The go/no-go process has changed and evolved over the years. Talk me through what the go/no-go process is like today.

Yeah, so for us, it's a much more gradual process. There's not really one moment in time where it's like, is this thing ready or not? It's more of a constant review of how features are progressing. And we like to be very hands-on with the team and with the features. So typically, we will be playing the features regularly, individually, as a group or with the team that's actually making the feature, giving feedback, trying to harden it, and also at the same time evaluating where it's at. So there are several moments in time as we're going where there will be a point where we have to decide, is it ready or not? But there's no line in the sand necessarily where we go, this is definitely not going, or this is definitely going. There'll be points in the process, like we had to do with Cargo, where it's basically like, this is a core feature, it's not where we want it to be. At this moment in time, we want to push it a bit further out to make sure we've got the time available for it. But ultimately, as I say, the process has changed for us in that there used to be more one date, everyone's working towards that date. We are constantly trying to be on the ball with all of these individual features and knowing exactly where they're at and if we're happy with them.

Well, for Cargo, it was... You came to me on the day of ISC and were like, is it in 3.23.0 or is it in a .x? And I was like, I'm still deciding.

You want to air the laundry? Okay. So the day that our cargo episode was supposed to air, or did air, was the day of the go/no-go. So I had two versions of hosting, as we often do: one where it's still targeting 3.23, and one where it's not going to be in 3.23, it's going to be in 3.23.x. And you might remember we were so good this season about getting all our episodes out at 8 a.m. Pacific on the dot, because we moved editorial over here and we've done a lot of structural improvements thanks to David and Tom over there, who've made things a lot better. We had a great streak going of every episode right there at eight. And then I was like, well, what version do we put out? And he's literally in there surrounded by everybody playing it. That whole story he's telling you, playing it, going through. And I'm just pacing out behind, breathing on the window. Which version? Just to see whether it gets the go or no-go. And it was: we still don't know yet. We're going to give them another two weeks, and then we're going to make the hard call. So it was like, all right, so we go with what we call the truth in the moment. The truth in that moment is that it was still targeted for 3.23, so we go out.

And I wanted to give the team every opportunity to be able to get it into the game. But it needed to pass a certain level of quality bar to get in there. And ultimately, they didn't manage to do that. And that was the decision I had to make.