So we have a stable candidate, QA has signed off and the US says this candidate is good, you can push it live. What happens at that point?

Well, certain things happen. Yes, you have a candidate, you feel pretty good about it. I think in the case of what actually happened, we had one issue, and so we had to rebuild for that one issue and did the testing to make sure it was fine. And that's why we pushed to PTU to everyone on the Friday and then actually went live on the Saturday, because we wanted to give it a full cycle of PTU with a lot of people in there to make sure it's good to go live. And then on the Saturday we felt good about going live. But during that period of time, while QA is signing off, obviously the LiveOps guys are making sure it's all ready to go, getting all the information together that we have to do. And of course there's a whole community side of making sure we deliver the letters out to everybody, letting them know this is all coming and so forth. So typically, while this build is going on, every now and then the build might break. And so you sit there in those last moments going, well, we're kind of out of time, hopefully this build will be fine and there won't be any issues. That's what's going through your mind.

Fortunately, just to talk about stuff we've done this year that made a massive difference to our development flow: the build system used to take us three, four hours to get one build done. And now, because of all the great work the guys have been doing, we can knock builds out in less than an hour. And that makes a huge difference. So rather than doing one or two builds a day and clogging up the whole system, now we can just churn out builds. And that makes a massive difference to the iteration and getting stuff done for everybody in the company, and also getting stuff out to the community.

I think that's something that's sometimes lost, that we're not just making a game, we're also making the way that we make the game.

Look, we're building a company, we still are, while building technology, while building a game. And so all the time we're refining our processes. We'll do something and go, that's too many steps to get where we want it to be. So we basically then go back and say, okay, we need to create a pipeline or a tool chain that makes it easier. For instance, to get a character in the game it has to go from this person to this person to this person, and you go, well, that's crazy. So how do we get this automated to knock out as many steps as possible, so why can't just one person do most of this and then maybe hand off? So all the time those discussions are going on, and this year we have a huge roadmap. We did loads last year, we've got even more this year, because the idea is we want to get much more efficient, so as we're bringing more content, it involves fewer people who can get more content out there.