It’s true and weird that while both Half-Life games are considered influential and important, hardly any of what was uniquely compelling about them made it into anything else. It’s still hard to find anything that feels like it does what they did: dynamically-chatty bit-part characters with interesting behaviours, NPC classes with specific attitudes to each other, dozens of elaborate one-off scripted behaviours, the continuous sense of not being the main character. Both games are full of stubs of situations that you could build a whole game around.
But a lot of their systems felt complex and amazing because they weren’t the primary focus. You weren’t engaging with a given system exhaustively enough to see the edges of it. These games weren’t afraid to give you a little of something that felt complex, and trust you to imagine a depth that wasn’t always really there.
Something that happens very often in game development is the encroachment of investment-style thinking onto the creative process: if something is at all complex, and it isn’t major, a pillar, something we’re sure about, it’s seen as expensive. And if it’s expensive, you need to make a case for it, and that means putting yourself on the line, and to do that convincingly, you need to make decisions up-front that rely on information you won’t really have until it’s implemented. You need to iterate, because that’s the only way design really happens, but people don’t really want you to. And if you do implement something cool, you better use it often. We paid for this!
But the more frequently you deploy a cool thing, the less value it has to the player. So very rarely, now, does a player of a video game get to see something that’s both cool and rare.
Some things are cool because they’re rare, and that’s the whole reason. You can implement a cool feature, say “wow, that’s cool, we should make that happen more”, and just like that, it’s not cool. It’s no longer a story players tell each other. The first time you play Splinter Cell Conviction, and you’re hanging off the side of a building waiting to pull another guy out a window, and a particularly savvy guard actually leans out a window and checks the side of the building – that’s amazing! What a smart, weird thing to do! But then they do it again, always. It’s not reactive or rare, like you thought, so it’s no longer cool. You won’t be saving that clip.
Compare that to FEAR, which you can play a million times, and then one day you’re chasing a guy who’s panicking because you killed his whole squad, and he tries to get away by diving under a truck and frantically crawling. He uses an animation for this that you’ve never seen, because this is your first time chasing a scared guy near a truck. Or getting cocky in Alien Isolation, because after 20 hours, you’d built a mental model of the alien’s locomotion that didn’t include it clambering over a certain type of desk, but it can, and now you’re dead. These things are powerful because they’re rare. AI feels smartest when it handles a situation gracefully that hardly ever comes up. Without that, no amount of clever behaviour will get you there.
I always feel that, after you spend a bunch of time working on a cool system or event or interaction that can occur, the best finishing touch you can put on it is to make it almost never happen. I don’t think this is as expensive as it sounds, as long as you’re able to avoid, in the first place, the pitfall of clotheslining your own design process on the arm of what is notionally diligence but is actually paralysis. You build systemically, and you empower developers’ individual creativity, and you trust the player to cooperate with an experience. I really reckon most of ’em want to.
Leave a Reply