But in the case of Dota 2, was it more liking DotA, or liking IceFrog?

EJ: Both, absolutely both.

GN: For us, it's like there's some people you meet and you just say, "I wanna work with this person." And IceFrog is that kind of person. There are a lot of people at Valve who don't ever have to work again. The reason we all go to work each day is we get to work with people that we do. The idea that you can go to work each day and see what this person is doing is pretty exciting.

It's the same way with Doug Church, who is working at Valve now. It's just a total blast for me, to go in and talk with him about enemy design or user generated content. It's just ridiculous how much fun you get to have by working alongside these people.

Have you said what he's working on?

GN: No, we have not said what he's working on.

Didn't think so. Don't expect you to either.

GN: It's cool! I'm excited.

So the strategy then is to harness talent, or collaborate with talent.

GN: Yeah, I'd go with "collaborate", more than "harness." Although in [Valve software developer] Adrian [Finol]'s case, I like the idea of thinking of him...

EJ: Leashing. In Adrian's case, we leash talent.

GN: In fact, I really suggest that you insist that you get a photo to accompany that part of your article. "This is Adrian. He's on a leash."

EJ: We have Photoshop.

Now, I'm assuming Dota 2 is free-to-play.

GN: So the primary focus for us at this point is not worrying about monetization, and it's instead worrying about getting the game right. So we started with a group of IceFrog's testers that he's worked with for all the different versions, and sort of got it to a point where we'd stopped making them crazy with all of the dumb things that we had done.

And The International [tournament at this month's Gamescom] is sort of the next step of that process. It's like, this is a very tough audience; there are a bunch of clear technology pieces and server pieces we have to get done. And the phase after that is, there's going to be an invitation beta, and then after that there's going to be an open beta.

But our focus is really much on building something that's cool, and then we'll worry about monetization. So we're not going to worry about that until later. Premature monetization is the root of all evil.

Conventional wisdom suggests that you have to be aware of your monetization design from a game design level.

GN: I think not sucking is way more of an important thing to pay attention to first. I think every gamer can point to shipping too early, or sucking, as being a way more dominant story in our industry than, "Oh, it was slightly cumbersome to give the company money." I mean look at Minecraft, right? Notch wasn't thinking through his incredibly precise monetization strategy.

EJ: It's also just, do the hardest stuff first, and make the game fun; making a game fun is so hard. It takes so much time. Figuring out how to make sense out of making some money out of it, that's not nearly as difficult.

How much time have you spent on this project so far?

EJ: I think it's been about two years.

GN: How many people are working on it now?

EJ: At this point it's probably 60 or so people, I think.

Did you recruit a lot for this game?

GN: No -- we don't recruit for games. We recruit people...

EJ: Well, we recruited one.

GN: Well, he recruited us. IceFrog. That's just not -- I mean, anybody who we hire, we don't hire to a specific position or to a specific project. The people at the company don't work on specific projects. Everybody's told, "Your first job is to figure out where you can create the most value." So when people end up working on Dota, it's not because somebody told them to go work on Dota. They go work on Dota 2 because they decided , "That's where I'm going to be the most useful."

EJ: It has this really great side effect, too. Instead of having some person review all the products that are going on at Valve, you can tell how a product's doing based on how willing people are to go and work on it. We know a product's pretty likely to be successful, or fun, or at least fun to work on, if lots of people are going to working on it. It's a good method.

Do people have to commit to working on a product for a specific length of time, or a specific split of time even?

EJ: No.

Do people split time?

GN: There are a lot of people who work on multiple projects simultaneously.

EJ: Yeah. People... they're just all committed to making sure that whatever they're doing they feel like it's productive.

GN: It's more interesting to come into, "I'm supposed to deliver this by then", and that's mainly because other people have dependencies. In other words, if you said, "I'm going to do this", in terms of localization, or in terms of this feature, people would be sort of annoyed if you just didn't do it. [laughs]