Game Development Reference
Interview with Exploding Barrel: Give Them
What They Want—cont'd
logging system too, which shows us in real time everyone who is logged
in and lets us track all of their behaviors. There is a crash reporter that
sends off a report and a call stack if there's a crash.
SB: But of course, you guys write bug-free code so that never happens.
JH: Exactly! [Laughs]
Q: Can you tell us a little about the technology in the game, especially
the modifications you've made to help people get into the game quickly?
SB: There's a good lesson here about why you need to pay attention to
data and pick one thing at a time and try to make it better. If you're trying
to fix ten things at once with a thousand different levers, it's just not pos-
sible to infer anything relevant from the results. There are too many differ-
ent variables in that equation to solve—ever.
When we launched our first Facebook game, we had a big problem:
70-80 percent of people who showed up didn't want to install Unity, and
they would leave forever. That's a big problem … in fact, there's no bigger
problem. So we really dug into that once we had our backend up and we
knew we were getting the right kind of data and could ask the right kinds
of questions. We worked carefully to turn one knob at a time to drive
that number down and reduce friction. We were able to make it a very
friendly, easy way to get in, and not have a big bad install screen hit you
in the face and bury a file somewhere that is hard to find.
HP: When you're looking at the data, don't let the pendulum swing too
far with your solution. When you look at the data and determine you
have an issue, don't think “Oh my god!” and try to go completely the
other way to try to fix it. You'll fail. What I'll tell everybody is to make
shifts in increments. Make sure you're asking the right questions and
don't let the pendulum swing too far.
SB: It's just like tuning gameplay. If you have a bunch of levers and you
change 49 of them and get a really awesome result, you will have no idea
how you got there. You lucked out.
The nice thing about social games is you get your data back immedi-
ately. The more people you have playing, the faster you get good data. It
doesn't take long to find out what is going on and test out a fix. You don't
have to wait days, because you get immediate feedback. For us, it's like a
stock ticker. Anything we want to track, what anyone who plays our game
is doing, we can see in real time. We can know the second someone buys
a specific item.