GDC Notes: Scott Jon Siegel, City of Wonder Post Mortem

These are my notes from Scott Jon Siegel’s GDC session, “City of Wonder: Postmortem.” Any mistakes or misinterpretations are my own.

Feels odd to give a post mortem on a social game, especially one that’s still live and active. This is more of a post mortem of the dev process leading up to the launch of the game.

CoW is an entry in the city builder genre of social game, modeled after Playdom’s own Social City. Released Aug 2010, 6 months of development time by Studio 24 in Mountain View and Bangalore. It has monetized very well, exceeding expectations. 10.7M MAU peak. Studio 24 also did Tiki Farm (6.4M MAU). Also did Fish Friends (1.2M MAU), shut down earlier this year.

Like a lot of Playdom projects, started with a pitch. The game as originally conceived was about expanding the city across military/economic/cultural power, guiding technology, and interacting with NPCs. That was a little grand. We were building off the Social City engine. We wanted to incorporate PVP and maximize on monetization through customization and PVP.

Social City helped establish Playdom’s post-RPG game titles, but we wanted to exceed the success of Social City. The problem is that we didn’t know if the CoW direction was good. I considered the pitch to be a crazy idea when I first heard it. I had very little interest in working on it. Social City defined a genre, and many other city games popped up with high-quality competitors. The market was saturated on city games.

Studio 24 learned about market saturation the hard way. Fish Friends launched to a saturated market of aquarium games and did not do well.

The concept itself was also pretty niche; not widely accessible and the original pitch sounded like a pretty hardcore game. We knew from Social City that city builders appeal to a nice, even, wide demographic. If the average gamer is a 43 y/o woman, I like to ask myself: would Nicole Kidman play our game?

So why did we go ahead anyway?

Well, the first reason was that we were combining the PVP of text RPGs and isometric builder games. In RPGs people buy stat boosters for player on player combat, but they don’t appear in game so they only really appeal to PVP players. In builders, people buy stuff to customize, but it’s only for the kinds of people who want to make pretty cities, the rest of the people stick with the free content. So CoW was a chance to do both of those things in one game.

The second reason was why I got involved in the first place: we can casualize the more “hardcore” elements. It meant making PVP almost optional, and that complex features scale with the player’s own interest.

The key mechanic of Social City was its city building loop. For SC, three metrics matter to the player: currency, population, and happiness. Currency increases population, but to expand you need more happiness which means you buy leisure buildings, which lets you increase population, etc. The core is a virtuous cycle.

We took the base mechanics of SC and made one tiny change: instead of gating the unlocking of new buildings by experience points alone, we added a technology tree that you needed to research.

Research as a mechanic was not arbitrary based on the source material. Tech tree lets us create additional long arcs for the player, and a path of progress for the player. Branching paths allow players to create diversity in their city layouts. The problem is that tech trees are very complex, usually. Balancing and tuning our tech tree was AESTHETICALLY driven just as much as it was SYSTEMICALLY driven. We designed the tech tree with different varieties of buildings in mind, with fewer options available at a given time, and and untangling the web of dependencies.

Still, even a simple tech tree visualization could incude panic in a player. We made the tree a power user feature, and we made the tech tree look like a standard purchasing UI. And if that was too much, we created advisor NPCs who would recommend specific research with some flavor text. For players who wanted depth, they could run research directly from the tech tree if they wanted to.

Tech tree gives a sense of progression to the player. It goes left to right, and you see the evolution of your tech. We also added events that trumpet the oncoming of the Bronze Age, etc etc. The thing is, many players didn’t understand that the game was about progressing through time. Most people thought it was permanently set in the stone age, so adding the events was an important way to get people clued into the fact that this game is about time progression. Even our loading screen shows the progression of time.

The most important thing we did was a “back of the box screen” that pops up and tells you what the game is about, complete with a slide show that shows the city evolving through the ages. But that animation wouldn’t work if the buildings themselves didn’t change. We drew inspiration from Children of the Nile (PC game by Tilted Mill) and how it changed the flavor of buildings between the ages. We kept our post-launch buildings adhering to the same flavor guidelines.

About the name: we ran tests to see what names got clicked on more in advertising scenarios. We were called Social Civ internally for a long time, but the clear winner was always City of Wonder. It’s not rhyming, it’s not a pun, it’s not clever… turns out there’s a Rihanna song, Disturbia, that says “city of wonder” in the lyrics. My new theory: name your game after a Rihanna song.

Embassies were way to deepen the relationship between neighbors in social games. We wanted adding friends to not just be a one-time bonus, but something that is a long-term benefit.

PVP expeditions: we have a second major system in our PVP game. There’s cultural, trade, and military “attacks”. A player is aided by the buildings they had. We had to think about ways to convey to the player how different buildings affect PVP but there were a lot of variables so it had to be very communicative. We went through a huge number of iterations (about a dozen major ones) until we landed on our final version. We did a clear visual/audio distinctionbetween win and loss. Prior to the player attacking, the advisors chime in to casualize the expedition experience.

Legends system: we added these originally because we wanted consumable item boost. We found that tying boosts to historical figures helped add personality to our game and tie it history more closely. Some legends affect reasearch, population growth, etc. As we grew the game’s features, we added new legends.

I’ve had the privilege of being in the social game industry for a “long” time, about as long as it’s been an industry. CoW has been the best experience I’ve had in this industry. But no project is without its low points as well as high.

What went right: the decision to be casually focused. We had a very even gender split in the game (53/47 M/F in the first week). I think we democratized our game like Apple democratized the MP3 player with the iPod. Steve Jobs insisted it would be the flattest user experience possible, fewest button presses to find a song. We democratized by minimizing the number of clicks you need to access the core features. More complex features were kept optional.

Iteration went very well. For example, we iterated a lot on the starting configuration of the city, using a lot of A/B testing immediately after launch. We had a system of continuous integration which made iteration possible, instant deploy of new features. We don’t have to wait until the end of the day for QA to test. You should also always have a stable build available for people to play. We did a lot of scrubbing on the builds: every single day we’d sit down and critique the build. As time went on these meetings went from long (3 hours) to short (<1 hour) as the game got better and better. We also ticketed every single issue that came out of the build scrubbing.

The team was great too. We’re called Studio 24 because development occurs 24 hours a day between Bangalore and Mountain View. CoW was blessed with some of Playdom’s best resources.

Things still went wrong, of course.

Our legends system was not the huge win that I hoped it was. Because the characters were so engaging, we were asked by executives to double down on the feature. We wanted to expand legends into a baseball card style system: collect, trade, redeem your legends for exclusive objects. On paper it looked great. Unfortunately the collection feature was implemented last minute. It lacked the iteration we were able to give to the other core features, so collections were poorly balanced and difficult or impossible to complete at launch. And only counted a legend as collected once you’ve USED it rather than OBTAINED it, which just confused players. The feature also didn’t have the same UI polish time. Many players didn’t realize we had a collection system. Our most successful legends were ones that we offered at certain pinch points in the game and popped up in players’ faces. Obfuscating boosts with charm may have done more harm than good in the end!

Another feature that could have used more love was embassies. These were one of the last features we implemented, a side effect of putting off platform integration and virality into late in development. The feature was solid but there was a side effect that we could have predicted: embassytown. People with tons and tons of embassies cluttering their city because they were really great buildings from a systems perspective. The problem was widespread, especially for the people who really liked our game and were active players! We added bulk embassy buildings to solve this problem.

We iterated very heavily on the PVP interface, but even then things weren’t perfect. People didn’t like that the system puts very heavy preference on the culture system because the rewards keep you gated by cultural points. We did like the idea of players coming across the gate and having to build their city more, but if it makes users angry, it’s not ideal. Also, PVP awards don’t really scale, not with overall experience, nor with risk-taking on the player’s part. There’s also no narrative to PVP so more casual players may not be interested.

Lucky for us, social games are a live service. So the problems addressed here we’re able to improve. Our roadmap is robust. We’ve introduced a colonization aspect where players can explore an island and find new buildings. It allows us to grow the game without changing the core. It allows us to reevaluate the PVP with things like naval expeditions, etc. This is a way for us to iterate on a feature without upsetting players who might like it just the way it is.