Yesterday morning Diablo III Game Director Josh Mosqueira gave an hour-long presentation on Diablo III and Reaper of Souls at the annual Game Developers’ Conference. You can watch the whole thing on video, but if you don’t have an hour for that, here’s our transcript, with numerous screen captures to show the key visuals.
The presentation was excellent, with Josh expanding and improving upon the Blizzcon 2014 panel “Evolving Reaper of Souls.” We wrote that one up from Blizzcon when it was good. Here it’s better, much deeper and more detailed, and Josh seems very rehearsed and prepared. The talk gives inside info about how much of a struggle it was to change launch features of Diablo 3, how the Console project helped a lot with that, and how often they went wrong by trying to be too much like Diablo 2.
The following is our rough transcript, plus a few sections of added info/rebuttal, as Josh was very honest and self-critical about the game, but a few places (in my “was there at the time” opinion) this presentation avoids trouble. Predictably, as those parts are about larger financial/institutional decisions, rather than about the game content, and they were made before Josh was even part of the team.
Diablo III’s Road to Redemption with Reaper of Souls – GDC 2015Josh Mosqueira. (Starts off with an intro and jokes about his accent. Content begins with a disclaimer:)
Disclaimer: Two points.
#1: Tempting to think since I’m up here on the stage and I’m the Game Director, I’m responsible for the changes. Nothing could be further from the truth. It’s a huge team project, all our team, other game teams offering feedback, management, and we listen to player feedback and adapt and adjust to the community’s desires. I am standing on the shoulders of giants.
#2: Contrary to popular belief, we do have dates and milestones at Blizzard. We can’t actually take forever. Reaper of Souls was supposed to launch in Nov 2013. We had to go all the way to Mike Morhaime to get more time to make it right. Everyone understood what was at stake, since we hadn’t quite hit our goals with D3v. We got a couple of more months to work on it and that made all the difference. Adventure Mode really came together in that time. (Eventually released in March 2014.)
Where’ is Diablo 3 today? We’re about to launch Patch 2.2.0, and we’re excited for that. So where was the game when I took over as Game Director 2 years ago. To really understand what a Diablo game was all about. Jay Wilson gave a great GDC talk about the seven pillars that defined Diablo 3.
Conveniently, we revisited Jay’s “Seven Pillars” in an article in early 2013 just after he left the Diablo 3 team, which created a lot of debate. What were those seven pillars that Jay and team built the game upon? Quoting Jay from a May 2012 interview, “Those seven things were: approachable, powerful heroes, highly customizable, great item game, endlessly replayable, strong setting, and cooperative multiplayer.”
How well did the community feel those 7 were handled in the game? Here’s the vote from our article about the Seven Pillars, revisited when Jay left the Diablo III team in January 2013.
Click through to see the grim results of that vote, and continue with the remainder of the lengthy presentation write up.
Well, that’s pretty dismal? One strong favorite, one 50/50, and 5 failed, with “None of them” actually ranking above three of the seven pillars. Happily, I think a revote today would draw much higher scores, with the game so much improved over the past two years. We’ll try that later, but let’s get back to Josh’s presentation.
Pillar 1: Focus on the Fantasy: In Diablo that means, “Epic Heroes… Against the forces of Darkness.” RoS did a good job pitting the Crusader against the only other figure in the universe who can rival Diablo in coolness.. Malthael the Angel of Death.
Pillar 2: Endgame for Everyone. Whether you play an hour, 10 hours, 100 hours, 1000 hours, we wanted to be sure there was something for you to do. This is what Adventure Mode came out of.
Pillar 3: Make Online Matter. We feel that when you’re playing with friends the magic of the game bubbles up to the service. We added clans and communities to encourage players to play together.
We constantly refer back to these three Pillars when we look at changes in patches and new content.
I’ll have to ask one of the devs about their constant “multiplayer is the way to play Diablo 3!” assertions some day. I tend to disagree. MP is “a” way to play, but lots of people never party and enjoy the game just the same or more. I party sometimes, but always feel that playing solo is a more pure and distinct experience. There’s a reason most of us judge class strength by their solo Greater Rift leaderboard, rather than how well they do in 2-4 parties.
The question comes up ever now and then, but probably not often enough. While MP tools and party incentives are great, do you guys think the devs focus too much on it? Might they be missing our or sacrificing some single player value with their focus always on a multiplayer mode that so many fans don’t even play?
I thought we’d run a vote/survey on preferred gaming mode; solo or party, but I can’t find it in the archives to we’ll have to post one in the days to come.
Part I: How did we get here?
7:10 — Why did we have to make all these changes? To explain, I want to go back in time, to the year leading up to Diablo III shipping. I joined the team in May 2011. I was there for the last year, and it was an awesome time to be there. There were people who had been working on the game over 10 years, all the way back to Blizzard North. I was really excited even though I’d only been there for a short time.
I could also see, as an outsider coming in, that the specter of D2 loomed large over the team. Trying to live up to it weighed heavily on the team.
But launch day is coming closer, we wrapped up the latest iteration of the skill system (#85, which was crazy.). Finally, after 10 years of waiting, there’s another Diablo 3 coming up. Even on Google we were the #5 ranked search of the year.
Launch Day came, and it was a global celebration for the team. Launch events around the world. People couldn’t wait to play the game. The team had been waiting for along time too. We had been growing these crazy beards, and we shaved them off on stage at our Irvine event.
Launch was great, incredible sales. The sales team had projected 6.66m units first year. We sold that many in the first couple of days. It was mind-bogglng. Early reviews were very good scores, but the servers couldn’t handle it.
We could see it from the opinion of the fans, “Come on you guys knew it was ten years of build up, how could you not have the servers working?” But even our most outlandish estimates were below the actual crush. It was a horrible time for us, and the wounds are still really painful. It hurts me still to say “Error 37.”
11:00 — We learned from it though, and Reaper of Souls was the smoothest launch in Blizzard’s history. But the damage had been done back in 2012. The game people had been waiting for, we got terrible User Scores and angry player sentiment. “Let me tell you, that felt shitty.”
The launch day/week server crush definitely a factor, but as someone who was documenting it at the time, there were several game issues that caused a ton of anger and controversy long before the launch. The biggest were the Real Money Auction House and the Online-Only DRM.
Enough people were railing on those we could do news posts collecting multiple such rants all the way back in 2011, and my impression of the Diablo III launch was that a lot of the haters were lying in wait, ready to pounce, and the Error 37 server overload gave them the opening. After all, it was the ultimate vindication of hate for Diablo III being an online-only game, so when Bnet was down, no one could play.
Josh’s presentation here was very honest in most ways, but it totally slid past this one without even a hint that there might have been a lot of festering resentment and anger at Diablo 3 long before the launch. After all, plenty of games have rocky online experiences for the first day or days. It’s almost to be expected, especially when a game breaks all time sales records. Most of the time it’s forgotten, or looked back on almost fondly.
Remember Diablo 3 at launch?
12:00 — The road to hell is paved with good intentions. Every decision leading up to Diablo III felt like the right one, but in retrospect we made some mistakes. One of the first ones was we struck dogmatically to the structure of D2, with three or four difficulty levels that repeated the same content while growing gradually more difficult. Players wound up hitting a brick wall of difficulty.
And on the story side, no matter how awesome it is, the hundredth time you see Cain die, and realize he’s being killed by Butterflies, you’re going “what the hell?”
The game was too hard initially, once you got into Inferno. Instead of being a massive hero destroying the world, you were a Barbarian smashing pots. The bravest heroes were fighting terracotta. And when players weren’t smashing pots, they were finding the best quest steps to find the best loot. Players were spending more time looking at loading screens than playing the game, as they kept repeating the same area or two.
And you’d bring back a whole inventory of rare or legendary items, but they were almost never upgrades. A true story, on my live Barbarian, it took me 104 hours before I found my first legendary. And you know what it was? A quiver.
And we sent you to Ebay to get the gear. The Auction House was another thing we had great expectations and hopes for, but it short circuited the item finding/upgrade process.
Launch was the Lowest Point for the team.
15:30 — Launch was our lowest moment. We’d let ourselves down and our players down. But instead of fracturing and bickering and losing faith, we rallied around and told each other we can do this, and that’s exactly what we did.
Where did we go wrong?
1: Items. Long item reward tail aspirations…. resulted in stingy loot drops. We wanted to emulate Diablo 2 and thought 10 years into Diablo III, players might still be playing and finding items they’d never seen before. We didn’t realize clearly that the Diablo 2 people remember was created after years of patches and an expansion. The drops were too stingy.
We also thought that players wanted a hard, challenging, brutal game. In truth people want fun and challenge, but in terms of efficiency and slaying monsters and getting loot. We tried to make the game hard and had a stingy item curve, and that was not what players wanted.
We also wanted sometimes Blue items to be better than Rares, and Rares better than Legendary. That ended up being confusing and players didn’t like it.
And smart loot wasn’t something we realized, and we had too much randomness on items. We thought randomness was the key. Today we realized that replayability is where it’s at. Randomness is a tool to create Reliability. We thought near misses would be fun, but players found them annoying and felt like we were mean, giving items that weren’t good, and weren’t even for your class.
Loot fountains are awesome! Except when the fountain is just junk.
Misjudged Player Psychology
Why we play: Fantasy vs. Efficiency. Fantasy is you’re a bad ass hero slaying all the demons of hell and getting all the great loot. Aspirational. How we play is very different. We go to Google and look up builds and strategy and follow those. Players short cut things right from the start. Back when Diablo 2 launched there weren’t such tools or streamers or YouTube tutorials showing all the secrets. Players had to explore more, while today people go right for efficiency.
Wrong kind of Beta:
Our beta consisted of the first 45m of the game and only up to level 12. No way to do in depth testing from beta testers or get any real actionable feedback. The reason we did this was we felt the spoilers were more important to protect than making sure our item-based game was properly balanced. On RoS we switched entirely, with the entire content except for the Malthael fight… and sure enough, when RoS lauched players found a bunch of bugs with the last fight and we had to fix it.
Part Two: Evolving into Reaper of Souls
So what changed?
25:00 — We patched furiously, listening to player feedback. To make the changes that players really wanted and that we wanted as well. Our team never gave up and we knew we could solve the problem. One thing that helped a lot was perspective we got from the Console version. We had just 3 people working on the console initially, and I remember sitting down with the Diablo III developers and they were telling is we weren’t sure console would work. “It was a test, a gamble. If it’s got to work, it’s got to be awesome. So change whatever you need to change.”
That was very exciting, but also terrifying. Getting the keys to the kingdom, but you have to not wreck it.
Pick up and Slay:
Hand built for the console:
One couch to rule them all.
The last was the most important, and we really worked on the multiplayer and co-op features and functions.
Perspective was key. The console allowed us to challenge the assumptions the core team had been operating under. Controls, loot, and looks. Moving the camera closer on console was a huge war and struggle, but we liked it.
A controller is not a mouse and keyboard. We’re going to put this beloved PC classic and put it on a dirty console controller. Yet the first time I tried playing with my Barbarian using the controller, and I felt much more connected. I feel like I’m walking my Barbarian, not controlling it. I felt like I was the Barbarian, rather than Josh clicking away on a computer.
What we learned is that Diablo is not about how you control it. What you’re doing is slaying demosn and getting loot.
The other perspective console gave us was Adventure Mode. This came from the console version with the difficulty select screen. At his point in the PC version we had the Monster Power system. We liked the system, but it wasn’t console friendly. So we put it right in front at the game start screen on the console. This started to change our thinking about difficulty.
In Diablo III at launch, monster difficulty was hard coded. The same skeleton in Act One normal had X hit points, and the same for Inferno. What we experimented with the idea of monsters leveling up with your character. But the fear was that the monster levels and the player levels going in parallel was a flat experience. No peaks and valleys. But we found that items were the vector that gave us that. It was fun to find a new item and become overpowered for a while, until the monster caught up.
We decided to allow players to unlock all the waypoints and let players go anywhere and slay anything.
The biggest change on the console was to itemization, in a system that eventually became Loot 2.0.
Less is More — We knew console players couldn’t spend the entire game in the inventory screen. So we cut down the clutter and the total amount of drops.
Items Define Builds — Encourage diversity. Creating interesting item choices.
Rarity = Power. We put in a clear item quality progression, with legendary items always better than rares.
This is a screenshot from early in the console, from 2011. I’d only been at Blizzard for a couple of months, and these two items dropped. I was playing a Wizard now, but I couldn’t figure which one was better. I took this over to the systems designers next door, and that turned into a 45m discussion about which item was better. That set off alarm bells in my head. If the designers couldn’t decide, how were players going to do so? And worse for the console players, who we didn’t want to spend their game time staring at the screen trying to work this math problem.
I’ve seen this problem in other games, they put green arrows up and red arrows down. We put that system in and it’s very useful for players to learn. The first backlash, “that’s not in D2. Why is it in Diablo III?” The next backlash, “we’ll give you the green arrows, but how to figure which is better? How do you simplify which affixes are best? Won’t you make decisions for the player.”
That could be a problem, but in a game that’s about killing, but any affix that lets you kill faster is the right choice. But it was a long internal conversation and debate to thresh it out. And while working on RoS, a lot of the fixes were trying to adjust to compensate for prlbmes with Diablo III’s launch…. were only addressing the symptoms. Not the larger system.
30:00 — The cause was our item philosophy at launch. Players didn’t really understand the concept of rarity vs. power. We were strapped for time so we made legendary items stat sticks instead of build changing. And we treated all affixes equally.
The fix: Drop fewer, drop better, make legendaries legendary, cheat, in short… Loot 2.0.
Smart drops. Affix groups. Flavor vs. function. Highlight the legendary affix. Make clear the ranges of quality.
Not out of the Dungeon Yet…
The Auction House.
Our intention and motivation was to provide a safe environment to trade. D2 was plagued by account thefts, item thefts, etc. Even though that was players doing their own thing, we wanted to bring it under the Blizzard umbrella and provide for it.
Our thinking was that only about 10% of players would use the4 Auction House, but the stingy item loot forced everyone to use it. It became a lose-lose issue, where players did nothing but the AH, or others hated it but felt they had to use it to compare. It short-circuited the item progression where even if a friend gave you a great item, or you bought one, then you’d never find better.
We had a big meeting about back end changes to Battle.net for Hearthstone and Heroes of the Storm, and all the talk was about technical issues and financial models, and I was day dreaming thinking about making an awesome Barbarian. But when Mike Morhaime asked me what I thought, considered saying just make minor fixes, but I bit the bullet and said we needed to shut down the RMAH and the GAH. There was surprise in the room, but after just twenty minutes or so of conversation, everyone agreed, both the game guys and the financial guys.
Three days later we made the announcement, and six months later the AH was gone. It really signaled to the fans and even the others at Blizzard that we were going to make Diablo 3 better, or die trying.
Blizzard has never shared any details on how much business the RMAH was doing and what their income from it was, but if it only took twenty minutes for the financial guys to agree, I have to think the cash inflow from Blizzard taking their share of RMAH transactions must have dwindled considerably by then.
Of course there was a trade off, and Josh and others must have argued that the RMAH was unpopular and hurting replayability and they thought the expansion would sell better if the AH went away, players saw how much more fun Loot 2.0’s drop rate was, and then RoS launched with more of the same.
Then and Now45:00 —
Diablo III: Efficiency beat fantasy vs. Reaper of Souls: Fantasy First: Game-breaking Legendaries…. lots of them.
Diablo III: Shop, not play vs. RoS: Loot 2.0 + shut down the AH
Diablo III: Flip farming vs. RoS: Variety is efficiency. Adventure Mode.
Part IV: Making Games is Hard
To Hell and Back
The Importance of Pillars: I presented two sets of pillars in this talk, for RoS and the Diablo III console. It’s really important that you can express to your team and to yourself what the DNA of your game is. They need to help you focus and make decisions. When you’re fighting fires an things get crazy, you need to go to your pillars of design. Not 20… no more than five. I like three.
Know Your Fears
All of us as creatie people and game designers make assumptions about hwo thigns will work. Those are often based on our fears. What don’t we know? The two fears that came back in RoS. “That’s not Diablo.” We were so afraid not to live up to the legacy of Diablo that we made poor decisions. Finally by RoS we felt what the inherent core was.
The other fear as an item based game was we were afraid to be too generous. We know people come back to games for more items. Drip-based. The fear was if we give them everything, they’ll stop playing. What we realize now is it’s much better for players to get loot and leave happy, since they’ll come back for a patch or an expantion. If you’re stingy they’ll leave frustrated and not get an upgrade for hours. You need to be comfortable being generaous.
Randomness =/= replayability.
Legendary pity timer. Make sure players don’t get too frustrated. Randomness can push people away.
Remember the Fantasy. This was big for me, since we constantly need to ask ourselves why does this matter for the players. Why are they playing my game instead of something else? You are trying to immerse the player. Fantasy first and foremost.
Your Team Defines Success. When Diablo III shipped we should have been on Cloud 9. Reviews were awesome, the sales were astronomical. But I remember on launch daywalking through the halls, it felt like a funeral. Success for us is about players being happy playing the game. We’re very proud that player satisfaction is up and players are excited about seasons, and we’re merging the console and PC gaming families.
51:30 — Before I close I want to give huge thanks to the Diablo team.
Q: Thanks for the talk. You mentioned the tension between the old guard and new guard. Does the tension comes from dealing with a bigger audience? The audience has access to the Internet? Is it because we’re getting old? Or the more streamlined console design?
Josh: Good question. The answer is a bit of all of it. As a team we realized we were slow to adjust. Diablo III asn’t the first loot game, but the biggest mass market. In the 10 years between D2 and Diablo III players adjusted their demands, and were used to games like WoW.
Q: How did a new guy on the team become game director so quickly?
Josh: I don’t know. I think at any moment they’ll ask me to leave. You should probably ask my team; it’s a hard question to answer. When the position was open, I was not going to apply for it. I just wanted to be a Lead Designer again. With my small problems to deal with. But I looked out at our team and I loved them and I wanted to take care of them and go forward with them. I applied and I was shocked when they chose me.
Q: More bigger beta tests?
Josh: We use the PTR a lot now. We get great feedback from patches coming up. Community is very important at Blizzard, and we like as much feedback as possible.
Q: When RoS launched and had the dynamic difficulty, I thought it was a good way to fix the rigid difficulty. But I notice players go through all the way on Normal but refuse to change. did you have trouble retraining players to notice the dynamic difficulty?
Josh: WE did. And I don’t think we’ve figured it out yet. We’d like more natural ways to scale it up and down for players.
Q: I started with RoS on the console. Will you unify the experience between console and PC?
Josh: We’re trying to release patches almost simultaneously now. But we have to treat them as separate games, but we’ve changed so many skills that it’s impossible to unify the games. But we keep them as close as possible.
Q: Why not PVP?
Josh: We’ve yet to figure how to make the ideal PvP, so we put some competitive elements into Greater Rifts, but we’re still in the process of evolving the game.
Q: Valve: More hats, everything is tradeable. Diablo III went the other way with nothing tradeable. Will change?
Josh: No… we feel that the fun is finding your own gear. We can imagine some social trading and exchange, but we want players playing and finding items, not ebay.
That’s the transcript/summary. Here’s the video.
I thought it was an interesting show with very useful info on all the game design decisions. The only real omission/error was acting like everything was roses and honey right up until Error 37, and if only they’d had enough servers then the launch would have been fine.
Which wasn’t true, for reasons stated above, but I’m curious if that was even technically possible. If they’d known in advance exactly how busy they were going to be, was it just a matter of adding more server capacity? So why didn’t they err way, way, way on the side of caution, just in case? Or were there technical problems with Bnet that went beyond just having enough machine capacity to log everyone in and let them play?
Anything that you guys noticed especially, or wished that Josh had spent more time discussing? Josh Mosqueira Diablo 3 Presentation: GDC 2015.