Blow doesn't like the contemporary "urge to control the player experience very tightly" much.
Current games are, he says, "all about getting every hour of work there on the screen for every player. And I think you do lose something in that. Like the Castlevania game that has the inverted castle [Symphony of the Night]. That's a huge, giant secret that everybody freaked out about, and it was cool!
"It's not just the budget thing, but it's like there's this idea that every player has to have this certain kind of experience, and if they don't, the game's broken somehow... I don't know how that happened, but that became mainstream design thought."
Another place he deviates from the mainstream is focus testing -- he derides it as "like letting people tell you what the game should be." But he's leery of even playtesting the game, although he has done it.
"Even non-focus testing is very dangerous," he says. "For me, it is very important in a way, because I have some abstract idea of how some puzzle is going to play, but I don't really know if my guess about how that's going to play is correct in reality. So it's a reality check."
But while he's seen players stumble over his puzzles "quite a number of times," he says, "I haven't changed the nature of the game or anything. But the danger -- even if I'm just using it as that reality check -- is that any time anyone has an experience that's even slightly different from my ideal version of what I would want, I'm like, 'Oh, I've got to fix that,' but I think that that's wrong."
When you're too reactive to playtests, says Blow, "You get this totally featureless game out of doing that. No game ever made by man provides a perfect experience to anybody. And attempting to do so can easily -- speaking for me -- carry me away from what's really important about the game. So I try to rein myself in about that."
That said, he hasn't abandoned the idea totally. "It's still something that I'm figuring out. Like, when do you decide to act on something and when do you not? I couldn't exactly tell you what my rule is. I don't know."
It's clear that, given his description of how the very fabric of the game is in flux -- some of the puzzles aren't even hooked up to anything right now, because "we keep moving stuff around the map", the structures are "just kind of plopped in the terrain", and the buildings are, so far, "for the most part, just blocks that I kind of just build in the editor. A few of them are a little more designed" -- Blow has set upon a specific, iterative production process which he believes in.
He says that it's hard to exactly define when The Witness is in "production." He's planning to recruit a few developers later this year, which may mark the beginning of production. But then again, he says, "we're kind of in production now. Sort of. I don't know. We're just doing whatever makes sense."
He looks at the mainstream industry's inflexibility as incompatible with his design sensibility as expressed in Braid and The Witness. "If they think they're being cautious, they'll build a vertical slice. And they'll be like, 'Okay, this is one level of our game, and now let's go into production, and crank it all out.'"
There is no section of The Witness that is yet "done". "I've reworked a lot of these puzzles a number of times, and they get better every time. Because I can't quite see, sometimes, what the best approach to a puzzle is. And sometimes it's like, 'Well, this thing that's up here on the second floor, it needs to be on a ground floor. And the top of this building needs to be open instead of closed. Oh, you know what? It needs to be a totally different shape of building.'"
It's well known that traditional production -- particularly asset generation -- can kill the ability to iterate. "And so for me the quality of this kind of game that I design so far -- this being the second one -- depends a lot on iteration like that. I'll make something and I'll not know how to make it better right now. It's kind of a cool puzzle, but I'm not totally happy with it. And then six months later. I'll wake up one day, and I'll be like, 'You know what? I know exactly how to make that puzzle better,' and I go over and I tweak it, and it's better."
Maintaining momentum on the development, however, is not a problem for Blow, because "there is just a lot to do, and everyone's aware of that," he says.
"I've certainly lost momentum on a lot of projects in my life," Blow admits -- recently among them, mainstream games he has consulted on. "But for me personally it tends to happen when I don't believe enough in what I'm doing. If I think something's really good, I can work on it for quite a long time."
Still, sometimes he does hit creative blocks. "I had a problem with the story. The version of the story that's in the game now is version three, and I was just writing it a couple weeks ago, really, and we recorded it in a rush," he says.
"I feel good about the story now. Like when people play it, and I overhear them playing it, I don't cringe. Whereas I did before, with the first two versions of the story," admits Blow. Given the surprising ambition of the narrative in Braid, the potential pitfalls for The Witness' writing are understandable, too.
"It was sort of a little bit a little bit posture-y, right? Like, this is a game about philosophical stuff, and so here's a recording talking about philosophical stuff. And it just was kind of dreadful. But I didn't think it was going to be kind of dreadful when I wrote it."
Those first recordings Blow got back, however, proved otherwise -- but at first he wasn't sure. "I thought it was just the same kind of thing like, well, nobody likes the way their voice sounds when they record it and play it back. I'm feeling weird about voice actors who are not that good actors, or maybe I'm feeling weird about seeing something I wrote."
It taught him to follow his instincts: "And in reality it was none of that. In reality it was some part of me knew that this story was not right for the game and I was uncomfortable with it, and that was coming to the surface as just this cringing."