"Even though we know how to do good design, we still make mistakes a lot," said League of Legends
design director Tom Cadwell, and in his GDC Online talk, he explained common pitfalls and how to avoid them.
"You can set up processes that facilitate good decisions," he said. "Once you know what the right behaviors are, it's very possible to coach your peers or subordinates. Some of these pitfalls are human flaws, and it is possible that some people don't manage those flaws well enough even though they're great designers."
Going With the Flow
One League of Legends
champion that created a huge problem was Omen, a character that ended up being cut 50 percent through the process -- "a lot later than it should."
"The excitement level was low and that's not enough," said Cadwell. "The really telling sign is that we put tons of effort into this character, but it never seemed to get a lot better."
There was "no clear decision point"; people tried to keep improving a design that never had great buy-in. "What we should have been asking those decision makers is, 'Would you bet this character is a 10?' They would have said 'Absolutely not,'" said Cadwell.
"It's really hard with a creative thing to say 'that's bad,' even with seasoned professionals, it's personal," he said. You have to "make it hard for decision makers to bless so-so work. It's really important that you do speak up and have a real discussion about it."
The Grand Unveil
"Getting lots of early feedback and iteration is a consistent path to quality. Anything in your organization that greatly impedes early feedback and iteration is also going to impede quality," Cadwell said.
When asking someone if you can see their work, often "the answer you get is 'Y'know, it's not perfect, I want to fix a few problems. Can we talk about it next Monday?' The right answer is 'I really want to see you work, I know it's early, but I want to give you some feedback,'" Cadwell said.
"Writers often refer to the 'crappy first draft' and this is really clever, because you're saying 'I expect your work to be bad when it starts off'. The goal is to make the early feedback positive and pleasant as possible."
"Ultimately with all design you're going to have to iterate a bunch to solve hard problems," said Cadwell.
"Designers are really worried that their great idea is going to be misunderstood or unfairly judged if it's seen too early," he warned.
Three key tactics:
- Don't let people hide progress -- scheduled deliverables of the crappy first attempt are key.
- Make sure that early process is positive -- use peer review. "Because there are no repercussions and it's positive, it often fixes problems."
- Have a mistake-forgiving culture. "You shouldn't openly punish mistakes that will obviously happen."
Too Awesome to Cut
"We probably shouldn't do it, but man, it's too awesome," is a dangerous statement, said Cadwell.
"Now you have a problem because you're bending over backwards to do this thing that doesn't fit."
Teams get jazzed by coming up with cool ideas -- "Now you're all excited... And you don't want to be the guy who's being a buzzkill" by pointing out it just won't work.
"Pretty soon, everybody's not objective," Cadwell said. This happens more often when "A series of small ideas... get blown out of proportion."
To combat it, make sure "decisions are reviewed by peers or supervisors who are not interested parties, who weren't at the kick-off."
Also, said Cadwell, "Structure process to assess cost to risk -- it takes six times the engineering resources. Is it worth it?"
The champion Wukong the Monkey King launched for China beta, he said, with three goals:
- "A character a global audience would enjoy"
- "A character true to the local Chinese Monkey King legend"
- "Make sure that because this character, which was going to be played by a lot of people new to League of Legends
... that the complexity was manageable... while making sure it's a deep character the rest of our community would like"
"Initial design had a persistent clone... Unfortunately it's really, really hard to play. Experts could handle it," said Cadwell. Newcomers could not. "We set out to create a character that's easy to grasp... we've actually created the most technical character in the game, ever."
You have to ask yourself, "Is the thing I'm designing consistent with the goals I originally had, or have I deviated in some way?"
Bear in mind that "If you state a bunch of goals, it's hard to honor them in complex work," Cadwell said. "And we sometimes under-assess secondary impacts."
"We need a lot of work done, so we typically need a long meeting," said Cadwell. However, psychological research shows that "creativity and attention drop after 15 minutes and they plummet after an hour, so long meetings to come up with creative ideas don't do that."
"Another fatigue is when you give the game a lot of constraints," he said. "Can't be blue, has to be human, has to be like this other thing, has to be 10 feet tall, has to be 400 polys," he joked. "Sometimes constraints really help, but sometimes if you over-constrain you have nowhere to go."
Make sure the constraints are actually mandatory, as a solution.
Also, time box meetings. For Warcraft III
expansion Frozen Throne
, which Cadwell worked on, Blizzard's Rob Pardo would "insist that level design meetings not go beyond two hours a day," he said, to avoid creative fatigue.
Another tactic: "Rotating people is really potent... If you swap out one of them, not only do they contribute new ideas but they will also force the others to do so because it changes the chemistry of the group," said Cadwell.
Also he believes that small groups are "more creatively productive" than large groups, so break people up.