That's what I find really fascinating. It's a well-known fact that what game designers might think is cool is not necessarily what users might like. But it's also true that potentially what users might think they want isn't exactly what they want.

AT: That's exactly right. If you ask users, "Would you want this thing?" If you just base your entire game design evolution based on strictly what your users tell you do, you'll probably fail. Any input has noise. Any data we collect has noise that needs to be filtered and cleaned up. With qualitative feedback, users also have noise that we can filter and then take that as input.

How do you obtain qualitative feedback? Is that just simply community interaction?

AT: Yeah. We have a very, very active forum, as you've probably seen. Community users usually give us feedback all the time.

They give us more feedback than we can process. So, there's never a lack of qualitative feedback from users, which is one thing that we really appreciate, because they are really passionate about our games, and they just tell us, "Hey, here's what I'd like you to change."

Do you think games can run indefinitely? Can FarmVille last forever, essentially, as long as you keep updating it? Or is there a certain saturation point?

AT: So far, it's been running for a year and a half, so we don't know. I guess we'll find out!

Popular television shows come to an end. MMO audiences drop off after years. It's an open question for the social game industry.

AT: It's an open question. And the one that we're also looking to find out, and we're also looking to expand our lifetime as much as users want us. So far, users don't seem to be complaining, so...

Retention is a real concern, obviously.

AT: Yes.

Have you honed in on like the right ways to retain users? Is it through content updates?

AT: Yes. Well, that's one of the basic things that must be done, but it's definitely nowhere sufficient. It's a basic thing that you need to have constantly updated things -- not only content, but also feature updates. Like I said before, like our farm game at launch had this many features [gestures, expanding his hands] it now has this many features. That's one of the reasons why you see the retention there.

And two, is to kind of be able to keep up the quality of games. Is it stable? Is it fast to load? Is it easy to play? Is all the content clearly labeled, to users? The basic user experience needs to be kept up to speed.

And third, it's just servicing on the community side. Are you listening to users? When a user complains, are you listening to them? Are you responding to them? Are you ensuring that there are no people trying to hack the game?

It's a lot of those things combined that eventually result in this thing called "retention". So, you need to do all these things right. I'm sure we could be doing a lot more things. Again, it's a matter of being in a young industry. We have limited resources.

Playfish has discussed having creative and metrics at the same level, so there's a feedback between them -- no one's telling the other what to do. Do you have a similar philosophy at Zynga?

AT: I think we are metrics-driven. It depends on what you mean by "creative". Like, what is creative? People's definition of creative is very, very different. We ask different people... What is creative to us? [If] people like it. Many times, to a game designer, what is "creative" is what is new. "What I think is creative."

I think we want to leave that judgment to the end users more. The end users will tell us what they like, what's creative. In fact, they'll give us a ton of ideas, too. So, I think that's where we differ. We want to drive as many things as possible through metrics and achieve, in the very beginning, not a balance, but a really, really integrated effort between metrics and creative. I think they can exist both in the same time. Very much so.

A lot of traditional game people sort of recoil from this idea, this sense that their creativity is being shut down, but I think if you look at it as, "I have some ideas. Now I can find out which one's right, which one people respond to," it's more appealing.

AT: Exactly. That's exactly right. Everything can be improved, because you're not doing a painting where everyone can just sit back and appreciate it. What you're building is a consumer product. Users have to use it, have to touch it, have to play it. As soon as that happens, you're in a different category than how creative an artist is.

And this is why I say game building is a craft; it's not painting. To build a cool looking bowl, first of all it has to be a bowl first. It has to be functional first. And different people have different ways of using that bowl, of looking at it. And you take that feedback and continue to improve it.

Everyone's creative. I have 10 million ideas if you ask me today. But whether or not users will like those ideas, well, let's ask the users. And metrics is a way to ask the user in the right way. They'll give you the answer to pick which creative idea, and once you've implemented that idea, how to keep iterating, keep on improving it.

I think in the traditional and console mobile industry... new releases are very, very expensive. New sequels are very, very expensive. But for us, we're Flash-based, PHP-based. We can change like that. That enables to continually improve an idea.