In a CastleVille postmortem talk at GDC Online, Zynga Dallas' director of design Bruce Shelley and senior designer William Lemons spoke about how focusing their design process on a fun core loop was key in making their first Zynga title a success -- but over-iterating and a weak central vision made development harder than it should have been.

What went right

Strong design focus during pre-production

"When you're competing in a space with so many productions, you can't do the same game again because nobody is going to buy it again," Shelley said. "You have to be different, you have to innovate at the gameplay level. So first, we didn't see any other games on Facebook set in the Middle Ages, and then we had to do things to make the game itself a new experience." Lemons felt that the studio's inexperience with social games actually proved to be an asset, rather than a liability. "I think one of the things that was great about our design process is that we were a brand-new team, still in the process of presenting the game idea, and just focused on making the best game we could, not the business aspects," Lemons said. "So we looked at the social games out there, like FrontierVille, and thought, 'This is something we like, but how do we make it better?' I think the fact that we were so free to design in those early stages meant that we didn't worry so much about whether systems in other games had failed in the past." CastleVille's crafting system, for example, was something that Zynga Dallas fought to include despite the fact that Zynga execs weren't enthusiastic about it. "Some of the company leadership was uncomfortable with it, but one of the lead designers insisted that the game wasn't worth doing without the crafting system," said Shelley. "After we shipped, one of the board of directors said that he was 'humbled' by how he had missed it completely and we were so right."

Powerful design tools

Also key to CastleVille's success was Zynga Dallas's in-house toolset, which enabled designers to exercise a high degree of control over the missions they built. "Once we started interacting with other Zynga teams, we found that we had a great deal of control over implementation than other teams had," Lemons said. "We had almost as much control over the quests as some of the coders did, and when we brought other Zynga veterans in, they were really impressed by our tools and how much versatility they had in quest systems. One of our best moments of user delight was when we had Santa Claus get shot down by a villain, and he crash-landed in the scene and his butt is sticking out, and we had total control of that on the design side."

MMORPG-style player progression works in social games

Using an MMORPG-style leveling system let designers pace the player's learning experience and exposure to new content. "We used player progression -- a leveling system -- to unlock new things, which was really important for me to teach players new things, like economic planning and the crafting system," Lemons said. "With the leveling system, we were able to teach them things in chunks, which helped us to keep from overwhelming brand-new players with the weekly content that was locked by levels."

Collaborative design review process

Both Lemons and Shelley stressed that their team's collaborative design review process allowed them to quickly and effectively test and revise designs. The design process started with a brainstorming session, after which they'd assign specific design elements to individual developers, who would build the design spec and pass them back to the group for a review over email. The email review then informed the next group discussion, during which the team would reach a consensus on an accepted design spec. That accepted design spec would then be play-tested, which led to further rounds of iteration. Shelley did note that certain game elements never quite made it through the whole design process -- chief of which was the castle itself. "We struggled with the castle," Shelley said, "How important is the castle? Is it the focus of the game, or just in the name? Originally, we thought that you'd be driving to build a big, beautiful castle, but we never reached consensus on that."

What went wrong

Execs gummed up the creative process

For social games, ensuring that first-time players have a smooth and enjoyable experience for the first ten minutes or so is critical to attracting new players and keeping them around. Zynga calls this the "first-time user experience", and Zynga Dallas had a bit of difficulty over-iterating CastleVille's. "We had a lot of user testing, we had several different demo versions that we showed people, and the FTUE was generally how we'd show the game to the company," Lemons said. "We had a FTUE planned out that we were really happy with, one of our first few iterations, but after we got our user testing results back [Zynga execs] wanted us to try several different things, and a dozen versions later, once we finalized it, we sorted of ended up back where we started with the first iteration. It was frustrating for me in particular because I was the one implementing this stuff." Shelley elaborated on how iterating from the top caused problems with their development process: "We had leadership above us who would come in and say 'I think this needs to be changed,' and they hadn't been through our collaborative process, in our meetings, and we found it discouraging when someone just came in and told us what to do."

Design tools had high learning curve

While the designers' tools were powerful, Lemons pointed out that they weren't really easy to pick up and use. "The implementation of our design tools was fairly difficult. Fortunately, there were a number of us with a CS or programming background, and everyone on the team had a different skillset, so people with more of a narrative background had a different time with that stuff. Our tools weren't created in scope, they were created by a programmer in his own time, if he hadn't made those tools we wouldn't have had anything."

MMO progression system made it hard to expose new players to new content

While the character level progression system helped designers introduce new game elements gradually, it also made it harder for them to expose their new content to brand-new players. "One of the things we weren't prepared for was the concept of the 'wide funnel,' which is that the goal is to release new content to as many players as possible," Lemons said. "We were really attached to the MMO progression, and it helped us in some ways, but we had a lot of things that were locked behind higher levels, which meant that there were a lot of things we couldn't let the players do. Half of our crops were locked behind level 15. When we were implementing our quests, we got frustrated because so much of it was level-locked, and they wanted our new releases to go to as many people as possible. Eventually we had to go back and bring some of those things to a lower level, make some of those quests a little bit easier, which was frustrating."

Royal pains

Despite CastleVille's misses, Lemons and Shelley were enthusiastic about how they brought a bit of their traditional game development skills and craftsmanship to the world of social games. "Fun is interesting decisions that have meaning and goals, that was what I think was missing from social games when we started playing them," Shelley said. "Early Facebook games used to put me to sleep."