Intermittent Mechanism

Er, that is, “Let’s Study Half-Life 2, Part 3″ confirmed.

(That is to say, it’s live now on Youtube.)

This video is the longest so far, and will likely be the longest video of the whole seven-part series. Putting these together has been taking a lot out of me, so expect somewhat shorter installments in the future, along with longer breaks between each installment. (I won’t be posting more than one a week, now.)

Transcript below the jump.

3.1 All Around Me Are Familiar Faces

So, I’m picking things up in an exposition scene, one that shows off Valve’s usual approach to exposition scenes. That is: it’s not a cinematically-edited cutscene, where the player gives up control of the view, and just passively watches the events play out. It is a scene where we’re free to wander around as NPCs deliver dialogue and walk around, hitting scripted marks on the map.

I’ve always admired Valve’s resolve throughout the Half-Life series. (And Gearbox’s, too, when they made the Opposing Force and Blue Shift expansions.) But I’ll fully admit that the results can sometimes be clunky. The faces still look good, even by today’s standards, but the bodies are wire-y. This was before the age of motion capture, and also before Valve updated its suite of NPC-staging tools, so the animations are stiff and un-dynamic. And Gordon Freeman doesn’t move at a pace that feels natural for these scenes, at all. Everyone else is trying to have a conversation, and every time you try to move it feels like you’re sprinting across the room.

In this scene, we’re introduced to the character Isaac Kleiner, who continues a trend we’ve seen so far in Half-Life 2. The original Half-Life didn’t really have characters, outside of Gordon Freeman and the G-Man. The other friendly NPCs that populated the Black Mesa Research Facility weren’t characters so much as generic character classes. There was the bald scientist. The black scientist. The stern scientist. The security guard. There could be multiple instance of this class wandering around on the same map at the same time. And over the course of the game, you’re likely to see each of this character types die multiple times—either in pre-scripted sequences, or because you lead them into certain doom after conscripting them to act as allies.

In Half-Life 2, Valve decided to turn these anonymous and abstract types into genuine, individual characters. So the generic security guard type becomes Barney, former Black Mesa Research Facility security guard and current undercover resistance member. Generic bald scientist type becomes Isaac Kleiner. Generic black scientist type becomes Eli Vance. And although we won’ t see him in this game, generic stern scientist type becomes Arnus Magnusson.

Anyway, Alyx heads off to her father’s lab, the so-called “Black Mesa East.” We’re supposed to go, too, but the teleport that Kleiner has built malfunctions, and we end up just a few feet away. We’re given a goal: reach Black Mesa east, by traversing City 17’s sewers and canals. That goal will provide the context of the action for the next several hours of the game, which I’ll be condensing in this video.

Anyway, we’re out of Kleiner’s lab, alone again. Barney throws us the first weapon of the game. It’s the same first weapon Freeman collected in the original Half-Life: a crowbar that can be used for melee attacks.

3.2 All By Myself

The moment that Barney throws you your crowbar and you set off into City 17’s sewers marks the beginning of a very long stretch of Half-Life 2 that Gordon Freeman spends alone. There are two big stretches like this in Half-Life 2: first, between Isaac Kleiner’s lab and Black Mesa East, and then again between Black Mesa East and Nova Prospekt. During both of these, we briefly interact with a small cast of characters, clustered in various resistance outposts. But, for the most part make our way forward alone, whether on foot, airboat, or dune buggy.

The game can get a bit lonely during these stretches. And they’re a missed opportunity. Whenever Gordon Freeman is alone, Valve doesn’t get to show off their top-shelf voice talent, and the Source Engine’s still-impressive lip synch and facial animation. They definitely wanted to avoid this trap in Episodes One and Two, which is why Alyx Vance remains a near-constant companion in both of those.

But you’re alone for long stretches of Half-Life 2. In this regard, Half-Life 2 is similar to most linear first-person shooters of its era. Even still today, it’s not uncommon for your player-character to not encounter any friendly non-player characters for large stretches of a single-player first-person shooter.

And there are obvious technological and business reasons for this: Making friendly NPCs is a very resource-heavy endeavor.

First, you’d have to take the time to create unique faces for your characters. Players are fine mowing down literally-faceless enemies in first-person shooters. It conveniently de-humanizes them. Mask-wearing enemies are ubiquitous in first-person shooters, whether those are the military-style face masks, KKK hoods, masquerade-style party masks, vaguely luchardor-style masks, or vaguely-luchardor-slash-ninja-slash-hockey player … I dunno, whatever this aesthetic is supposed to be …

But players probably aren’t as keen on having friendly characters wearing masks. Masks are kind of creepy. If a friendly NPC wears a mask, players are probably going to guess that they’re untrustworthy. So truly friendly NPCs mean resources spent on faces, and facial animations, and lip synching to dialogue. This is intensive work, and it’s cost effective to skip it. Survey the history of 3D games, and you’ll find a lot of friendly non-player characters that we only ever talk to on the radio. Or maybe we only ever talk to them through a windowless door. By the time we reach them in the flesh, there’s a good chance they’ll have already died. Or, they’ll be revealed as our enemy, which means they’re a hostile NPC, rather than a friendly one. Sometimes they’ll have transformed into some sort of monstrosity. This lets developers get away with giving them really dumb, monster-level AI, as opposed to smart, friendly-character AI.

If developers are willing to spend more resources, you go a few steps up. Friendly characters might converse with us, in scenes where they have a face and moving mouth, but no body. Or, they might have a motionless body with no jaw. (Horror games give developers creative ways out of needing to do lip synch!) If developers do splurge on a fully-animated  body, it will often limited to a non-interactive cutscene, or, if it’s in a user-controlled part of the game, it will be enclosed behind a wall of unbreakable glass. In these cases, they’ve modeled a whole person, and animated them. But at least the animation doesn’t have to dynamically adapt to the player’s movement in space. And you don’t have to make friendly-fire contingency plans for when players inevitably try discharging their weapons during dialogue, just to see what happens.

Anyway, back to Half Life 2. In this section of the game, you’re alone. You’ll be alone many times throughout the game.

The pacing of Half-Life 2 isn’t perfect. People complain about the vehicle sections, in particular, as being far too long. I don’t personally share that sentiment, but I think I understand where it comes from. The vehicle sections in Half-Life 2, alongside most of the foot chase business, are very lonely. Half-Life 2 doesn’t have the stretches of friendly camaraderie that you get in some of Halo’s vehicle sections. I can understand if this sense of loneliness makes these traversal sections feel aimless and over-long, for some.

For my part, I think Valve deserves a lot of credit for sticking with their guns here, and not going the obvious route of adding a voice on the radio.

3.3 The Voice on the Radio

The voice on the radio is one of the most ubiquitous tropes of first-person shooters. And videogames, really.

The voice on the radio is convenient way to add a character to the story, while foregoing much—or all—of the work of animating them. It’s also developers’ go-to tool for any and all exposition needs. Need to explain what this place is, and why the player is stuck there? Voice on the radio, reporting for duty. Need to tell the player what they need to do next and why they need to do it? Voice on the radio, to the rescue. Go ahead—open a ticket. Tell us what thing we need to fix. Or need to destroy.

(Ah, thanks, Doomguy. Finally, some peace and quiet.)

The “voice on the radio” is something added by a writer at a late stage in a game’s production, after the developers have decided on the string of set pieces they want. It has a function: to provide necessary exposition, and context for the actions we’re supposed to take in the game. It does its job, but it’s awkward.

The voice on the radio posits that somewhere in a world, someone’s watching your character, monitoring your actions so closely that they’re basically omniscient. But the flip side to their omniscience is that they’re holed up, hiding. They refuse to come out into the open and do things themselves. Everything, from the tightening of a bolt to the blowing up of a nuclear power plant, needs to be done by our character. They are the eyes, the ears, and the brain. We are merely the hands.

Theoretically, the voice on the radio is supposed to emphasize the importance of what the player is doing. But too often, it does the opposite. It makes us feel like we’re being assigned busywork. We’re just a handyman, pressing the buttons that our more-important bosses can’t be bothered to press. We’re an errand-runner … sometimes literally.

Half-Life 2 even makes fun of this, when we’re asked to operate the teleporter at Isaac Kleiner’s lab. The joke here is also a lesson, illustrating what not to do in game design. Don’t have an in-game character explicitly assign the player a simple task, like pressing a button. It robs them of a feeling of agency.

And for God’s sake, don’t to hide how simple the task is through gratuitous technobabble. You’ll just end up making yourself look worse. This is especially bad when tasks are nested into other tasks—here, the technobabble tends to become really over-complicated and pedantic, to explain how this particular sub-task is different from previous sub-tasks.

Now let’s look at what Valve does, in the absence of a voice on the radio.

So, this is just a simple valve puzzle. It’s Valve’s namesake. It’s old-fashioned, directly tied to the old water physics they were happy to show off in the first Half-Life. It’s in no way revolutionary. But it’s handled gracefully. The player is just barreling forward, the only way forward, and lo and behold there’s an obstacle. And they have to find some way to get beyond it. That’s it. That’s the entire setup. There’s no one blabbering to them on the radio, telling them how important it is that they fix these pipes, so that they can restore the water at a resistance camp, or whatever. We solve this problem not because we’re told to, but because it’s there. It’s in our way.

And even when Valve does use voices on radios, it’s in a completely different way from the traditional “voice on the radio.”

3.4 Push vs. Pull

One way to describe the difference between how Valve motivates the action in their games, as opposed to how many other developers do, would be to differentiate between pushing the player vs. pulling the player.

I’m borrowing some buzzwords from the field of education, here. Just to briefly sketch out this dichotomy:

“Push-based” educational styles are guided by the teacher. The teacher asserts their authority to decide what’s important, and constructs a curriculum around these decisions. This curriculum is pushed onto students, via content and assignments, which all arrive with the implicit assertion “you need to know this stuff.”

“Pull-based” educational styles are guided by the student. This is hard to enact in a traditional classroom, but it’s a frequent goal of things like e-learning and career re-training. Whether your’e developing educational games for adolescents, module-based adult-oriented career development courses, you have a common goal: you want the student to be able to latch on to what interests them the most, to follow what they’re passionate about, or captivated by, or simply find fun. This is “pull” learning.

Now, I don’t intend to defend the track record of educational games. But I think this philosophy of “pulling” is useful. It’s something that a lot of story-driven single-player games still struggle with, to this day. As players, we’re constantly being assigned stuff to do, by a voice on the radio. We get explicit instruction. Arrows tell us where to go. Waypoint markers count down the meters to our next goal. We’re constantly being hounded to hurry up, and being berated when we fail.

That’s why Valve’s work back in 2004 still stands as such a shining beacon. To go back to Station 7: I can think of a lot of games where, as the player approaches this station, a voice on the radio might come on, and say something like:

Vozzy McRadio: “I’m getting radio silence from Station 7. Would you Kindly use your EMP to re-route the phlebotinum injection manifold into their comms array?”

Then the player would feel like they’ve just been assigned to go explore this station. But Valve doesn’t assign their player tasks in that way. Any motivation to explore it isn’t imparted to the player-character by the game’s other players. It’s left up to the player themselves—either their sense of curiosity, or their need for supplies if they’re running low—that sort of thing.

Station 7 is an optional little side-area. But Valve employs this “pull” strategy consistently, even in the non-optional areas of the game. So, for instance, at a certain point, the player’s way is blocked by a canal lock closing. In another game, you might get a voice on the radio here:

Foni Sans-Wire: “Looks like the canal has been closed! According to the plans I’m seeing here, there should be a way to open that canal lock inside the building to your right.”

But there isn’t any. There’s just silence, and an environment around us. And so we look around. Might as well dock at this little docking area. Then, going inside, and—what this? There’s enemies in here! Oh, but there’s a cool new gun, too. The magnum! Sweet. Guess I might as well keep going this way. Gotta dash between the crates … away from that helicopter … man, that thing’s annoying. And then I shoot a bunch more guys … oh, look at this. Okay, I can just … problem solved.

Every single step of this feels like a discovery. It feels like we, as players, didn’t know what to do, so we just explored and stumbled upon solutions. This is pure pull. We’re never assigned any tasks. We just discover what we need to do on our own. We don’t get confused and lost, because the solutions to our problems are simple. We might even happen on a solution without even fully realizing the problem. But it’s okay that the solutions are simple, because the game isn’t being patronizing. It’s not explicitly assigning us menial labor, and then patting us on the back when we do that labor.

Half-Life 2 is filled with puzzles, but almost none of these are announced as a puzzle, arbitrarily assigned to us by some character as busywork. Our way forward just happens to be blocked. And since we want to keep going forward, we poke around. And we find that, by manipulating physics objects in the environment, we can find a way around this blockage, in a very straight-forward cause-and-effect way. These puzzles are all incredibly simple, but because they follow a basic physical logic, they make us feel like we’ve applied some actual intelligence.

“Pulling” the players in this way is tough to pull off. Even really good games stumble at it. For instance, I really liked Arkane’s 2017 Prey game. It’s a big and expansive first-person shooter, falling much more in the exploratory and open-ended lineage of System Shock than with the strict linear rails of Half-Life. At a certain point, several hours in, the map really opened up, and, excited by the freedom afforded to me, I decided to to follow up on a side-objective, “The Corpse Vanishes,” that had been hanging around in my to-do list for some time. It required me to go to the Psychotronics lab, which had just opened up to me, so I decided to follow my own curiosity, and take a break from the main story missions for a beat. And this is what happened:

I’m going to pause right here to point out that I did not bring the map up at this moment and interrupt December’s radio call. The game did this itself. It interrupted its own exposition, by forcing me to look at the map at that particular moment. Anyway, moving on …

That is such an atrocious pile-up of voice-on-the-radio exposition, interrupting my attempts to explore on my own, and genuinely interfering with my efforts to stealth my way around that monster. Arkane had already successfully pulled me into exploring their world, but then all of a sudden the soundtrack and UI got cluttered with multiple strands of pushing. So trust me when I say that even masterful developers stumble when it comes to leveraging curiosity rather than rotely assigning tasks.

3.5 Player, Heal Thyself

Alright, there’s a topic I want to get into right now. But I think I’m about to die. So just … give me a moment to … catch my breath … okay. Everything’s good. I got a good rest, not dying anymore.

One of the key ingredients that makes Valve’s “pull” strategy possible is the fact that the game uses health packs, rather than regenerating health. Regenerating health became really popular in the years following Half-Life 2’s release. And there are reasons for this. In the wake of Halo, first-person shooters were finally becoming a major genre on consoles. Call of Duty sold absolutely insane numbers on the Xbox 360 and PS3. And if you’re sitting on your couch, and just want to be immersed in some spectacle, rather than sitting at a desk, ready for some twitch-precision aiming, regenerating health is a nice feature. It streamlines things. It’s completely forgiving of small mistakes, only really punishing huge mistakes. But it has the side-effect of shaping player behavior in really distinct ways.

For an example, take a look at some footage of me playing Resistance 3, which release on the PS3 in 2011. Insomniac Games, the developers of the Resistance franchise, had experimented with different player health mechanics over the course of its three mainline entries. In the third game, they decided to go the pure, old school health-packs route. But in this footage, I’m playing the game in the mindset foisted upon me by the norm of regenerating health in console FPSs. This means I’m playing very cautiously. I’m favoring sniper-type weapons. I have the high ground, and I’m refusing to concede it, even as a friendly NPC invites me to jump down to him. I’m sticking to the cover that I have. And, in a game with regenerating health, this would have worked! Nothing could have flushed me out of this relatively safe position. But because this is a game with health packs, my health is slowly being sapped away this entire time. No matter how much luck I have on my side, I can’t completely avoid getting hit when I step into the open. And so my health bar … slowly … dwindles to nothing.

I’m behaving this way because that’s how regenerating health encourages people to play. But Resistance 3 isn’t a game with regenerating health. It’s meant to be played in a different way. It’s designed to encourage players to be bold. The correct way to play this sequence is to take the zip-line down, so that you’re right there in the enemies’ faces. Blow them away at close range, and you find that they drop health kits, which you can then pick up. The health kits are little carrots, beckoning me out of safety, forcing me to be bold, altering the risk-reward location. There they’re to make sure that, even when you have a sniper rifle, sometimes you’ll still need to emerge from your foxhole and run to another bit of cover, hoping to find a health pack along the way. There they’re to promote outright recklessness—to get you to run toward an approaching enemy, rather than run backwards, because you see that one you just killed dropped a health kit. The “glory kill” mechanic in the 2016 DOOM game is really the apex of this design strategy. Before it came out, people were worried that id’s current focus on consoles might lead to the incorporation of things like regenerating health. But instead they went with a system that rewards you for recklessly rushing at oncoming demons, to reap the health rewards that close-up kills give you.

Anyway, that’s how health kits can affect player movement in combat. But they do more than that. They also encourage player exploration. James Paul Gee, an education theorist who has studied games a bunch, looking for lessons on how they motivate players, includes a really wonderful passage in one of his books, where he describes, in first-person, the experience of a certain moment in American McGee’s Alice:

[James Paul Gee quote]

This passage really gets to the heart of the negative effect regenerating health has on level design. You need your players to need something. If they don’t need anything, then they don’t have any motivation to go anywhere. You take away their basic drive to ask, “what’s over there?” And you end up replacing that sense of curiosity, that direction through a space motivated by reward, with a bunch of dictatorial commands. And I guess this works, in military shooters like Call of Duty, because realistically you’d always be following the direction of your commanding officer, anyway. But it also leads to the ugly excesses of voices on the radio.

In Half-Life 2, Valve trains players early on that a specific type of crate—ones that look like this—will have useful items in this. There’s actually a bunch of interesting things going on under the hood with how these crates function, including the fact that their contents actually change based on where the player’s health is, and where their shield energy is, and what weapons they’ve been using recently. But the point I want to make is really simple: they’re the carrot dangled in front of us. Or, rather, they’re an array of carrots, forming a breadcrumb trail. A carrot breadcrumb trail. Crumbs of carrot cake? Whatever, this analogy is getting too belabored. But the basic point is, they are something that the player wants and needs. And they allow Valve to pull players toward the relevant corners of levels, rather than push them. The motivation we have for moving toward a health pack is our own. It’s not something that’s hoisted upon us by a chattering NPC.

3.6 Pulled into a story

One final thought before I end this video. At the time it came out, Half-Life was widely regarded as a revolution in game storytelling. Which is a bit counter-intuitive, when you really get down to it. Because Half-Life doesn’t tell that much of a story. We can go through its basic beats right here:

• Young theoretical physics postdoc Gordon Freeman is at the scene when an experimental scientific procedure goes awry.
• Said procedure opens up a dimensional rift, allowing aliens from another dimension to teleport into the vast facility where he works.
• Initial hopes are pinned on a military rescue, but then it’s revealed that the military has arrived to silence the scientists by killing all witnesses, rather than save them.
• As he struggles to survive a multi-front conflict, our protagonist discovers that this is not humanity’s first encounter with these aliens. In other parts of the lab he didn’t have clearance to, separate teams had been abducting these creatures from their own home and performing experiments on them.
• Finally, Freeman reaches the Lambda complex, which characters had previously told him might hold the answers. He discovers that the lab had long been doing experiments in teleportation, which landed them in the border dimension of Xen. It was these experiments that provided the initial contact between humans and the aliens.
• In order to close the dimensional rift, Freeman must travel over to Xen and defeat the alien who is commanding the forces of the alien dimension. He does so. Upon success, he is offered a job from the mysterious trans-dimensional authority known as the G-Man.

That’s definitely a story. It’s not a particularly revolutionary one. It borrows a lot from the output of id Software at the time—both DOOM and Quake were about monsters invading in the wake of scientific experimentation with teleportation technology. In terms of complexity of themes and character development and overall emotional palette, it can’t hold a candle to what RPGs were doing around the same time, in both Japan and in the West.

So it’s important to make a distinction. I don’t think that people really laud the Half-Life franchise for its story. I mean, its story is fine. It’s not bad, or anything. But I think that people mostly laud it for its storytelling. And that’s because Valve’s storytelling is, like everything else in these games, based around the philosophy of “pulling” players.

Sure, there are dialogue scenes, and we can’t skip those. So I suppose that’s a demerit on Valve’s part.

But aside from those, player engagement with the story of these games is entirely guided by player curiosity. It is entirely possible to run through these levels without caring one whit about the story that’s embedded in them. But if you are interested, there are all sorts of details to notice. The best examples of these are the various appearances of the G-Man, around the edges of various levels, always keeping an eye on Freeman’s progress. It’s completely ignorable, but if you’re interested, it’s there: another little carrot-cake crumb trail for you to follow. (Wow, now I’m weirdly attached to that analogy.)

I’ll be going into more detail about Valve’s approach to storytelling in the next video. That’s all for this one. Thanks for watching!