like different |
3448s | players and like it's it's very easy to |
3451s | come up with much more precise matches |
3452s | if you're you know silver to platinum |
3454s | you know you're getting kind of our the |
3457s | high population experience right you |
3459s | know that's like we have a whole lot of |
3461s | players there but you know once you get |
3463s | really high up once your Grand Master |
3465s | Level like you know the best way to |
3468s | increase match quality is not any change |
3471s | we can make it's like |
3472s | message your friends and tell them to |
3474s | get online because otherwise like |
3476s | there's just nobody to play against you |
3479s | know yeah it's just unique problems you |
3482s | need to solve in the different ranks |
3483s | then right I mean it's one system but it |
3485s | can't really fit all very much like hero |
3487s | balancing you have different balancing |
3490s | issues in bronze as you would have in a |
3492s | GM game yeah absolutely it's actually oh |
3494s | go ahead yeah we know that match quality |
3496s | High GM takes a nosedive whenever owl |
3498s | players are screaming yeah |
3501s | it's absolutely a thing |
3504s | um |
3505s | yeah but it's like once they're done |
3506s | with scrims and like you know start like |
3507s | you know hopping on and you know playing |
3510s | Solo or someone or someone else and yeah |
3513s | all of a sudden it's like wow badge |
3514s | quality is like so much better yeah |
3517s | it's really um it's really kind of |
3519s | interesting you know like at like gold |
3522s | level you know it's actually pretty easy |
3524s | to make good matches most of the time |
3526s | you know we still have edge cases |
3528s | sometimes but at you know that kind of |
3530s | like really high MMR |
3533s | um really the only way you know kind of |
3535s | is like this fundamental trade-off |
3537s | typical in match in matchmaking where |
3540s | it's like either you have better match |
3541s | quality or you have |
3543s | um and longer queue times or you have |
3544s | shorter queue times and worse match |
3546s | quality and like there's kind of this |
3548s | fundamental you know trade-off there |
3550s | like |
3551s | you know we could make match quality |
3553s | incrementally better by putting people |
3556s | in queue for twice as long like the |
3559s | relationship isn't exactly you know one |
3561s | to one right so like you know Pro |
3563s | players will sometimes tell us like I |
3564s | want you know better match quality and |
3566s | we'll be like well |
3568s | I could make your Q times twice as long |
3571s | but that doesn't sound very appealing |
3573s | right |
3574s | um but like there are some things you |
3576s | know long term that we're looking at |
3578s | where we can not just make matches you |
3581s | know better in One Way by pulling from |
3584s | the other one but we can actually be |
3585s | smarter and the problem there is it's |
3588s | just so hard like matchmaking at that |
3590s | high MMR it's like it's like this |
3592s | impossible problem but we can be more |
3595s | and more sophisticated you know like |
3596s | I've been you know kind of looking at it |
3598s | as like okay so if like if this is what |
3600s | the queue looks like and I was like a |
3602s | human trying to make a match what would |
3604s | I be trying to do it you know it's hard |
3606s | because it's like we have to encode kind |
3607s | of like the wisdom that we have into you |
3611s | know a computer program that's running |
3612s | that's making you know hundreds or |
3614s | thousands of matches a minute like you |
3616s | know in this real-time way but you know |
3619s | long term there are some things that we |
3621s | would love to do to try to kind of like |
3624s | start improving match quality at the |
3626s | High Mr it's just so hard and so |
3628s | experimental so like you can expect us |
3630s | to definitely definitely try you know |
3632s | over the next year or so some things |
3634s | that hopefully will help but I you know |
3636s | I can't actually make any promises |
3638s | because it's like I really have you know |
3641s | when before we we push like a change you |
3644s | know like the role Delta change was like |
3645s | I didn't actually expect the the queue |
3649s | types to get a lot faster I expected |
3651s | match quality to maybe get a little |
3653s | worse but you know it's like now I'm |
3655s | like okay it didn't exactly turn out |
3657s | that way |
3658s | so yeah there's a there's just a whole |
3661s | lot of work that has to go into really |
3663s | making match quality better at that |
3665s | crazy high MMR |
3667s | you can deal with |
3669s | with form uh matches then as well |
3672s | um so in uh ranked we just don't let |
3675s | players play cross-platform so like |
3677s | either you're playing uh you know mouse |
3679s | keyboard or you're playing in the you |
3681s | know console pool and those are just |
3683s | totally separated and and you have |
3685s | separate ranks too so if you're playing |
3687s | on the same account on both of those uh |
3690s | both of those ranks like you can you |
3692s | know in theory you can be like a gold |
3693s | level mouse keyboard player and a |
3695s | diamond level controller player |
3698s | that's me |
3701s | laughs |
3703s | I think it really speaks to how many |
3706s | different variables go into like our |
3709s | Matchmaker and making these like |
3711s | changing the system and the different |
3713s | things that the system can affect like |
3714s | just by one small little tweak so thank |
3718s | you for going into that and giving us a |
3720s | little more insight on that would this |
3722s | be a good time to kind of look at some |
3724s | of the data that we've observed from |
3726s | this roll Delta change and maybe how |
3728s | that's affected the game up into this |
3729s | point yeah sure so um so here's a chart |
3733s | you can see here's the here's the |
3735s | average uh Delta between |
3738s | um between two tanks in a match a |
3741s | typical match that we make and um damage |
3744s | and support we don't show them here but |
3745s | they're very similar same Dynamic so |
3750s | um you can see when when we shipped |
3752s | OverWatch to uh you know we were making |
3754s | very good matches in general but |
3756s | um you know we weren't really paying |
3757s | attention to this and like I said this |
3759s | isn't really something that kind of you |
3761s | don't naturally get matches that |
3763s | minimize this role Delta Without Really |
3765s | Trying for it so by default in |
3769s | competitive are |
3772s | you know we were making matches that |
3774s | were like uh the the average tank Delta |
3778s | was something like two uh divisions |
3781s | apart and in quick play it was three on |
3783s | average and you know average is like |
3786s | you know about half of the matches are |
3788s | better and about half the matches are |
3789s | worse and there was definitely like if |
3792s | you look at the worst matches you know |
3793s | you don't see them here but the worst |
3795s | matches were way way worse than this and |
3798s | so um |
3800s | you know at this point it's like the the |
3802s | teams are approximately balanced but um |
3804s | especially one of the things that we saw |
3806s | at ship was you know you might create a |
3809s | match where like the you know the tank |
3812s | on one team is the worst tank or the |
3815s | worst uh player in the match and the |
3817s | support player is the best player in the |
3819s | match and then on the other team it's |
3820s | swapped and so like that just felt kind |
3823s | of weird so |
3824s | um |
3825s | so we made a couple rounds of changes uh |
3828s | the first round came in the season one |
3830s | mid cycle and that got us a good chunk |
3833s | of the way there |
3835s | um but it wasn't until season three |
3837s | where we really uh were able to kind of |
3839s | like nail both the average case and also |
3842s | make the worst case quite a bit better |
3844s | too so you can see now |
3846s | um in competitive our average is is one |
3849s | division which is very very close |
3852s | um it you know that's really within the |
3854s | uh the bounds of you know are you at |
3857s | that point it's you know it's not really |
3859s | necessarily like an average skill |
3861s | difference it's like you know are you |
3862s | playing well today or you know did |
3865s | you know are you like exactly where |
3868s | you're supposed to be in terms of your |
3869s | MMR things like that so it's much it |
3871s | feels much much closer yeah |
3874s | of course |
3875s | um by minimizing this you know you don't |
3877s | really get anything for free in |
3879s | matchmaking so we can go to the next |
3881s | chart and you can see um |
3883s | by minimizing this roll Delta we |
3886s | actually uh made our expected win rate |
3888s | of our matches slightly more so ideally |
3892s | we like to make matches that are as |
3895s | close to 50 50 win rate as possible you |
3897s | know that's a really important uh |
3899s | principle for us of match fairness |
3903s | um and you know at OverWatch 2 ship we |
3906s | were very close to that |
3907s | um you know this is median so you can |
3909s | think you know 50 of your matches are |
3912s | better than this and 50 of them are |
3914s | worse and unfortunately there is kind of |
3916s | a tail a long tail here of of |
3918s | progressively worse |
3921s | um but you know we were very close to 50 |
3924s | for most of our matches at ship |
3928s | um the first world Delta change we made |
3930s | it was actually a little bit unintended |
3932s | uh we didn't realize how much worse it |
3935s | was going to get so you can see in quick |
3937s | play especially in the season one |
3939s | mid-cycle |
3940s | um |
3941s | you know the median expected win rate |
3943s | went up to 55 55 from my perspective is |
3946s | still not horrible like you know a 55-45 |
3950s | match |
3951s | the losing team or you know the the team |
3953s | that's not favored still wins 45 of the |
3956s | time so you know that's not really |
3958s | something that I think |
3960s | you know if you're playing a 45-55 match |
3962s | you probably won't feel like one side |
3965s | really has a noticeable Advantage |
3969s | um but like I said half of the matches |
3972s | are worse than that so it got worse um |
3974s | and you can see in season three we were |
3976s | able to somewhat |
3978s | um improve that you know once we saw the |
3980s | problem we were able to start kind of |
3982s | bending that curve back down and this is |
3984s | actually you know this is now a higher |
3986s | priority for us going forward like we're |
3988s | going to keep making changes to kind of |
3989s | minimize this even more because it's |
3992s | really important for us to to make |
3994s | matches that are really as Fair as |
3995s | possible the other weird thing about |
3997s | this is actually this expected win rate |
4000s | it it's role agnostic so that means it |
4003s | doesn't take into account the role mmrs |
4005s | that role Delta and so this measure |
4008s | getting worse is in a sense |
4011s | um it looks worse than maybe it is |
4014s | because by minimizing that difference |
4016s | between the rolls that's also increasing |
4019s | match fairness just in kind of a |
4020s | different dimension so you know it's |
4023s | it's a little bit hard to say you know |
4024s | did matches get significantly less fair |
4027s | I don't think so they might have gotten |
4029s | a little bit less Fair |
4031s | um and that's a trade-off that we're |
4032s | making in order to make each individual |
4034s | match up within the match feel more fair |
4036s | so you know when we talk about match |
4038s | fairness it's really hard you know it's |
4040s | kind of this holistic thing where we're |
4042s | looking at a lot of different factors |
4043s | and trying to balance them and it it's |
4045s | very hard to make this this balance you |
4048s | know ultimately we kind of have to |
4049s | decide which one wins sometimes and um |
4052s | you know we can't always make every |
4054s | match perfectly but at this point I |
4057s | think you know the world Delta changes |
4059s | seem like they're definitely making a |
4061s | positive impact on the game I mean when |
4063s | you're saying you know that the way it |
4065s | feels in the game if my individual |
4067s | matchup is closer as a tank uh and I'm |
4070s | not just being taken for a walk by the |
4073s | other tank and completely stunned even |
4075s | if I lose the entire match if the |
4077s | individual matchup feels closer then I |
4079s | don't feel like I'm getting the short |
4081s | end of the stick either right so yeah |
4083s | totally I think that it kind of makes a |
4085s | big difference but we're seeing in those |
4087s | graphs um the frequency going from like |
4089s | season to the next season uh what's the |
4093s | philosophy for the uh systems team here |
4095s | uh how often how frequent do you make |
4099s | changes bigger changes smaller changes |
4101s | any kind of changes to the matchmaking |
4102s | without disrupting too much yeah |
4105s | um so I mean going forward matchmaking |
4108s | is just a huge priority for us in |
4110s | general |
4111s | um you know with OverWatch 2 we we've |
4113s | kind of you know the team's larger we |
4115s | have more time and and energy to really |
4118s | devote to making matchmaking as good as |
4119s | it can be and also it's just been a |
4121s | learning opportunity for us too like |
4124s | honestly since we shipped roll Q back in |
4127s | 2019 |
4129s | um at that point we had no idea what |
4131s | made roll cue matches really good you |
4134s | know we we had our intuitions from |
4135s | before roll q but it's a lot of that |
4137s | turned out to be wrong and so you know |
4140s | at this point I'd say you know we've |
4143s | really over the past several years |
4144s | really learned what makes OverWatch |
4147s | matches good and now that we have the |
4151s | time that we can really devote to doing |
4153s | the work I mean there's I would say we |
4155s | have probably like a Year's worth of |
4157s | work that we've talked about in terms of |
4159s | you know this road map of all of these |
4161s | things that we're aware of |
4163s | um so you know in the long term like |
4165s | this is going to be a huge Focus for us |
4167s | going forward is making large changes |
4169s | large improvements you know almost on a |
4171s | patch by patch basis and then even |
4174s | smaller than that like you know the |
4175s | season |
4176s | um three mid-cycle has I don't know |
4179s | maybe three or four kind of like smaller |
4181s | changes in it um including the one that |
4184s | I'm really excited about which will um |
4187s | it will shrink the you know the the gap |
4191s | between the best player and the worst |
4193s | player in the match |
4194s | um it'll shrink that somewhat |
4197s | um and the way that one works is |
4198s | actually very interesting because |
4200s | um what that change does is all it does |
4204s | is try to match parties that have a |
4207s | similar Delta with each other and you |
4210s | would say like oh you know that why do I |
4212s | care that only affects you know parties |
4214s | but it actually kind of shifts the |
4216s | entire curve of match quality in the |
4218s | direction in in a better Direction |
4220s | because it's like concentrating these |
4223s | parties that are a little bit wider |
4225s | together and then we can make more |
4227s | matches that have like only solos or you |
4229s | know uh you know Duos that are like well |
4231s | paired things like that |
4238s | I said it'll be more so in quick play |
4239s | but even in competitive like I think |
4242s | it's upsell what Diamond that we allow |
4245s | to like up matches up to two ranks so |
4248s | like you know if you're a gold level |
4250s | player you can play with your diamond |
4251s | level friend that's still a pretty large |
4253s | disparity you know that's a that's a |
4255s | much wider range than what I think what |
4258s | players would consider an ideal match |
4260s | right players definitely tell us like |
4261s | yeah |
4263s | sure they do and you know I I think |
4266s | ideally |
4268s | um you know we want to make matches that |
4269s | really have as as you know title abound |
4273s | as possible and so you know putting |
4275s | those parties together I think is going |
4276s | to do a lot to kind of improve |
4279s | um that average case yeah but that's you |
4281s | know that's an example of like I would |
4282s | say a smaller change that um you know |
4285s | we're thinking about those types of |
4287s | changes kind of constantly and um and |
4290s | then in addition to that we've got you |
4292s | know we have so many tunable parameters |
4294s | now on our Matchmaker you know you can |
4296s | kind of imagine it as like an instrument |
4298s | or something where like every little |
4300s | knob has to be kind of just right and so |
4303s | you know we've got probably hundreds of |
4305s | those at this point and we're kind of |
4307s | constantly adjusting those I would say |
4309s | you know last week we adjusted twice I |
4312s | think the week before it was like three |
4313s | times you know it's almost kind of like |
4315s | a daily uh thing sometimes especially |
4318s | you know after this the season three |
4320s | patch where we had a whole lot going in |
4323s | um and so there's kind of just like you |
4325s | know we really want you know every week |
4327s | matchmaking to get a little bit better |
4329s | and you know that's kind of the goal for |
4331s | for the foreseeable future yeah |
4334s | and I think one of the sort of important |
4336s | thing is |
4337s | we're not just looking at like the |
4338s | averages like it's very easy for us like |
4340s | from a data perspective to look at like |
4342s | hey here's what the average is but like |
4344s | the reality of matchmaking is something |
4346s | we've talked about together is |
4348s | the community's perception of |
4349s | matchmaking is usually defined by the |
4351s | absolute worst cases like those 95 those |
4355s | 99 percentile like cases and |
4357s | a lot of adjustments we're doing are |
4359s | just to the average but are just |
4361s | specifically like make those like worst |
4363s | cases better in fact you know we've |
4365s | talked about it how |
4366s | we're willing to give up a little bit of |
4368s | like quality for the average just to |
4371s | make those worst cases better so that's |
4374s | a lot of and that's like just one of the |
4376s | many things that we've been like trying |
4377s | to like tune around and |
4379s | this entire system is so complicated and |
4381s | we've added this like major third axis |
4384s | of quality not just in win expectation |
4387s | and uh actually that queue time but now |
4390s | there's this like a getting the roles to |
4392s | like you know match together like |
4393s | appropriately |
4395s | that finding the right knobs a lot of |
4396s | times we're like we'll make a change and |
4398s | we're like |
4399s | that didn't do exactly what we thought |
4401s | it does but it's like it's cool that we |
4403s | know that like if we mess with that knob |
4404s | but that's what it does because the |
4407s | system is so complex that like and |
4409s | there's just constraints everywhere |
4410s | inside by changing one constraint |
4412s | the thing you didn't think what that |
4415s | really meant something all of a sudden |
4416s | like become super important and like so |
4420s | but we're trying we're we're just like |
4422s | it's and a lot again it's just like |
4425s | almost daily like changes you know |
4427s | making sure it's like you know a lot you |
4429s | know we get immediate like responses |
4431s | from live Ops it's like uh we really |
4433s | mess with Cue times oops |
4435s | or or you know just we just get like a |
4438s | lot of like anecdotal feedback from |
4439s | everyone on the team and everyone on |
4441s | Reddit and like everyone like you know |
4443s | all the streamers like the entire |
4445s | community and all that feedback is super |
4447s | useful because again that's |
4449s | at the end of the day what we're really |
4450s | trying to do is just make these matches |
4453s | there's like the concept of mathematical |
4455s | absolute fairness |
4457s | which I think we we really like in |
4458s | OverWatch one like really concentrated |
4460s | on but like when OverWatch 2 we're |
4462s | really trying to say it's like all right |
4463s | there's more to match quality than just |
4465s | that |
4467s | and that's what you're seeing like you |
4469s | know with all these with these like roll |
4470s | Delta chaining out changes other changes |
4473s | that we're doing like in future like we |
4475s | know that |
4476s | you know there's a lot of like you know |
4478s | people saying it's like wow Q times an |
4479s | archaic can be really bad |
4481s | um yeah that's something we need to fix |
4482s | we're looking at that uh |
4485s | there's a lot there's like a there's |
4487s | there's other like things |
4489s | um we've been like sort of looking at |
4491s | it's like |
4492s | you know it's like if you play quick |
4493s | play and then go in competitive it's |
4495s | like the different like cues sometimes |
4497s | you have like different really different |
4498s | experiences and we want to make make |
4499s | them more holistic sort of view of like |
4501s | how skilled you are |
4503s | um there's a lot of things that we that |
4505s | we really want to improve going forward |
4506s | that again I think will no matter how |
4508s | you play OverWatch whether it's click |
4510s | play whether it's competitive whether |
4511s | it's whether you're in bronze whether |
4513s | you're in GM |
4515s | will give you a better experience |
4516s | through matchmaking yeah I really |
4518s | appreciate how uh your team seems to be |
4522s | taking like |
4523s | like a really analytical approach and |
4525s | part of that is the data that we see in |
4528s | simulations before things go live part |
4530s | of that is looking at data after things |
4532s | have gone live part of that is listening |
4534s | to community feedback and seeing what |
4536s | the player experiences part of that is |
4539s | US experiencing it ourselves it sounds |
4541s | like there's so many different routes |
4544s | and so many different avenues that |
4546s | affect our decision making and for me |
4549s | it's comforting to know that like this |
4551s | is a system that like we don't see |
4553s | ourselves like reaching the absolute |
4556s | perfect like this is something that |
4558s | we're going to continue to iterate on |
4560s | and will always be growing and |
4562s | developing as you said almost daily |
4564s | Matchmaker changes go in absolutely I |
4568s | think um you know like I I've thought of |
4571s | like what is the ideal you know |
4573s | Matchmaker honestly we are like so far |
4577s | away from from what that could be |
4580s | um you know it's really exciting because |
4581s | I think we just have you know we have |
4584s | like a road map that could span years |
4586s | that's like meaningful improvements |
4588s | every step of the way and and yeah you |
4590s | know it's really important to stay |
4591s | grounded I think |
4593s | you know like if you asked us a year ago |
4595s | what we thought would be important today |
4598s | in terms of matchmaking like we wouldn't |
4600s | have given the same answer |
4602s | um you know staying grounded like you're |
4604s | saying reading all the data |
4606s | um you know running hypotheses testing |
4609s | you know changes live |
4611s | um all of those are really good ways for |
4613s | us to you know test our theories you |
4616s | know we we um I I feel like our our hit |
4620s | rate on uh you know like if we make a |
4622s | guess of like oh we we think this is a |
4624s | good change I think maybe we get it |
4627s | right about 25 of the time or something |
4630s | like that |
4631s | um almost inevitably when we're when |
4633s | we're working on a change something |
4635s | about that is going to challenge our |
4637s | initial conception of of what we thought |
4639s | was going to happen and so |
4642s | um yeah very important to to yeah stay |
4645s | mindful of that it's part of the science |
4647s | right like Game Dev being like an art |
4649s | and a science like we're we're out here |
4652s | trying new things and not proving things |
4654s | to be true but finding out what's not |
4655s | true and then iterating off of that and |
4657s | I think that's really really neat and |
4659s | really interesting and |
4661s | um I also really appreciate you framing |
4663s | up Matchmaker from the perspective of |
4665s | different game modes |
4667s | um I know that for me when we first |
4670s | started having discussions about |
4672s | Matchmaker I was approaching almost all |
4674s | of those discussions from a competitive |
4675s | perspective but Matchmaker exists |
4678s | everywhere in OverWatch and I really |
4680s | appreciate you painting the different uh |
4683s | like places in game like how it affects |
4686s | quick play different than it affects |
4687s | competitive and |
4689s | um yeah that's that's really interesting |
4692s | for me just for the uh General |
4694s | understanding for everyone who might not |
4696s | be so familiar intimately familiar with |
4698s | all those different |
4700s | um systems in place for a Matchmaker how |
4703s | are you using the same system for each |
4706s | and every different mode or do you have |
4708s | to make different kind of systems or |
4710s | like cut something out of one to make it |
4713s | work for arcade to make it work for |
4714s | competitive Etc yeah that's actually |
4716s | that's a great question and actually |
4718s | that kind of speaks to kind of the |
4721s | complexity of the overall system so we |
4723s | do have a single converged architecture |
4725s | that runs across all of our cues the |
4728s | thing is you know because we want |
4731s | something different out of competitive |
4733s | role queue versus you know arcade |
4736s | um we have a lot of configuration that |
4739s | is kind of per queue and we're we're |
4741s | actually working on kind of a new layer |
4743s | of that even |
4745s | um that gives us even more fine-tuned |
4747s | control so |
4749s | um because yeah at the at the end of the |
4750s | day like there's some stuff that only |
4752s | runs for roll cue you know like |
4755s | ironically I made this roll Delta also |
4758s | work for non-roll cues because I really |
4761s | wanted to reuse the tech so the way it |
4765s | works for non-roll cues it's an |
4766s | interesting idea in roll cues you only |
4769s | ever have two of a single role on a |
4771s | given team at most but in non-roll cues |
4774s | you have you know five players of |
4777s | no role right |
4779s | um so I made the system care about the |
4782s | the highest and lowest MMR player of |
4785s | each team and and kind of in that way |
4788s | you can think of it as |
4791s | um you know a 5V it's like five of roll |
4794s | none versus five overall none and and |
4797s | you know we're still reusing the system |
4798s | it's it needs to be tuned differently |
4800s | but like we can kind of reuse some of |
4802s | the text so you know sometimes it's kind |
4804s | of like you know can we can we find a |
4807s | way to make the same system make sense |
4809s | across multiple cues because it you know |
4812s | there's kind of actually a |
4813s | configurability nightmare if we have to |
4815s | manage like 30 different cues all of |
4818s | which have subtly different Behavior |
4820s | it's like you know if there's a problem |
4822s | with one we're never even gonna find |
4823s | that because it's like a needle in a |
4825s | haystack at that point yeah so yeah it's |
4827s | it's a that's a great question and it's |
4829s | it's a major |
4831s | um it's something we think about a whole |
4832s | lot |
4833s | awesome if I'd love to hear like any |
4836s | final thoughts on matchmaking or if |
4838s | there are any other questions that you |
4840s | have Zoe um or anything that you you |
4842s | guys want to input otherwise I think I'm |
4845s | gonna shift gears and talk about |
4847s | competitive I I'd love to dive into |
4850s | competitive ones yeah because all right |
4852s | let's go um so yeah let's do it awesome |
4856s | so |
4857s | um first I'd like to do kind of like a |
4860s | retrospective on the season three |
4862s | updates that we saw and kind of explain |
4863s | like why uh we made those changes |
4868s | sure so season three |
4871s | uh sort of we've done a few things like |
4874s | for competitive play |
4875s | um you know we changed the UI a little |
4877s | bit we made the um we added specifically |
4880s | like well season three I guess |
4882s | this is where like |
4884s | so many patches |
4885s | almost right in my head we've uh added |
4889s | the ability to see |
4890s | um your competitive update screen at |
4893s | like any time uh and then the mid-cycle |
4895s | patch that we're adding in like newer |
4897s | buttons for that to make it like easier |
4898s | to get to like made that like you know |
4901s | really pretty |
4902s | um |
4902s | then |
4904s | sort of like we've also like made some |
4906s | changes in anticipation of what we're |
4909s | doing in season four |
4910s | where the seasonal rank Decay we didn't |
4914s | put as much decay in we yeah we nerfed |
4917s | it yeah it was Morgan said it best we |
4918s | basically nerfed it uh because in season |
4921s | four we're going to be removing that so |
4925s | we didn't want to make it as much of a |
4928s | you know what what just happened you |
4930s | know it's like one of the things that |
4931s | we've learned about like competitive |
4932s | play like an OverWatch 2. we made some |
4934s | pretty big changes |
4936s | and I honestly think we just didn't |
4938s | communicate those super well so like one |
4941s | of the big sort of takeaways from the |
4944s | competitive play leading up is just |
4946s | people don't understand how the system |
4947s | works so |
4949s | a lot of the changes that we're working |
4950s | towards especially like with season four |
4952s | is |
4953s | making things more clear for players |
4956s | um yeah getting you know |
4958s | because people want to have confidence |
4960s | in |
4962s | you know what what they're playing you |
4964s | know what they want to be able to set |
4965s | goals they want to be able there's all |
4966s | these things they want to do and when |
4968s | they don't understand how the system |
4969s | works it sort of works against all that |
4971s | so that's one of the big things oh and |
4973s | yes season three we also changed the |
4975s | number of like wins from seven to five |
4977s | right |
4978s | um which that wasn't that was an |
4980s | interesting change and that was |
4982s | something where it was originally seven |
4983s | for like a very interesting reason that |
4985s | we I think we really talked about too |
4987s | much what we shipped with as far as |
4989s | competitive play for OverWatch to wasn't |
4992s | everything that we wanted to ship with |
4994s | one of the major |
4995s | um features that we were looking forward |
4997s | to was this concept of being able to |
4999s | provide players with periodic |
5002s | uh information about how well they're |
5004s | performing |
5005s | even potentially like on a hero basis so |
5007s | the idea of like hey you have seven wins |
5009s | out of those seven wins you played like |
5011s | these Heroes hey look at these how well |
5013s | you performed on these Heroes let's say |
5015s | you played like |
5016s | um Soldier 76 you know Cassidy and |
5019s | sojourn well you played the you played |
5022s | those Heroes during those wins like |
5024s | let's look at your accuracy it's like |
5026s | are you preferring better or worse than |
5028s | your particular than like other players |
5030s | that you're like skill rank and we |
5032s | really like that concept because we want |
5034s | to provide information to players to |
5036s | help to help give them goals to provide |
5039s | them with like information about hey how |
5041s | do you get better at competitive at |
5043s | um you know competitive OverWatch too |
5044s | right because we went from like we went |
5047s | from Sr to something extremely granular |
5050s | to something that was a little more |
5053s | broader with the skill tier and division |
5055s | so it sounds like you really want to |
5058s | contextualize like not just take away |
5060s | that granularity for the sake of making |
5062s | it less granular but also it's very |
5065s | difficult or it was very difficult to |
5067s | contextualize Sr like what you're |
5069s | talking about with like how well am I |
5071s | performing compared to other Cassidy |
5073s | players like that that context it wasn't |
5077s | there for Sr and it sounds like that's |
5080s | something we're moving towards yeah and |
5083s | becoming better at over at OverWatch too |
5086s | like if you're the type of player who's |
5088s | actively trying to improve that's a |
5090s | long-term process |
5091s | and we really didn't feel like |
5093s | and that's our value that was changing |
5094s | after every game was really sort of |
5096s | setting the right signals like for that |
5098s | because |
5099s | when you think about it it's like |
5101s | you know if batch tracking is doing an |
5102s | awesome job you're going to win and lose |
5104s | about 50 of the time right and it's like |
5107s | if you're you know it's like if you're |
5108s | improving it's like that 50 50 will |
5110s | still sort of be there but you'll the |
5112s | quality of the teams that you're playing |
5113s | with like the quality of the opponents |
5115s | it sort of like goes up over time like |
5117s | as you've improved so |
5119s | that was another sort of big thing about |
5121s | sort of changing the entire like Sr |
5124s | change after rematch to this competitive |
5125s | update system is |
5127s | really just sort of like framing that in |
5129s | terms of |
5131s | a longer term Journey for for players |
5134s | not concentrating on like individual |
5136s | matches which are highly variable for a |
5139s | huge number of reasons I mean that's you |
5142s | know we talk about OverWatch 2 |
5144s | competitive play there's just I think |
5146s | this idea that it's like oh I'm gold |
5148s | three so I'm going to play at a gold |
5150s | three like level whatever that is |
5152s | every single game and in reality it's |
5155s | like |
5155s | all right some games I like got super |
5157s | lucky hit a bunch of headshots other |
5159s | games it's like oh my goodness I made |
5161s | this most horrible misposition |
5163s | and it was just like we lost like we |
5165s | lost a team fight |
5167s | there's a lot of variation variation in |
5169s | that which is what I like which again we |
5171s | sort of just wanted to spread out like |
5172s | the updates because |
5174s | again recognizing that like any |
5177s | individual game |
5178s | doesn't actually mean that much really |
5180s | right yeah um so |
5184s | yeah I mean so that was sort of a big |
5187s | thing and again to get back to the like |
5189s | another question we get from the |
5191s | community is like why is it wins and not |
5193s | wins and losses well because we wanted |
5195s | to do this report |
5196s | one thing we know just from looking at |
5199s | Stats is the stats you have when |
5201s | whenever you win games are completely |
5204s | different than when you lose them |
5205s | and if we're going to do any kind of |
5207s | like apples to apples comparison you |
5209s | know because again these reports were |
5210s | like something that like over time you |
5212s | would see them change and you get you |
5215s | you get you know positive reinforcement |
5217s | saying it's like oh I'm getting better |
5218s | at this you know because the idea was |
5220s | you see this report it's like oh this is |
5222s | something I can get better at you |
5224s | actively try to get better at that you |
5226s | know you play not just to play but you |
5228s | play with like oh I'm gonna like work on |
5231s | my like my headshot accuracy or I'm |
5233s | going to work on you know particular |
5235s | like you know Sportability making sure |
5237s | that's more effective you know making my |
5239s | old support making my olds better |
5241s | um that I think was like interesting but |
5243s | unfortunately we didn't ship with that |
5245s | so that sort of like |
5247s | we had to like read it's like okay if |
5249s | we're not gonna do that even though we |
5250s | want to still do it in the future but |
5252s | like for now |
5253s | let's like me you know seven seven wins |
5256s | felt like too much |
5257s | so that's why I changed it to five right |
5260s | um and that's sort of where we are of |
5262s | season three |
5264s | and now season four |
5266s | we can start talking about that |
5268s | are you ready ah yes yes okay cool cool |
5271s | season four so season four the big |
5274s | change that we're doing there is we're |
5276s | going to be removing the seasonal rank |
5277s | Decay |
5278s | um |
5279s | so |
5280s | that was another aspect that we were |
5282s | trying to add in with OverWatch to |
5286s | um going back in time to OverWatch one |
5288s | um |
5289s | we gotta leave a lot of feedback from |
5291s | players like over time where they felt |
5293s | like there wasn't really that much |
5294s | changes between Seasons |
5296s | [Music] |
5296s | um |
5297s | there's a lot of like sort of pent-up |
5299s | demand for this idea of it's like I want |
5302s | sort of this |
5303s | I'm going to use a strange word because |
5305s | if I say if I say MMR reset Morgan will |
5308s | freak out because that's like an almost |
5309s | horrible idea ever don't do it but |
5311s | there's this idea of it's like |
5313s | a new season starts I want to like you |
5315s | know I want to refocus my efforts you |
5318s | know I want to like start this journey |
5319s | again towards working towards Greater |
5321s | Heights of my skill |
5323s | and there was no real way to do that |
5326s | because for most players what what |
5328s | actually happens is they reach a certain |
5330s | skill |
5331s | and then they'll just sort of change |
5333s | around like different like you know |
5335s | tiers or now sub you know divisions |
5338s | um |
5339s | so the idea was we're going to like |
5341s | lower your |
5343s | lower your division or your skill tier |
5345s | and division at the beginning of the |
5346s | season |
5347s | to provide that sort of moment of it's |
5349s | like oh I've got another goal to work |
5350s | towards I can start this journey again |
5354s | so that didn't quite out work the way we |
5357s | wanted to for a couple of big reasons |
5360s | um and this is the big Maya culpa moment |
5362s | of you know of what shoe competitive |
5364s | play one |
5366s | we didn't again we didn't explain what |
5368s | was happening |
5369s | uh and that was sort of like sort of a |
5371s | big thing people came away it's like |
5373s | wait a new season just started why is my |
5375s | ranks much lower did I suddenly get |
5377s | worse I don't I don't feel like I got |
5379s | any worse than like OverWatch why did |
5381s | this just go down and |
5384s | so we didn't communicate what it was |
5385s | trying to do and the other thing was for |
5388s | this sort of like system for to create |
5390s | that Journey |
5392s | have making doing this every two months |
5394s | is probably too short |
5396s | um we looked a lot of data with regards |
5398s | to like how many people how many games |
5400s | of competitive people play there's a lot |
5402s | of them that maybe only play like 15 or |
5404s | 20 games which isn't even like |
5407s | enough the way we tuned it it was going |
5409s | to take more than that to work your way |
5411s | back up to your quote-unquote real |
5415s | um skill skill level |
5417s | and because they didn't get back to that |
5420s | instead of this like virtuous journey of |
5422s | like hey I'm like gonna get back I'm |
5424s | gonna like restart I'm gonna like you |
5426s | know try hard this time I'm gonna get |
5428s | like so much better at OverWatch too it |
5430s | was like we lowered them back down they |
5432s | got back up a little bit and then we |
5433s | lowered it back down and got about it |
5434s | then we lowered it back down |
5436s | and it didn't feel virtuous it felt |
5438s | punitive which is like not at all we |
5441s | didn't want that at all |
5442s | um so |
5444s | for those reasons and again just to make |
5446s | just to make things clearer |
5448s | we're just removing it so |
5450s | when season four starts |
5453s | um when you first go to play competitive |
5454s | play there'll be a little there'll be a |
5456s | message uh that pops up saying hey we're |
5458s | removing this we're also just sort of |
5460s | removing like all the effects of it so |
5462s | we're like resetting everyone back to |
5463s | like what their quote unquote |
5466s | actual like skill is so we'll pop a |
5469s | little up dialogue indicating hey this |
5470s | is what your previous ranks were back in |
5472s | season three and here's your new ones in |
5474s | season four |
5475s | again with the idea that hey we need to |
5478s | fully explain what's happening because |
5479s | when we don't explain things uh |
5482s | it's not good I'm glad that you know you |
5486s | guys are here to talk to us about this |
5488s | like being here and communicating with |
5490s | with us and and the players is it means |
5493s | a lot yeah I mean same sentiment here |
5495s | like the transparency is really really |
5497s | so important especially with systems |
5499s | which are just so complicated so |
5502s | intricate and as you said you turn one |
5504s | knob and then everything else changes so |
5505s | like there is a lot of communication |
5507s | which needs to happen in order for the |
5509s | people who play the game to really |
5510s | understand why things are happening the |
5513s | way they are I do want to ask though |
5516s | why do people or shouldn't people want |
5520s | that full MMR reset because still like |
5523s | that that is what people like I see that |
5525s | so often in the community it's like why |
5526s | don't I just reset it please tell them |
5528s | why that's not what they want yeah |
5531s | that's a it's a good question so like |
5533s | you know I can understand the sentiment |
5535s | right it's like you know if you feel |
5537s | like your MMR is not where it should be |
5540s | or something like that you're like ah |
5541s | just reset it all right |
5543s | you know it's really actually kind of |
5544s | fascinating we have millions of players |
5546s | and if you think about it most of those |
5549s | players have never actually played each |
5551s | other right so |
5553s | um by default if we throw all of the |
5558s | ratings that we have out we're losing a |
5561s | whole lot of information like our |
5563s | current view of of everybody's skill you |
5565s | know your individual skill may not be |
5567s | calibrated but the system as a whole |
5570s | embeds a whole lot of knowledge about |
5572s | the relationship between all of these |
5574s | players most of whom who have never |
5575s | actually played against each other you |
5577s | know like in in chess we have you have |
5579s | like ELO or something |
5581s | and you know the community of chess |
5583s | players is somewhat small and like you |
5586s | know you'll go to these tournaments and |
5587s | like everybody's kind of playing each |
5589s | other and it you know it kind of makes |
5591s | sense |
5592s | um but in in OverWatch like we don't |
5595s | really have that like you know unless |
5597s | you're on top 500 or something then |
5599s | you're actually playing with with other |
5600s | you know all the other people |
5602s | um that's a separate problem but uh but |
5605s | generally with ratings it's like you |
5607s | know we really want to have some view of |
5611s | everybody's skill and even a rough view |
5613s | is much better than no information |
5614s | because what happens if we throw all of |
5616s | it out is you'll get games where like |
5619s | bronze players are playing against top |
5622s | 500 players |
5623s | and we've actually done that internally |
5625s | right because like we have bronze |
5627s | players on our team and top 500 players |
5629s | on our on on Team Four right and so like |
5631s | in those play tests like those are not |
5633s | fun to play tennis yeah uh it turns out |
5636s | that uh that matches only really feel |
5639s | fun in a very very narrow difference of |
5643s | of skill otherwise you know one team's |
5645s | just gonna Stomp the other team you know |
5646s | OverWatch is a very decisive game and |
5649s | being like Platinum versus gold like |
5652s | that's a very noticeable difference in |
5654s | skill |
5655s | um and so you know all of that |
5656s | information that the ratings have in |
5658s | aggregate it's extremely valuable like |
5661s | you know whenever somebody asks about |
5663s | MMR I feel like No don't touch the mmrs |
5667s | break the entire matchmaking if you |
5669s | if you throw out the enemy |
5671s | yeah because the information is not |
5673s | going to be there that means that the |
5674s | roll Delta graph is going to go all the |
5676s | way up again |
5678s | exactly but not in a good way yeah I |
5681s | think this sort of this there's a little |
5683s | bit of like misunderstanding about like |
5684s | how long it would take for the system to |
5686s | readjust |
5688s | um |
5688s | you know when these conversations come |
5690s | up the people will be like oh it'll just |
5691s | probably take a month and it's like |
5693s | no you don't understand like this months |
5697s | like years like yeah it's it's actually |
5699s | interesting in the sense of you know |
5701s | with OverWatch 2 coming out you know |
5703s | we're going free to play it's like all |
5704s | these like new players |
5706s | there is a little bit of like our |
5707s | matchmaking has been has technically |
5710s | gotten worse just because we have this |
5712s | mass number of people that we don't know |
5714s | much about them right |
5717s | you know people try to compare like how |
5719s | OverWatch one matchmaking was right |
5721s | before we shipped OverWatch too and |
5723s | that's where we had players who had |
5726s | played a ton of games they played a lot |
5728s | of Heroes they've played all the maps |
5731s | they've gone through like multiple metas |
5732s | they were all like very knowledgeable |
5735s | and like the matchmaker's view of their |
5737s | skill was |
5739s | reasonably accurate but when you bring |
5742s | in new players think about like all the |
5743s | things they potentially like have to |
5745s | have to learn again all the heroes on |
5747s | their particular role within just one |
5748s | role all the different Maps |
5751s | um again like they haven't been through |
5752s | like multiple different Meta Meta swings |
5755s | and |
5756s | when we talk about like what the |
5758s | matchmakers view of the skill is |
5760s | you know it's not |
5762s | it's not as simple as like hey I won |
5764s | game it like I get better I lose game it |
5767s | is but like the reality is the amount is |
5770s | like different but like the other thing |
5771s | is that like as a new player |
5774s | you know |
5776s | the Matchmaker can think you're of a |
5778s | certain skill because you've been |
5779s | playing let's say like you're you've |
5781s | been playing soldier 76 and you've been |
5782s | having a good amount of success right |
5784s | put your new player you're like all |
5786s | right I feel like I've I'm doing well on |
5788s | Soldier 76 I'm going to switch to |
5790s | Cassidy |
5791s | all right at that point the matchmaker's |
5794s | view of their skill is wrong again |
5796s | because you're on a completely like new |
5798s | hero right we also see other like sort |
5800s | of changes again and it's been great to |
5803s | see all this like we love it you know |
5804s | all these like new friends bringing |
5806s | other friends to play OverWatch to who |
5808s | hadn't played it before but when you |
5810s | play with like different groups of |
5811s | friends |
5812s | that can sort of change |
5814s | what your skill is because you might |
5817s | play with your friends differently than |
5818s | you would play like solo Warriors like |
5820s | super try Harding right yeah like the |
5822s | Matchmaker like all it really you know |
5824s | it really just has this like one view of |
5826s | what you are |
5827s | and with new players it's |
5829s | Lots can happen to that so |
5832s | that's again why we're like so scared |
5834s | with us about like you know like this |
5837s | talk of like MMR |
5838s | um MMR reset because |
5841s | it would just it would it would |
5843s | literally take like like |
5845s | you mentioned it years yeah really |
5847s | delicious build up this like combined |
5849s | knowledge about like all the players and |
5852s | when you think about like all the |
5853s | different accounts they have on all the |
5854s | different roles |
5856s | um it's yeah it's it's mind-boggling |
5860s | I'm not sure if we want to share |
5862s | information on this but I'm curious if |
5864s | you can give us a little more insight |
5865s | onto into like new players versus |
5870s | veteran players like you're mentioning |
5872s | like for OverWatch one we had all this |
5874s | like data and then we launched OverWatch |
5876s | too and a lot of these players carry |
5878s | over but we also have new maps we have a |
5881s | new game mode we have new Heroes and we |
5883s | have a lot of new players so I'm curious |
5886s | what your Insight is there yeah that's |
5889s | interesting like we definitely like some |
5891s | new players are something we've |
5893s | definitely like looked at pretty closely |
5894s | like we want to try to |
5896s | again because the Matchmaker has such a |
5899s | so little understanding of you know just |
5901s | how good they are and everything that |
5903s | matchmaker is based upon is this idea of |
5905s | like how good are you like so we can |
5908s | match you up with other people of that |
5910s | same like sort of skill |
5911s | and that's something where |
5913s | we've like try to do our best we've done |
5916s | things with regards to trying to like |
5918s | start them off at like a better place so |
5920s | they're not like like the situation we |
5922s | don't want is like hi I'm a new player |
5923s | OverWatch and we totally like |
5925s | misevaluate your skill and you lose a |
5927s | bunch of games because we think you're |
5929s | way better than you are so |
5932s | we've done work on that regard yeah we |
5935s | did a whole lot actually like you know |
5937s | when we were working on the launch of |
5939s | OverWatch 2 we knew you know that we |
5941s | were going to have a ton of new players |
5943s | show up and we also knew that we were |
5944s | going to have a ton of people who had |
5945s | like played OverWatch years ago come |
5947s | back and so you know we built systems |
5950s | specifically to |
5953s | um you know try to make those players |
5954s | experience as good as possible and you |
5956s | know as a result of that make everybody |
5958s | else's experience playing with those new |
5960s | players as good as possible too |
5962s | um so yeah we did the um you know we |
5965s | very carefully tuned you know kind of |
5967s | the new player experience in terms of |
5969s | like what their ratings look like when |
5970s | they enter the game kind of how much |
5972s | uncertainty we have about them how fast |
5974s | we move their ratings one one thing |
5976s | that's like the bane of my existence is |
5978s | when people do like unranked to GM you |
5981s | know like they make a fresh account and |
5983s | then like play to GM you know because |
5985s | like their first several matches are |
5986s | really bad so like we added a system to |
5989s | try to kind of like hey you're you're |
5991s | like on a win streak that we don't |
5993s | expect you to be on your MMR should |
5996s | probably be way different than what it |
5997s | is |
5998s | um so thankfully you know we've like |
5999s | Shrunk the number of matches that it |
6002s | takes for somebody to like kind of |
6003s | calibrate when they're just coming into |
6005s | the system by quite a bit |
6015s | yeah I |
6017s | yeah and then also like for returning |
6020s | players you know we um we did a lot of |
6023s | work to kind of model like we we kind of |
6025s | do Decay players when they come back if |
6027s | they've been gone for a really long time |
6029s | just to try to kind of like ease them a |
6032s | bit in and then also we we increase our |
6034s | uncertainty about them too so it's like |
6036s | we're kind of like much more willing I |
6038s | mean we treat them a little bit more |
6039s | like a new player because it's like you |
6041s | know we don't know what you've been |
6042s | doing for the past two years maybe |
6043s | you've been you know not playing video |
6045s | games at all maybe you've been like |
6046s | playing a different shooter and you're |
6048s | even better right so yeah a whole lot of |
6051s | work that went in kind of at launched to |
6053s | really try to get ahead of a lot of |
6055s | those things |
6056s | but you know still a whole lot of new |
6058s | players a whole lot of new uncertainty |
6059s | in the system yeah and we've been |
6061s | dealing with it I think um you know |
6064s | that's probably the type of thing |
6065s | that'll just get better over time |
6069s | just yeah it's uh the the answer to |
6071s | every question is always just play more |
6073s | games yeah |
6074s | yeah |
6077s | and to give the Matchmaker a better like |
6078s | view of like how good you are yeah it's |
6080s | like that's sort of like the two things |
6082s | that sort of solve all like you know |
6083s | matchmaking things like play more games |
6085s | and have more people playing games so we |
6087s | can there's even more people to |
6088s | potentially match you with yeah right |
6091s | I I was just wondering what the what the |
6094s | team's stance is on actually displaying |
6096s | uh |
6097s | the team ranks like we're heading into a |
6100s | game |
6101s | um in competitive like is that something |
6104s | discouraging because you know like very |
6107s | often uh back when there was like the |
6110s | team MMR in the versus screen right at |
6111s | the beginning uh your team immediately |
6114s | if they have a lower Sr would go like oh |
6116s | well we lose so like immediate defeat a |
6119s | kind of attitude |
6121s | um what's this the stands from your team |
6123s | about displaying uh those kind of sure |
6126s | mmrs so that brings us to another change |
6129s | in season four haha there he goes |
6133s | um not quite like everything you I think |
6135s | you're asked for but like we are going |
6136s | to display like |
6138s | um before like a competitive match and |
6140s | that likes our team screen at least what |
6142s | the average like for this particular |
6144s | like game is so you know you can see |
6147s | it's like hey this is a |
6149s | Platinum 4 match |
6151s | um it just to give you sort of like a |
6154s | better idea it's like am I maybe I'm a |
6156s | little bit better than everyone in here |
6158s | or maybe I'm a little bit worse you know |
6159s | it gives you some information about that |
6161s | um again like |
6163s | we definitely like seeing the feedback |
6164s | with regards to like hey I just want to |
6166s | see like what everyone's like sort of |
6167s | values are |
6169s | um |
6169s | some of that I think is is positive like |
6172s | we we do |
6173s | we do get very pointed feedback about |
6176s | the Matchmaker when that when that |
6177s | happens which that you know that doesn't |
6180s | necessarily that doesn't really bother |
6182s | us but I think it's more just that we |
6185s | don't want like players to necessarily |
6186s | like hone in on like |
6189s | precisely like what those like values |
6191s | are because a lot of times they'll sort |
6193s | of like they'll make judgments about the |
6195s | match before they've even played it yeah |
6196s | like a self-fulfilling prophecy right |
6198s | and that could be like that could be |
6200s | really negative right it's like because |
6201s | again like we're not going to put you in |
6203s | like a horrible like a super horrible |
6205s | match you know Morgan was talking |
6206s | earlier about it's like you know 55 45 |
6209s | like it sounds horrible but it's like no |
6211s | you win 45 of the time right and it's |
6213s | like |
6215s | it's the kind of thing of like it's |
6216s | still work it's still worth going in and |
6218s | trying your hardest yeah right it's |
6220s | something that you know I think over |
6221s | time and especially like as we're making |
6222s | like all these like matchmaking changes |
6224s | you know and we're trying to do like all |
6227s | these like tuning things it's something |
6228s | that you know we might like put back |
6230s | like in the future like but for now |
6232s | um we're gonna like take you know baby |
6234s | steps which is like putting in the like |
6236s | you know what is the like sort of rank |
6238s | of this particular match yeah we've been |
6240s | talking about it a lot I mean we've |
6242s | there's a challenge because it's like we |
6244s | want to show players more about matches |
6246s | right like we want to kind of show you |
6248s | know how good these matches are but it's |
6251s | hard because it's like finding ways you |
6253s | know like the charts that I was showing |
6254s | earlier like those are kind of hard |
6256s | especially if like it's on a screen that |
6258s | you're just seeing for a few seconds |
6260s | like you know part kind of understanding |
6263s | some of that information can be |
6264s | difficult and also yeah like Scott was |
6266s | saying we don't want to create a |
6268s | situation where somebody's going to like |
6270s | you know not try their hardest because |
6273s | we've already lost |
6275s | like it's interesting like in in |
6276s | Starcraft actually they used to show |
6278s | this where it was like oh you know your |
6280s | your team or your you know you are |
6282s | favored to win or the enemy is favored |
6284s | to win and like that you know that kind |
6286s | of works in a 1v1 game where you can be |
6288s | like okay you know I might lose this |
6290s | match but in a team match where it's |
6292s | like your team is not favored to win you |
6295s | know then it's like well what you know |
6297s | are my teammates bad like |
6303s | though like that's a that's a Content |
6306s | booster yeah yeah |
6308s | so if we could only get the good the |
6310s | good side of it yeah if we can only get |
6312s | the positive without the negative then |
6314s | sure but yeah again it's it's hard to |
6317s | it's hard to it's hard to talk about a |
6319s | lot about like magic like this is the |
6321s | matchmaker's view of your quality and |
6323s | like |
6324s | we can't just hit you with a bunch of |
6326s | math and like weird numbers that you |
6328s | know like that we're looking at and |
6330s | because again like |
6332s | we really want to make competitive play |
6334s | understandable right you know |
6336s | so that's |
6338s | so we just have to like we were just |
6339s | we'd have to be like be very careful of |
6341s | what we display to players that they |
6343s | don't inadvertently |
6345s | you know take that information like the |
6347s | wrong way you know it's like yeah |
6349s | because so much of it needs to be |
6350s | contextualized right like it's so much |
6353s | information yeah and again it's like |
6356s | none of these numbers are absolute at |
6358s | the end of the day right it's like this |
6361s | is the matchmaker's current like best |
6363s | view of like what this match is but it's |
6365s | like |
6366s | who know it's like maybe like like once |
6370s | it makes the match what everyone plays |
6372s | has a way bigger effect than like just |
6375s | putting the people in it's like is |
6378s | everyone playing like the super like |
6380s | hard meta like picks what if someone's |
6383s | like trying something like different you |
6385s | know it's like there's all these like |
6386s | different effects that once you actually |
6387s | get into the match |
6388s | that yeah it's yeah yeah we've been |
6391s | trying to study that a lot lately |
6392s | actually yeah like what causes one-sided |
6395s | matches you know because it doesn't feel |
6398s | fun to like lose and you know you feel |
6400s | like you got stomped and people blame |
6402s | the Matchmaker for that yeah 100 and |
6405s | it's funny you know like we've been |
6406s | trying really hard to you know like |
6408s | really the problem there is like what |
6410s | signal can we see that is actually |
6412s | predictive of like a Stomp and so far we |
6416s | really haven't come up with a whole lot |
6418s | we found a few things you know that |
6419s | we've been able to kind of fix at the |
6421s | edges but kind of the core problem just |
6423s | seems you know like if if one person's |
6425s | like having an off day 45 is not you |
6430s | know you don't win a 4v5 so and even if |
6433s | and again even if it's not even if |
6435s | someone's not having a a bad day it's |
6438s | like |
6438s | you know it's like two teams trying |
6440s | their hardest like someone's gonna win |
6441s | that first team fight yeah you know and |
6444s | then I drove OverWatch it's like you win |
6445s | the first team fight that gives you a |
6446s | little bit of like you know ultimate |
6448s | economy Edge that potentially like gives |
6450s | you like an advantage like on the second |
6451s | you know on the second team fight like |
6453s | there's there's things |
6456s | that again the Matchmaker can't solve |
6458s | all those like kind of problems |
6460s | um it really can't unfortunately you can |
6462s | see it on like control Maps where like |
6464s | you know on on point on the first point |
6466s | like one team might just like roll the |
6468s | other team and it's like oh my God like |
6470s | this is such a one-sided match and then |
6471s | the you know the next point it happens |
6474s | the other way and it's like yeah okay |
6475s | wait you know that happens all the time |
6477s | and uh you know so it's like in that |
6479s | case that's probably actually a balanced |
6481s | match |
6483s | I mean you can only do so much math at |
6485s | the end of the day it's people playing |
6487s | right and it turns out humans are not |
6489s | super consistent no we're not robots |
6492s | right I think I am excited for it though |
6494s | I think yet oh my goodness |
6498s | um for season four aren't we providing |
6500s | more uh context to the competitive |
6503s | updates to kind of help people |
6504s | understand how they're doing |
6508s | thank you Jody yes we are |
6511s | uh so another thing we've been asked for |
6514s | is hey we show you how many wins you |
6517s | have give me see how many losses dies |
6519s | and we're gonna provide that so again |
6522s | and not providing that is one of the |
6524s | things that made it seem like you know |
6527s | it's like a competitive update happens |
6528s | and sometimes I went up a division |
6531s | sometimes I went down a division and a |
6533s | lot of that was confusing because we |
6534s | didn't actually show you like the losses |
6536s | and |
6538s | um yeah and again because because we're |
6540s | all human |
6541s | we sometimes forget about losses just |
6544s | like we forget about like that leave |
6545s | that we might have had because you know |
6548s | because I you know with family emergency |
6550s | or something like that you know and we |
6552s | do penalize for leaves as well |
6554s | um so |
6555s | yeah we're including like more |
6557s | information there specifically like the |
6559s | the losses and ties and the the |
6562s | breakdown of where you are oh yes well |
6564s | okay |
6566s | and specific not so much like so once |
6568s | you actually complete an update |
6570s | um |
6571s | again because like within a division |
6573s | there's actually like a there's a fair |
6575s | amount of wiggle room there we got |
6577s | feedback saying it's like hey I went |
6580s | I went five and four |
6582s | and I didn't go up I didn't go up a |
6584s | division it's like |
6585s | blizzard why |
6587s | um oh it's because it's like okay you |
6589s | were actually like just barely into one |
6591s | Division and like you actually like you |
6592s | did go up it's just it was in all within |
6594s | one division so again to try to make |
6597s | things like more clear about like what's |
6599s | going on and that like yes |
6601s | it's like your five and four record does |
6603s | mean it went up at the point where you |
6606s | actually get that like get that rank |
6608s | update after the five wins we will be |
6610s | showing you like where you are within |
6613s | that particular division |
6614s | and |
6616s | if you're someone who's between like gm1 |
6619s | and like the bottom of top 500 because |
6622s | there's this point where it's like |
6623s | okay how far do I need to get before I'm |
6626s | actually in top 500 and that's a not a |
6629s | lot of people but it's like enough |
6629s | people we've definitely heard the |
6630s | feedback that's like I want to know like |
6632s | how close I am also indicating how close |
6635s | you are to achieving top 500 and I have |
6638s | three percentage so again more |
6640s | information about like your updates more |
6643s | clarity as far as |
6645s | here's how you played and these are the |
6647s | results |
6649s | I really appreciate that and and |
6650s | something else that we're doing for |
6652s | season four that I'm really excited |
6654s | about are the competitive reward changes |
6659s | um so can we talk a little bit about |
6660s | like what that's going to look like and |
6663s | why we made that decision to change |
6665s | competitive Rewards |
6666s | yeah so we're changing the competitive |
6669s | rewards uh and again to be specific here |
6671s | these are these uh competitive points |
6673s | that you're getting for playing |
6674s | competitive uh competitive OverWatch |
6676s | that you use to purchase the golden guns |
6680s | um this isn't anything with challenges |
6681s | uh specifically but like |
6684s | um one thing we've |
6686s | always noted is that like if you're |
6688s | playing in bronze we were like the |
6691s | difference between like how many |
6692s | competitive points we give you for like |
6693s | a bronze or even like a gold and like |
6696s | someone who's a grand master |
6698s | was really huge |
6700s | so a lot of what we're doing is just |
6701s | like providing uh more competitive |
6703s | points at the end of the season for |
6706s | everyone like up through like up through |
6708s | Grand Master Grand Master's going down a |
6710s | little bit like at the |
6712s | um at the end of the season but |
6714s | the flip side is we're also just we want |
6716s | to reward people who play more like |
6718s | during the season |
6719s | um and we definitely like saw sort of |
6722s | thing of it's like oh |
6724s | it's like I've got my like go three rank |
6726s | or I'm done like there's no real reason |
6728s | to play like competitive play anymore |
6729s | it's like hey |
6731s | um if you enjoy playing if you enjoy |
6733s | playing OverWatch it's like hey here's |
6735s | some more competitive points uh so we're |
6737s | upping the number of competitive points |
6739s | you get for a win and also Tai um |
6743s | because you do get a little you do get a |
6745s | little bit like for a tie we're upping |
6746s | that as well again just to |
6749s | um you know give people a little bit |
6751s | more hey if you're playing a lot we want |
6753s | to like respect the fact that you know |
6755s | you are playing a lot because hey by |
6757s | playing a lot you're making the |
6759s | Matchmaker better yeah |
6764s | that's awesome I know another thing |
6766s | that's coming in um |
6769s | actually this is the mid-cycle patch |
6772s | this is live right now but we also have |
6774s | competitive mystery Heroes coming to the |
6777s | game |
6778s | oh boy competitive Mr Heroes we are like |
6781s | yes we are really excited about this |
6784s | because |
6786s | um |
6787s | the chaos is going to be so Exquisite I |
6791s | know |
6792s | Scott to like explain competitive |
6795s | mystery Heroes and he just said it's |
6798s | chaos yeah it was just gonna be great |
6800s | you know it's the kind of thing of like |
6802s | you know it's like people like really |
6804s | like trying hard and like trying to win |
6806s | this but at the same time it's like |
6808s | they have no control over the heroes |
6809s | they have no control over there like um |
6812s | um their team compositions so in some |
6815s | ways it's actually the people who do |
6816s | well on competitive mystery Heroes we're |
6819s | gonna like Flex their like skill in a |
6821s | way that like actually normal |
6822s | competitive play doesn't because there's |
6823s | going to be super flexible and you gotta |
6826s | be good at everything you got to be good |
6827s | at everything in all situations |
6829s | all the heroes all the maps but that |
6833s | being said we are like doing something |
6834s | we do know that like competitive mystery |
6836s | Heroes like it's like we love chaos but |
6838s | like we don't like just there are limits |
6840s | to our like our levels of Chaos |
6842s | we did put another rule that basically |
6844s | says you can only like have as when it |
6847s | randomizes the heroes it's only going to |
6849s | create team composite like the maximum |
6851s | of three if any role |
6853s | um so that was again sort of like |
6855s | prevent some of the like the worst or |
6857s | weirdest compositions |
6860s | um and then also we are doing um because |
6864s | this isn't like a roll cue setup this is |
6867s | also going to have the change that we |
6868s | made with regards to tanks and open q or |
6871s | like non-roll cue cues that sounded |
6875s | weird but it's accurate uh it's like I |
6877s | have like lower health so again trying |
6879s | to make that like overall like |
6881s | it's like we want fair and balanced but |
6883s | chaotic crazy but crazy |
6887s | balance like in a competitive sense like |
6890s | mystery here is like how do you balance |
6892s | those teams from uh I can't like a very |
6894s | competitive standpoint I mean any |
6896s | individual game |
6898s | the result is going to be like even more |
6899s | like even more random right yeah in the |
6901s | sense of it's like yeah you'll have some |
6902s | total you have some games where someone |
6906s | happened to like you know roll into like |
6908s | their best hero |
6909s | and like you know they they play really |
6911s | well on it and maybe like dominate you |
6913s | know and like oh it's like oh all I got |
6915s | were my off-road Champions and so forth |
6917s | yeah that's fine but everyone's having |
6919s | that experience yeah |
6921s | so there will be there's always like |
6923s | some balancing out there and sort of |
6925s | It's actually kind of like sort of an |
6927s | interesting thing about like Mr Heroes |
6928s | it's like when you do lose |
6931s | it's like you can just blame it's like |
6933s | oh we just had bad cops yeah |
6936s | it's like you won't even blame the |
6939s | Matchmaker yeah right it's great you |
6941s | know it's like it's like yeah just like |
6942s | you know I didn't roll a great set of |
6944s | Heroes but if you play enough |
6946s | yeah they're like all that kind of |
6948s | variance right we'll balance out and the |
6950s | people that are better at playing more |
6951s | of the heroes should rise to the top of |
6954s | the leaderboards right that's the theory |
6957s | well and um also dealing with |
6959s | uncertainty like I I actually play Mr |
6961s | Heroes I love the mode it's such a fun |
6963s | mode |
6963s | um and one of the things that's really |
6965s | cool about mystery Heroes is so like |
6967s | normally in OverWatch it's very |
6969s | deterministic right like you're you have |
6971s | control over kind of everything or your |
6973s | team does at least |
6974s | except whether you win I guess so um you |
6977s | know like in games like HearthStone for |
6979s | example kind of like dealing with |
6981s | Randomness and uncertainty is like a |
6983s | very inherent part of the game and so |
6984s | you know that is is like a skill |
6987s | actually you know it's like if you roll |
6990s | um a mercy and your team already has |
6993s | three supports or something like |
6995s | like kind of like how do you play that |
6998s | do you like is it do you want to stay on |
7000s | Mercy yeah |
7003s | just jump it yeah so there's like a very |
7007s | interesting kind of like strategic layer |
7010s | that you know it is a little bit crazy |
7011s | but at the same time there's a lot of |
7014s | thought that you can kind of put in |
7015s | that's you know it's very dependent on |
7016s | like how good am I with this hero you |
7018s | know what's my my team's team comp |
7020s | what's the enemy team comp and kind of |
7022s | doing that whole like calculus to try to |
7024s | figure out like is this actually a good |
7026s | hero for me to like kind of try to hold |
7028s | on to or not |
7030s | or should I just yeah so we think at the |
7032s | end of that competitive season the |
7033s | people are at the top of the leaderboard |
7035s | they'll just don't deserve to be there |
7036s | yeah |
7037s | Swiss Army knives of OverWatch yes |
7040s | absolutely |
7042s | um |
7043s | so yeah we'll see we'll see we'll see |
7044s | how that uh we'll see how that goes like |
7046s | I said there's a lot of excitement about |
7048s | it I mean there's lots of fans of Mr |
7049s | Heroes on the team |
7052s | um it's I think it's it's still like our |
7054s | most favorite it's like our most players |
7055s | it's a really it's almost played our |
7057s | arcade |
7058s | yes |
7061s | yes embrace it this this might be a |
7064s | really stupid question uh the answer |
7067s | might be super obvious and I just don't |
7068s | know but |
7073s | uh like obviously that's locked on my |
7076s | account like that's my MMR for mystery |
7078s | Heroes does that bleed into other |
7081s | competitive uh like normal competitive |
7083s | play or like quick play so if I say like |
7087s | a really unlucky very often in mystery |
7089s | Heroes by Rolling off Heroes constantly |
7092s | and then obviously I'm losing all of |
7094s | those games but I technically do good in |
7096s | normal competitive play will the other |
7099s | MMR affect that MMR and my account |
7102s | overall that's a really good question so |
7104s | for competitive modes each each rating |
7108s | because it's tied to a rank |
7111s | um it's you know player visible those |
7112s | are very distinct so like you can do |
7115s | whatever you want in competitive mystery |
7116s | Heroes and it will not have any impact |
7118s | on the other game modes for unranked we |
7121s | do |
7122s | um have a little bit more kind of like |
7124s | of a of a you know shared understanding |
7127s | of your ratings |
7129s | um because you know in those modes we're |
7130s | really just trying to optimize for you |
7132s | know like Fair matches with a 50 percent |
7134s | win rate but for ranks you know we |
7138s | you know we're showing you those ranks |
7139s | you know we're not changing those out |
7141s | from underneath you so like that rank is |
7143s | you know just for competitive mystery |
7145s | Heroes yeah so when you lose a |
7148s | competitive Mr Heroes you're not gonna |
7149s | you're not gonna need to get upset and |
7151s | say like oh my am I going to lose you am |
7153s | I going to lose GM |
7156s | oh don't worry about that I don't have |
7157s | to worry about it yeah I'm pretty |
7159s | excited to try out competitive mystery |
7161s | Heroes here for the chaos so could we um |
7165s | could we have OverWatch League Play |
7166s | competitive Mr Heroes do you think just |
7168s | first I mean yes |
7174s | that idea that's I think that's that's |
7177s | gonna go down really well |
7179s | I'm excited |
7180s | awesome |
7181s | this has been this has been really |
7183s | helpful and you know we really |
7185s | appreciate the insight and I know uh |
7189s | throughout this we've kind of talked |
7190s | about how you know we should have |
7192s | communicated this or like we should have |
7194s | done this better and I just really |
7196s | appreciate you guys taking the time to |
7198s | be here and like really outline this for |
7200s | the community |
7202s | um and you know I'm really happy with |
7205s | everything that you guys said you know |
7207s | um I appreciate the insight and it's |
7209s | been epic talking with you guys these |
7212s | are really important topics to our |
7213s | community so yeah we should do this |
7215s | again yeah yeah absolutely thank you for |
7218s | joining thank you guys so much yeah and |
7220s | and sharing your thoughts and you know |
7223s | um |
7224s | we talked about all the factors that go |
7227s | into uh changing Matchmaker and like |
7230s | this is true for Matchmaker it's also |
7232s | true for our game but Community |
7233s | sentiment and Community feedback means a |
7236s | lot to us we see we hear you and we are |
7240s | implementing that into our plans and it |
7242s | helps us make decisions so you know |
7245s | please keep sharing your feedback |
7248s | um we're really happy to have these |
7249s | conversations with players and you know |
7252s | thank you guys so much for watching and |
7254s | we'll see you in the mid-cycle patch |
7257s | bye everyone thanks bye thanks |
7271s | [Music] |
7288s | [Music] |
7298s | [Music] |
7300s | foreign |
7301s | [Music] |
7309s | [Music] |
7351s | thank you |
7352s | [Music] |
7371s | foreign |
7377s | [Music] |