How did you convince people internally thatThiefwas promising?
We got some very early
prototypes that were
super, super-rough and
only tiny bits of that code
were used in the shipping
game. But that was just
guys on patrol, noticing
you or not, and you duck-
ing out of the way. And we
did do a bunch of mission
write-ups for the Dark
Camelotstuff, not that we
used them in the real
game, but we thought,
yeah, that would be kinda
cool. You’d sneak in and do
this and this guy would do
this, and in the end you could make this happen. I think we had a critical mass of cool lit-
tle elements. That said, it wasn’t until right when we shipped that it all came together
into something that worked at all in a way that players might actually want to play as
opposed to intellectually “this could work” way. We had a habit by that point of pulling
games that were lots of different elements together so that you had to connect the dots
in your mind until near the end, which isn’t exactly how the industry likes to work these
days.
Would it have been better to be more cohesive earlier?
Well, certainly it would have been great to be more cohesive earlier, though I’m not
sure what we would have had to sacrifice to do that. Obviously, these days in the indus-
try we try to get a lot more working early on, but I think that means it’s harder to take as
many risks. More to the point, it’s much harder to do games that require a lot of sys-
tems. And in a way that’s good because super-overcomplicated games rarely work. But
if you do it right I think you can have a lot of systems that work together in a very ele-
gant and transparent way, but that’s the kind of thing that’s very hard to show right now
because you’re told, “All right, well, before we do any real development, we need a pro-
totype that works,” and that means you’re only going to be able to do a few new things
at a time built upon whatever you did last. I think that makes it very hard to do some-
thing as simple asThief, which is an incredibly focused game, but it still requires light
and shadows that work, and shadow detection that works, and AI that can understand
shadows, and a speech system so that the AIs can communicate to you about the shad-
ows. It’s not rocket science, but it’s a lot of stuff. Like I said, we hacked together some
stuff at the beginning to give us an idea that, OK, this could probably work, but I’m not
sure it would have ever convinced a publisher it was going to work; obviously it gener-
ally didn’t convince Eidos. It convinced them enough to fund it, and hey that’s great, but
I’m not sure that would be true any more. I think the stuff we used back then would be
518 Chapter 26: Interview: Doug Church
Thief