Oh, this hits the goods in me.
I just got affirmation that my ramblings are not on the wrong side of tracks. I mean, Ken Levine of Bioshock fame recently floored the audience with the latest iteration of Bioshock franchise. He (ok, the team) didn’t do that by reinventing pixels, but by giving the artificial characters incentives, agendas and ability to act true to the context they’re in. Giving them, hopefully, traits and unexpected people personalities that make the journey from begin to finish a little less lonely tube affair. Well, in practice, it’s not that rosy but damn close as it’s not scripted into unconforming timeline. It rather tries to react to where the player goes, what he does and what’s around. Mirroring and angling the surroundings.

I mean to say, ahem, damn. Yes. This is how narrative and being-there experience and involvement and all the other once-vapour golden ideas will be done. By recognizing the need for them, and then shoveling resources at them like it was nobody elses business, because it’s not impossible unless you keep blindly listening to grey-faced suits who project future core targets based on what has sold in the past five years. Don’t look there. Future ain’t in the past, unless you take into account other mediums and forms of storytelling. Like, I don’t know, books or television series with people in them. Possibly interacting with each other. Ok, that was slightly on the trolling side but how else can it be said?

I’ve often heard how a properly done AI companion could feel like a buddy on coop-mode at your side. Watching the AI / NPC gameplay video, that goes right down the drain. Know why? Your buddy is never present.

Let me open that one.
When your best mate evar is playing alongside with you, you see only his actions and the resulting effects. Solved puzzles, blown guard towers, whatnot. What the videogame and the mission requires. However, when you turn to look at the guy or the gal you’ve spent your childhood with, falling in sync from trees and hitting the curb face first while barreling downhill with crap bicycles, you don’t see him/her. Just a badly animated videogame character that slides around and repeats the videogame motions.
No visual connection or context to tie with, nor personality shining through. Your bud can do just what the player character is limited to do, and that’s always in minority compared to AI characters who need to connect with at least the context of the story and dialogue when interacting with the player. Player model, who incidentally, stands proud and motionless like a big tree, deferred light glimmering in his normalmaps.
Point being, you can’t properly connect with a stiff slidey videogame construct that has less naturality to its movements than an average kitchen appliance.

Your NPC buddy AI companion thing is not limited, though. As in the gameplay video, he/she can project very human traits — constructed, of course, but if they’re triggered by the surroundings and situation, they can become human. They become something player can relate with. Almost human reactions, if you may. If there’s an underlying structure and balance between predictability and unpredictability, they start to give off a whiff of a personality behind the actions.

Of course, there are logistics underneath. Building personality through animation, context sensitivity, AI, sound design, dialogue and all the other cogwheels of the machine is a massive task and there’s no sense nor chance to populate whole game world with such characters. It’d be awfully nice, of course, but then designers and writers would break their heads trying to make the key characters stand out. The mass and weight of it just needs to be recognized and placed accordingly in the game, to have it impact the world and story.

In regular co-op, as fun and blast it may be, your best bud fighting alongside you isn’t going to humor you by sticking his head into a barrel and testing the echo for the fun of it. That’s not acting out in a world together. It’s more akin to perhaps scooting radiocontrolled cars around a track together. Bloody good fun, yes, but try and stick that into a narrative context and something’s gonna be missing, unless just watching events unfold from synchronized actions counts. Sometimes it does, but even then it has to be done from the get go with that in mind without shooting for what can’t be done. Recognizing the means, etc.

Of course, getting back to the gameplay video, nothing’s done right until it’s in the hands of everyone and receiving actual love and tears. So far there’s only a glimpse of gameplay video, and cynicism is easy. I for one try and be optimistic about this, as I take this bloody personally. Now, that camera and some of the strained sort of animation.. Ah, can’t have everything in one go, can I?

It’s been a while, and I feel like a worthless punk for not writing since ages. Luckily, now that I got laid off, I have free time and a distinct lack of work stress and
I feel like some proper bloggery and ranting is in order.

Also, the original blog post about communicating game worlds to the player found itself amazingly an alternate home in printed form at Gamesauce magazine, so this kind of works as a live sequel as well. As always, you’re free to disagree, and I hereby encourage you to voice your opinions and thoughts in comments. I’m not writing down the truth, just my view of things as I feel about them right now.

I have this urge to rant about how wrong game developers are with faking humans. I mean that damned AI that’s almost a trope in itself these days. Disregarding the few very good examples out there proving that NPC characters can be done right, I’ll concentrate on the bad shit since it’s much more rewarding to fling around.

So, those rigidly moving robot-like depictions of humanity that developers try to make more human by adding more bones, more polygons, more shaders, more realistically simulated hair blowing in wind and more this and more that. Yet they still keep acting very much unlike human, while looking more and more like human. No, that gets just creepy.

More graphical fidelity you shovel on screen, higher it sets the bar for animation, facial expressions, voice acting, effects and .. well, anything. More it represents reality in our headspace, more we spot the inconsistencies and you can be sure they stand out and break that careful world construct. I’d rather see blocky graphics with fluid likeness of life infused all over than all the graphical provess with limited animation on those photoreal humans. Of course, gamers are used to the videogame looks and accept a certain level of “videogameness” as part of the media, but if you want to push the narrative limits of visual information, you need to have a broad vision to account everything it’s begging to be pushed with. Uncanny valley is most often associated with human realism, but in reality it looms over everything we’re able to associate with any close matches in our own experiences. It could be from our daily lives or from our favourite movies – object of association doesn’t matter. Just the very existence of association matters. If you try to deliver associable visual experience, it damn well should match as a complete package and not just as isolated masterworks of programming that’s easy to present as slides at GDC.

I honestly believe we could get away with less graphical fidelity, and instead the fake humans should pose at least a few common human traits, or flaws if you want to play dirty with words. Human flaw – or what makes us human from outside observers perspective – is how we react, mostly.

Yeah, I’m distilling observable “human” qualities to observed reactions, disregarding polygon counts and texture resolutions and other whizbang trickery engineer-driven development so commonly focuses on. You can have the world-class animation but keeping it alive is the hard part. We bang our toe, we break from being perfect beings (observed as rigidity!) for a split second, reacting with a sound, facial expression, and some gestural motions. We are animals, for a second. We come upon a closed door in unknown hallway and we react by glancing at the door handle, maybe trying if it opens. We are curious monkeys. We don’t just walk against it and tread the ground with futile steps against the collision. If the door doesn’t react back, we disregard the door and focus our attention elsewhere. We don’t have the patience of statues. If the leader of our group isn’t going anywhere, we start looking at some things, fiddling with others. We keep ourselves busy even when we’re idle by definition, by reacting with glances and manipulation of random stuff. We are playful children! Of course, by “we”, I mean a random bunch of AI posse following the leader, i.e. player. Remember, player observes you, the AI, and gets suspicious if you look real but act like a robot with four hinges and a three-line script for brain. Player starts to look at the videogame flaws instead of human flaws.

What’s human, anyways? Scratch that. What’s human in drama? We want to play (and make) games that brings a feeling of something larger we can lodge ourselves comfortably into. It’s a vague term in itself, but sense of drama is what we want to push into the undercurrents of gameplay – the communicable game world player “gets” without having to read it up in the manual. Rules of drama dictate the basics, and few of them could be considered here. Character has a goal, and is driven towards it with a motivation. We don’t have to go to deeper stuff with midpoint-slumps and conflicts with antagonists and so forth, because we’re not dealing with primary characters here. Primary character is the player, and that bastard breaks all our careful rulesets anyways. Player is the uncontrollable variable we hate to incorporate to our games. Anyhow. AI represents the secondary characters, and because gamestudios are so often tech- and sellingpoint/marketablefeature-oriented, AI – as a companion – is in most cases a blind, daft sidekick that stops to loop its idle animation when player stops, and will resume moving when player moves. Unless it gets stuck in obstacle, in which case you’re again playing with videogame flaws, not human flaws. ANYHOW. If you have AI companion throughout the game, or even only during parts of the journey, it should appear to have its own reasons to tag along, its own reasons to be suspicious of you, ambitions to play you for its own goals and even backstab you after calling you dear in the night. All that because that’s human. We’re mentally dirty, incomprehensible beasts with varying views of world and of each other.

Yes, it takes more care, more focus and more work to develop but in the end, wouldn’t it feel more rewarding than just witnessing next-next-generation shader tricks or ten thousand polygons more on-screen?

Which route do you take, technical or human?

It doesn’t take huge elaborate systems to bring a feeling of those things happening around the player, that sense of hidden layer beneath the more readily apparent game surface our player and world is limited to interact on. It’s that layer under game world where player can’t go, from where he should get this feeling of being subject to observation. Perhaps even a feeling of being judged, silently.

It’s in the script, in dialogue, in animation and in the AI system, with some thought to tie everything together.
We observe details consciously, but sums of details we understand subconsciously. Emotion is built from subconscious, no?