where you could see the AI’s path. A lot of that game comes down to how do you get clo-
sure moments and how do you get scope.
What do you mean by closure moments?
In most games you’re killing everyone, so you get nice moments of closure when you
win each battle, whereas inThiefyou’re going past people. Which means all those
encounters are open-ended. It’s like all the sentences start but they don’t end. So a lot
of time was spent on how do you get closure from these encounters that didn’t really
have closure. And a lot of that had to do with building spaces where the player can really
see what’s going on. If it gets too claustrophobic, the player doesn’t really have any
idea. In a lot of the early levels the AIs were on these incredibly interesting and com-
plex paths which looked great on the overhead map but when you were playing the
game it felt like it might as well be random. Because you’d just have no idea what was
going on: a guy would show up and then he’d be gone and then he’d show up some other
place and you’d forget to hide a body but you’d have no idea if someone was going to find
it. Because the player only had a very local sense of what was going on, we had to
change the scope of the AI behaviors to be very local as well. Otherwise it just felt like
randomness. And so a lot of the designers’ challenge came down to how do you build
these spaces that can run in the engine fast enough, which certainly had a whole set of
constraints about size and so forth, but at the same time big enough with clear enough
line of sight or clear enough iconography. You had to be able to say, “Oh, this is that main
hallway and it looks just like that other main hallway so I bet the guard’s on this rotating
patrol through this hallway. OK, I get it. I better go hide the body off the hallway.” Ways
for the designers to make it possible for the player to make rational plans, given that the
player couldn’t bring up a radar or switch to the God’s eye view and go, “Oh, I see.”
How to keep that first-person immersion of “Here I am, what’s going to happen” with-
out making it so opaque that you might as well flail around randomly and hope you win.
As with your previous games, was having fairly non-linear environments one
of your primary design goals?
Yeah, definitely. There’s
no question that we were
always about maximizing
the player choice as much
as possible and I think that
inUnderworldin particu-
lar we had a lot of
watching people do things
we hadn’t expected or be
clever in ways we hadn’t
expected. Or just watch-
ing where a couple of
systems would come
together in interesting
ways. You know, “I’m
Chapter 26: Interview: Doug Church 521
Thief