"Some of your parents might be aware of our previous work," jokes Remedy’s Matias Myllyrinne -- the company’s last game, Max Payne 2, shipped in 2003. According to Myllyrinne, it's been such a long time in part because "it’s easier to iterate on a product than to bring something totally new." The company has been working on Alan Wake ever since, and the project's scope and gameplay style have evolved over its many years of production. Myllyrinne spent most of his talk at DICE last week illuminating why and how the company had spent so long in development -- which is now finally finished. "Our big mistake was to try to mix a thriller into a sandbox design," he said. When the gameplay was reined in to a more linear experience, the company needed to make the painful decision of "giving up six months of work, and throwing away a few million." "We were going to do a tight preproduction stage [on Alan Wake], then do a production stage which was about twice as long, then take time in postproduction, polish, iterate, and get a great game out there. Then reality happened," he said. Toward the end of the prolific prototyping period, the team tried to smash all the features together -- and from there came certain innovations brought on by the necessity of constraint. Preproduction was much longer than they had anticipated, but it took about as long to build the game as they thought. "You’ve got to be prepared to put your money where your mouth is," Myllyrinne said, noting that Remedy financially backed the game itself. "Luckily with the success of Max Payne we’ve been able to do this." Still, partnering with the strong and powerful ensures survival, he said: "Microsoft is really helping us to punch above our weight." During preproduction they experimented a lot, taking concept photos of actors, working with concepts of light and dark. They also wanted the enemies to have an effect of ink dissipating into water, so they actually dumped illustrations into water buckets to see what would happen. The company took over 60,000 digital photographs to put into textures in the world, taking lots of inspiration from real world locations like Astoria and Crater Lake in Oregon. "When you’re building a thriller, you need a base level of reality," he said. From there, layering scary things on top creates a distinction between sheer horror and subtle thriller, where "the real parts of the game feel real." "For a thriller, we wanted to apply a lot of the techniques we used in Max Payne," said Myllyrinne. So they included the concept of a "near miss," in which the game slows down when players do an evasive move, like in the Max Payne series. Still, "it took ages for us to make the mechanics fun," he admitted. "As often happens with our prototypes, we overshoot it. We go way too far, and then we have to dial it back two or three times and we’ve got something." "I wish I could say we design a lot of things on paper, but the way it works in the real world is until we get it running on screen and can play around with it, that’s when we can really make improvements," Myllyrinne said. The company does have design documents, but not a lot of other on-paper design. Remedy refines its games after user research and data analysis -– for example, Alan Wake is shipping with version 20 of the control iteration. Not everything happens quickly. For instance, in 2007 they still had the HUD wrong: "It’s almost embarrassing for something so basic," he says. How does a company pull through such a long development period? "You have to have a lot of faith in yourself and your team," he said. "You need to celebrate the little victories you get along the way. It’s a long distance run, it doesn’t matter so much how you finish one lap, it’s how you finish the race."