Job Simulator 2050 from Austin-based Owlchemy Labs is bright, vibrant, and looks like a cartoon-ized Little Tikes play set in which you can, for example, throw raw steaks at a robot, drink a bottle of wine, and make a sandwich made of solely bread slices to serve to waiting customers. It's goofy in the best way possible.
But just beneath the humor and disembodied white gloves is a virtual reality game that incorporates some seriously solid VR interaction design fundamentals. At a recent Steam VR event in Seattle, held by Valve Software, I had a chance to talk to Alex Schwartz, CEO of Owlchemy Labs about how his team develops and designs for a new generation of VR.
What are the guidelines you follow when creating interactions in Job Simulator?
It’s not Job Simulator unless you’re grabbing it or throwing it or eating it. A menu can’t just be a menu, it has to be a physical thing. So we put cartridges on a table – you grab a cartridge, pop it [into the ‘job machine’], then pull the lever.
The game evolved a lot since I last played [in March 2015 at GDC]. Back then, how much of this new stuff did you know you were going to design into the game?
None, we’re designing this as we go. It’s crazy.
As you’re adding stuff, what are you keeping in mind? How are you thinking the interactions should work?
We found fun fairly early, and we realized it’s all about visceral hand interaction and feeling like you’re doing the thing, rather than that being abstracted. Video games have always been about hitting a button, and there’s always been some sort of delta between what I do and what actually happens. ‘Oh there’s a crouch button,’ and [you push that] then the character crouches. Now, you just fucking crouch! [In VR], everything has to be the thing that you do, and we’re trying to remove as much abstraction from everything.
"You can never know how good something's going to be in VR until you try it out with a headset, with your hands, and it either clicks or it doesn't."
As we design, it’s first about what feels really hands-on, and second, it’s about whether people are really going to understand it.
Because if we design an appliance or something, and then people read it in their mind as ‘Ok, how do I do this?’ and then they reach for the wrong thing or it’s not working “right,” we have to go iterate on that.
It’s pretty much like little kids toys, or a Little Tikes version of a thing, because that’s really what you’re like in VR—you got these mitts and you’re just smacking things like a kid does [laughs]. So we really have to simplify things into the basic core interactions.
When you find that somebody doesn’t “get it,” what’s usually standing in the way?
Usually it comes down to affordance design. You know when you have a door that you’re supposed to pull, but it says “push” on it, that’s like the ultimate failure of affordances. That’s the kind of roadblock you have, and we don’t want people to have a frustrating attempt as they try to do something.
The [food] blender went through three or four iterations to the point that it made sense to pull that lever. Blenders don’t have a slot machine arm, yet we iterated on it enough that people get it and it pretty much works pretty much every time.
Playtesting is so incredibly important, and once somebody has playtested a job, they’re almost useless to us, because they see it and they get it, so we need new people constantly coming through. We just watch for hesitations, little signs of confusion, and usually you can figure out, ‘Oh, that was supposed to be X.’ Then we take notes, go to the drawing board and try to rejigger something so that it just clicks.
Eat a burrito to exit the game
So you just playtest throughout the entire development process?
Yeah, we’re trying to ramp up playtesting because now we’re in the final two-and-a-half months of development, trying to get the core jobs done for the [HTC] Vive launch. We do Friday night bring-your-friends-to-the-office-type playtesting stuff, just have dinner catered, drinks, and just have people play, because it’s super valuable for us to have those new testers. And you can never playtest enough.
Even having press come in today [is valuable]. We have two people taking notes over little things that we see. If you played it a week from now, anything that you might’ve gotten stuck on will have probably been smoothed out. We’ll keep smoothing out the edges.
So on one side, there are the things that impede you. On the other side, there are the things you try to do that we didn’t think of, and we need to make that the magical moment where we take it into account. Everybody reaches for that little bubble pipe that the little British guy has. People just want to steal his pipe!
So the number one high priority task—and you don’t see this much at [other studios]—is “make the pipe grab-able.” … We go down the rabbit hole of building out things that people want as we see them, and we just want that magical moment to happen.
It seems that non-VR developers might not have the same attitude towards interaction. Is that something that you had to get past?
Yeah. Anything that we think of as “this is the way this design should design this” is always wrong. Anything we think of on paper that’s gonna be a great idea turns out...not amazing. The answer really is to try it and prototype it. So we set up like five different ways a thing might work, and then we try them.
You can never know how good something’s going to be in VR until you try it out with a headset, with your hands, and it either clicks or it doesn’t. That means the fastest you can iterate is the best route for development. We’re not going to sit down and write a [game design document], because it’s going to be wrong. We just try it out in VR.