"Our mission is to be the most customer-centric game company in the world," said Tom Cadwell, design director of League of Legends, opening up a postmortem of the game at Game Developers Conference's Social & Online Games Summit in San Francisco. "We go direct-to-customer, we follow a free-to-play business model, and we focus on an adaptive ideal for development. Our overall goal is to build a globally relevant game that will be successful around the world," he continued. He and senior producer Steve Snow shared "key lessons" the team has taken away from launching and running the game. One major concern is beta. Beta periods of games were, in the past, used to iron out bugs, but Snow says "the reality is that this doesn't work anymore. Beta can not be about finding critical or major bugs. They're going to get found -- but you have to do everything to find them before your potential customers do." This is because early adopters matter. "The people who come to your product first matter... Meeting their expectations is key to your product's success." And to do that, "you have to be able to follow through on your commitments." Game as Service Says Cadwell, "When you are operating a game as a service you are never done with feature development. You are always going to be introducing new fundamental features to your user experience to make sure that you are leveraging your customer base to the fullest extent you can. The more your user base is into your game, the more they tend to hang around," and the more important this becomes. The nature of running a live service complicates things greatly, says Snow. "Those bugs are blockers to [the game] experience and on top of that you have to do feature development. "New work is coming in every day," says Snow, and this throws off people who are used to deciding on what they need to improve and then working toward that. "This is what we thought we were going to do today, and based on what happened yesterday, this is what going today," The priority: "the highest quality experience you can have for [the players] that day." Says Cadwell, "Whenever we release a new feature we put in multiple hooks that allow us to turn off some of the feature, or all of the feature, if it's blowing up in our face." 
He also notes that if you handle these changes well, "and you have a fun and engaging game, your user growth will be a lot more than you expected." To combat this, says Snow, "we would triage daily -- we instituted daily mandatory company playtests. Tons of feedback was coming out of it. We playtested the build that would go out that week." The company faced a decision: "Do we allocate senior engineering leadership" to crashing servers or game balance issues? "This was happening on a daily basis," says Snow, so "we ended up separating the teams -- the core feature development team and live teams." This structure emerged across the entire organization -- and constituted not only a physical but "a psychological" separation. "By separating the two teams, the people on the live organization knew that what they thought they were going to do yesterday is not nearly as important as what's coming in today," says Snow, which resulted in "much happier people" than staffers who are asked to reprioritize. Rather than have people assigned to specific tasks, says Snow, "we would put our smartest guys" on any problems. "If we don't figure it out as soon as we can, tomorrow will not be a good day for our customers." Riot calls this their "ICE" team -- standing for "improved customer experience." "Kanban ended up emerging as our primary tool" for ICE, while the game development teams stuck with Scrum with sprints. 
 Says Cadwell, "It's important that different groups have slightly different cultures sometimes... The guys responsible for making the game scale need to fundamentally be more methodical and conservative than the guys who are responsible for making the game fun." It's a leadership challenge to make people understand why procedures are different on different teams, however. DOTA League of Legends is inspired by popular Warcraft III mod DOTA, and says Cadwell, "While you do leverage the customer base, most sequels tend to become a little bit more niche. We wanted to have a path to broad appeal." "No one had commercialized DOTA," which made the ideal appealing, "but on the other hand it's a really hardcore game." The goal was to make LoL More accessible than DOTA, but support a hardcore audience as well. Cadwell outlined the mechanic of "denying" -- killing your own allies to prevent your opponent from racking up the kills. This is popular with hardcore players, but unappealing to the team at Riot because "you're denying your opponent of a satisfying game experience... Denying that moment is not a very good experience, in our view." Unfortunately, however, high-level DOTA play requires skill at this tactic, despite it making the game "more passive". To high-level players -- many of whom will be early adopters -- "removing this, we felt, was going to make those people feel effectively nerfed." It was removed in the alpha or beta period, but "it didn't go great," because players were used to the advantage and familiar with it. "It also provided something just point and compare, this something that LoL doesn't have that DOTA has." 
 The problem is that "if you come into a game it feels really alien, then you are not going to attach, and if it has a long learning curve" you might drop it, because "learning takes energy." "We added more differentiating features," giving the player base positives to look at rather than negatives, says Cadwell. And when removing things, the best tactic was "to explain to the community in a transparent way why we did what we did... A lot of mindshare went into this for several months. But because what we were saying was logical and the benefits we were promising were being delivered," ultimately the community accepted it. The team also put a lot of effort into optimizing the user interface. As Cadwell says, "the controls for Warcraft III are not optimal for DOTA" due to the change in game scale -- armies versus single characters. However, says Cadwell, "we ran into a lot of troubles with UI and it really comes down to the learning curve thing... If people come in and their muscle memory gets juked and they have to learn something they previously mastered... There's less chance they're going to get engaged in the game intellectually and emotionally." 
 A rule? "When you change UI, change is usually bad." Another? "Don't just change UI for small improvements -- it's not worth it." Another major design change over DOTA designed to bring in a bigger audience was changing game mechanics to encourage more teamwork and less adversarial play between team members. DOTA "is considered to be adversarial," says Cadwell, "But one thing that will churn out players fast is a terrible emotional experience." "The mechanics of the game incented negative action." Its "incentive structures are really based on the individual," while its team incentives are weak -- "there are not a lot of shared rewards." In DOTA, "Playing badly causes your team to lose more than playing well causes your team to win." However, says Cadwell, in general, "positive team experiences make people want to stick around and continue to play the game." "Most" of the changes the team made were accepted by the community. Says Cadwell, "Making things more pleasant wasn't a problem... The problem was when we didn't have something they had the expected. Adding is easier than subtracting."