A look at why the gaming industry is going backwards while patting itself on the back
Last year I wrote an article listing a few simple ways in which the gaming industry could continue to make us excited without needing to rely on new hardware. I want to revisit some of the topic expand on the ideas. Part one deals with the shift from dreaming of AI programming to VR hardware.
Once upon a time, our society was interested in something called “Artificial Intelligence”. It was the idea that we could eventually create an “electronic brain” on par with the human mind using computers and advanced programming. It was the lovechild of two new fields of science that caught the public’s imagination in the 1950’s: Psychology and Computers.
Psychology attempted to sort through the brain and reduce everything to inputs and outputs, which in turn allowed us to believe we could program computers to imitate the mind within a few decades. But like so many hopes born out of the 1950’s, these beliefs were fueled by ignorance, imagination, and an unrealistic trust in science. The hybrid field known as Artificial Intelligence suffered every time either of its parents did, and eventually it was put on the shelf until something practical could be done with it.
That opportunity came with the arrival of video games. Suddenly we could watch little pixelated “people”, and they’d even show up on the screen and entertain us when we interacted with them. It wasn’t exactly the dream of an electronic brain, but it was real and had plenty of room for potential. Interacting with things that (sort of) looked like intelligent creatures was enough to fuel the prospect of Artificial Intelligence once again.
During the 80’s and 90’s, experiments with video game AI didn’t seriously further the idea of replicating the mind, but it also didn’t matter much. What did we need an electronic brain for, anyway? Game programming only needed basic behavior systems, if any. They gave a highly specific context to everything, because by necessity it was the realm of half-finished illusions, cobbled together with just enough on display to make it seem like there was a world to enjoy. Worlds became more important than programming better characters; so did gameplay mechanics, graphics, etc. When we don’t have a narrow context for behavior, we imagined it as omnidirectional and neutral, just like the human consciousness in a resting state, but once you add a man with a gun in a room, suddenly you have a very limited (and probably temporary) set of behaviors that you need to create. That sure makes everyone’s job easier.
To some extent, developers continued to experiment with better AI systems, and as late as 2003 at least one leader in the game industry was keeping the dream of the electronic brain alive. Gabe Newell, in an interview with PC Gamer, gave a statement on it. It speaks volumes about what I consider a fundamental turning point in the industry, especially when you read the little paragraph at the bottom:
Unless Valve is developing revolutionary AI behind closed doors with Half-Life 3 (which is possible) it appears that the dream of advanced AI has been abandoned by Gabe Newell and the rest of the gaming industry. It’s been 10 years since he made that prediction, and I dare say the closest we’ve come to a “synthetic character” who acts like your friend is Siri, the keyword-recognizing application included with iOS devices. Take away the human-sounding name and it’s nothing but a voice-controlled task manager.
A few days from the time of this writing, Gabe Newell is going to be revealing an exciting line of “Steam Machines” that will attempt to save PC gaming from Windows and Microsoft’s death grip. Valve is now obsessed with new business models and operating systems, and Newell has become fascinated by wearable computers, input methods, and VR. He wants to improve our ability to interact with software, instead of improving the software to become more worthy of interaction. Display headsets like the hotly anticipated Occulus Rift (which is also due to make an appearance at this week’s Consumer Electronics Show) are much more important to him now. Somebody should ask him what happened to his belief that AI would become so important.
The Kickstarter video for the Occulus Rift is packed with quotable lines. The founder of the system, Palmer Luckey, talks about how he dreamed of being “in the Matrix”, and existing inside a game. He tells us he’s passionate about games, but “even more than playing games, I’m passionate about bringing games to the next level.” Gabe Newell shows up in the video to say that it looks “incredibly exciting”. Why exactly it’s incredibly exciting to feel like you’re inside game worlds that may or may not provide interesting things to do, apparently doesn’t need to be explained.
The steady migration away from programming intelligent systems toward building prettier things using more efficient hardware shouldn’t be surprising. After all, which is easily noticeable? Which will attract consumers? Which generates better headlines in the tech world? AI is an invisible force discovered through experimentation and choices, not something you can blast on screen in a promotional trailer. Sometimes, the economics of innovation make decisions easy.
Mobile devices’ ascendance has lowered the cost of components for VR headsets while also making them more comfortable to wear, but what innovations have made AI and programming easier? The fact is, designing intelligent, dynamic characters (or systems in general) is just as hard as it ever was, if not harder.
Visual fidelity is the bane of complex game systems. The need to see in full detail what’s happening between characters and the environment guarantees that not much interaction is going to happen beyond destroying things or being destroyed. This is why simple-looking games are often more complex than complex-looking ones. Being able to walk, point, and shoot at things within immersive virtual reality is the perfect metaphor for how visual fidelity hamstrings what we can actually do.
Let me boil it down to a question. Which is more exciting to you: Dwarf Fortress, a hideous ASCII-art game which randomly generates and simulates an entire planet, including thousands of years of civilization, historical figures, battles, monuments, cultures, seasonal patterns, creatures, water physics, hunger, thirst, drunkenness, sleep, character moods, relationships, subterranean layers of all manner, crafts, trading, military organization, decoration, landscaping, and individual physiology right down to specific bones and organs that get damaged in combat–or the graphically updated version of Doom 3 playable on the Occulus Rift?
I’d say both games are terrible in their own way, but Dwarf Fortress is much more exciting to me as an example of what’s possible. If you put its visual fidelity on a scale of 1 to 10, it would be a one, but in terms of depth and interactive richness, it’s a 10. The programmer is free to design things as complex as he wants, because he doesn’t need to show you what’s happening in detail. Makes sense, doesn’t it? And, in this sense, our traditional definition of “progress” has become counter-productive. Because when we finally do have VR goggles that we can afford, we’re going to want even prettier places to walk around in. We’ll need new 4K or even 8K resolution displays shrunk down to the size of a playing card in front of our eyes, along with higher frame rates, immediate response times, and an order of magnitude improvement in the visual department–especially animation. What was that quote? “Ten years from now…..”
As we go higher on the visual fidelity scale from Dwarf Fortress, we find more balanced example of great systems and simple graphics. Ultima games jump to mind immediately, and so do early Fallout, X-COM, Planescape: Torment, and even the Pokemon series. There are hundreds of characters and abilities in Pokemon games, each with basic sounds and animations. This breadth of content was only ever possible because of the shortcuts in the art department. The latest Pokemon jumped into the land of 3D, and does a good job compensating for its small selection of animations by using lots of special effects during combat. It is thanks to a very simple presentation scheme that programmers can create all those interlinked systems we are drawn into. We don’t need to see every activity played out in gorgeously rendered animations, because we’re focused on the strategy and significance of it all.
What is the appeal of VR, anyway? What’s this bright future? Can anyone articulate the next phase, or even the eventual goal? John Carmack has said in an interview that in a “few years” we might have Occulus Rift hooked up to mobile devices, and thus become free to have a paintball match that we can digitally “skin” however we want. This is a stupid idea. He says himself that he wants to move away from “augmented reality” like Google Glass, but instead use fully-immersive VR, which closes off your senses from the outside world. Excited, half-finished bullshit is not good enough for me.
The only type of game I can imagine myself enjoying with the Occulus Rift are racing games, because the temptation to behave like a normal person is absent. I won’t feel like I should be able to reach out and touch something in the world, because I’ll be sitting inside a driver’s seat, with a physical steering wheel in front of me and physical pedals on the floor. I’ve seen too many people convince themselves that the omni-treadmill solution is exciting too; it’s garbage. If you’re going to offer me walking and running, I’m going to feel the instinct to sit or crawl too. I’ll look down at my glorious virtual legs, and feel like I’m trapped. We should have more freedom, not more forms of restriction! It’s aggravating to see such enthusiasm for a superficial, shallow, and dead-end movement when we could be working on the real systems of interaction that challenge the mind.