this space intentionally left blank

November 10, 2018

Filed under: gaming»design»aaa

Loot Pack

If you wanted to look at the general direction of AAA game development, you could do worse than God of War and Horizon: Zero Dawn (concidentally, of course, the last two titles I finished on my usually-neglected PS4). They're both big-budget tentpole releases, with all the usual caveats that come with it: graphically rich and ridiculously detailed, including high-priced voice/acting talent, but not particularly innovative in terms of gameplay. But even within this space, it's interesting to see the ways they diverge — and the maybe-depressing tricks they share.

Of the two, Horizon (or, as Belle dubbed it for some reason, Halo: Dark Thirty) is the better game. In many ways, it's built on a simplified version of the A-life principles that powered Stalker and its sequels: creating interesting encounters by placing varied opponents in open, complex environments. The landscape is gorgeous and immense, with procedurally-generated vegetation and wildlife across a wide variety of terrain with a full day-night cycle. It's pretty and dynamic enough that you don't tend to notice how none of the robotic creatures you're fighting really pay attention to each other apart from warning about your presence — you can brainwash the odd critter into fighting on your side with a special skill, but otherwise almost everything on the map is gunning for you and you alone.

In fact, Horizon's reliance on procedural generation and systems is both its strong suit and its weak point. It's hard to imagine hand-crafting a game this big (Breath of the Wild notwithstanding), but there's a huge gulf between filling a landscape according to a set of gameplay rules and depicting realistic human behavior. One-on-one conversations and the camera work around them are shockingly clumsy compared to the actual (pretty good!) voice work or the canned animation sections. Most of the time, during these scenes, I was just pressing a button to get back to mutilating robot dinosaurs. But you can see where the money went in Horizon: lots for well-rendered undergrowth, not so much for staying out of the uncanny valley.

By contrast, God of War is really interested in its characters, as close-up as possible. The actual game is not good — the combat is cramped and difficult to read, ironically because of its love affair with a single-take camera (which is kind of a weirdly pointless gimmick in a video game — almost every FPS since Half-Life is already a single-take shot). It's small and short by comparison to open-world games, with its level hand-crafted around a linear story. I was shocked at how quickly it wrapped up.

But there's no awkward "crouching-animation followed by two-shot conversation tree" here: it may be a six-hour storyline, but it's beautifully motion-captured and animated. When Jeremy Davies is on-screen as Baldur, it's recognizably Jeremy Davies — not just in the facial resemblance, but in the way he moves and the little ticks he throws in. There's maybe five characters to speak of in the whole thing, but they come through as real performances (Sindri, the dwarven blacksmith with severe neat-freak tendencies, is one of my favorites). In retrospect, I may wish I had just watched a story supercut on YouTube, but there's no doubt that it's an expensive, expressive production.

Where both games share mechanics, unfortunately, is a common trick in AAA design these days: crafting and loot systems. Combat yields random, color-coded rewards similar to an MMO, and those rewards are then used alongside some sort of currency to unlock features, skills, or equipment. It extends the gameplay by putting your progress behind a certain number of hours grinding through the combat loops, as this is cheaper than actually creating new content at the level of richness and detail that HD games demand.

If your combat is boring (God of War), this begins to feel like punishment, especially if it's not particularly well-signposted that some enemies are just beyond your reach early in the game. It bothered me less in Horizon, where I actually enjoyed the core mechanical loop, but even there playability suffered: the most interesting parts of the game involve using a full set of equipment to manipulate encounters (or recover when they go wrong), but most of that toolkit is locked behind the crafting system to start. Instead of giving players more options and asking them to develop a versatile skillset from the start, it's just overwhelmingly lethal to them for the first third of its overall length (a common problem — it's tough to create a good skill gate when so much of the game is randomized).

Ironically, while AAA games have gone toward opaque, grind-heavy loot systems, indies these days have swung more toward roguelikes and Metroidvanias: intentionally lethal designs that marry a high skill ceiling with a very clear unlock progression. It may be a far cry from Nintendo's meticulous four-step teaching structure, but since indie developers aren't occupied with filling endless square miles of hand-crafted landscape, they've sidestepped the loot drop trap. Will the big titles learn from that, or from the "loot-lite" system that underlies Breath of the Wild's breakable weapons? I hope so. But the economics of HD assets seem hard to argue with, barring some kind of deeply disruptive new trend.

December 13, 2016

Filed under: gaming»design

Guns and glory

If you'd told me a few years ago that my favorite shooters in 2016 would be reboots of Doom and Wolf3D, I'd probably have been surprised, or depressed, or both. Surprised, because both of them were very much games of their time, and it would seem impossible to recreate their peculiar arcade-oriented chemistry now. Depressed, because I probably would have seen it as a part of the stagnation of the shooter genre, with which I have a love-hate relationship.

But it's true: after I upgraded my graphics card as a birthday gift to myself, I've been going back and replaying a bunch of FPS games (for better or worse, they're the graphical showcases in my Steam library), and the two standouts have been DOOM and Wolfenstein: The New Order. I'm as shocked as anyone! They're both excellent revivals of old id Software franchises, and part of what's so interesting about them is that they're excellent in such completely different ways.

Of the two, The New Order (and its prequel mini-expansion, The Old Blood), have a heavier lift: although it's vaguely connected to the 2001/2009 games, most people are only aware of the original, which was (despite mind-blowing graphics for its time) two-dimensional in both gameplay and character. They weren't great games even back then, and they haven't aged particularly well (TNO includes "nightmare" remakes of the 286 levels, in case you forgot how tedious and confusing Wolf3D could be).

The team behind TNO is the same group that made Chronicles of Riddick and The Darkness at Starbreeze, both of which took ridiculous licensed properties and stretched them past both the source material and the Steam category: Riddick asks players to engineer multiple prison-breaks with not much more than a knife, a black-market night-vision mode, and a lot of Vin Diesel dialog. Half of The Darkness is shooting light bulbs! The shape and flow of a Starbreeze game could be odd — linear chunks connecting free-form adventure hubs — but they were almost always as interesting in the downtime as they were in the action sequences.

By virtue (such as it is) of this being a Wolfenstein game, going too far outside of "open door, shoot Nazis, repeat" was probably too much to ask. But The New Order does pack a surprising amount of pathos into what is otherwise a totally bonkers 1960's alternate history in which the Fourth Reich won the war via mad science. It tilts on the action side of things, but there's still definitely a Starbreeze flavor, if you liked their other titles. Parts of it are brilliantly cinematic, including a ten-year flash-forward sequence that separates the first and second acts.

And while the gameplay doesn't go full retro, it has elements of old-school flavor. There's no regenerating health system or trendy cover-hugging mechanics here, and the use button gets a workout. My favorite refinement is the commander system, where many of the battle setpieces will introduce one or two radio-equipped officers. If you're spotted, they'll call in reinforcements, so a good approach is to stealth-kill them before mopping up the troops. Or you can play the way I do: take out one officer and then charge the other with guns blazing, before they can stack the odds too far. That this is still a viable (and fun) strategy strikes a nice balance between players who want the stealth experience, and those (like me) who are mostly in it for the instant gratification.

DOOM has no such such balance, and does not suffer for it in the slightest. There is no clip reloading for any of the weapons, and the run speed is entirely unrealistic. Unlike most shooters since Halo, everything in DOOM is designed to encourage non-stop movement toward (and through) enemies. Which is a big part of the reason why, even though it changes from its predecessor in significant ways (mouselook, a jump key, lots of upgrades and collectibles), it still manages to feel like playing deathmatch in 1996.

The two primary mechanics for maintaining player momentum in the game are a fast mantle, which gives players the ability to very quickly ascend vertically, and the "glory kills," which reward players with a half second of invulnerability (during the animation) and a quick piñata pop of health and ammo. There's no reward for staying still — in fact, like a bullet-hell shooter, players are immediately punished for remaining stationary. The goal is not to block or absorb damage, it's to avoid getting hit at all.

DOOM levels are primarily structured as a series of loosely-connected arenas, which also keeps the deathmatch feel: while hallways and platforms are used to set the mood and create checkpoints, the most intense gameplay is set in wide, 3-D spaces with multiple "circuits," just like the best Quake DM levels. Gears of War is also famous for "combat bowls" as a gameplay tool, but it strongly emphasized cover over movement, whereas DOOM almost never places waist-high walls to serve as partial cover (and they'd be useless in a game with lots of high vantage points anyway).

There's a moment at the very beginning of the game where your character, having ripped out of a set of shackles and punched through the initial set of undead, pulls up a video screen reading "DEMONIC INVASION IN PROGRESS." As mission statements go, this is pretty much DOOM in a nutshell: crank everything up to 11 and embrace the inherent, b-movie absurdity of the thing (a similar process took place music direction, which started with no guitars at all and ended up a metal shredfest).

All combined, the end result is about as pure a video game as you can get with a high-budget, AAA studio product in 2016. It's the interactive equivalent of a Fast and Furious movie: mixing comforting aesthetics (as far as the intended audience is concerned, at least) with the maximum amount of intense parallax motion. Nobody's going to mistake it for fine art — it has none of the thoughtful playfulness of, say, Dishonored — but not everything should be fine art. It's a great game.

August 19, 2015

Filed under: gaming»design

Free Electron

When GitHub released Atom last year, there were a lot of people who thought the underlying technology — a web-based mashup of Chromium and NodeJS — was a terrible idea. They may yet be right, but Atom has gotten steadily better, to the point where I can recommend it to my students. Meanwhile, GitHub made the app runtime, Electron, available for hacking around, and it's becoming increasingly common (along with NW.js) as a way to build simple, network-savvy desktop applications.

While I'm hardly sympathetic to the people who complain about web-over-native, one complaint that does seem reasonable is the size of these applications. Electron takes up 113MB even before any user code gets loaded, and while that's not a huge footpint in today's world of terabyte hard drives, it may seem redundant when users probably already have a browser installed. If you run multiple Electron apps, they don't share runtimes, either.

That said, there is one genre of software that routinely ships its own dependencies this way, and nobody bats an eyelash: games. Looking at my Steam library, there's at least five titles there that run on Unreal Engine, each with its own copy of the engine and supporting libraries. And engines like Unreal or Unity are huge cross-platform monoliths these days, used by AAA studios and small teams alike. Many of these games probably embed a web browser anyway, since it's a great way to build UI. What if we flipped that around? Could Electron be a viable development platform for independent games?

I started thinking about this question earlier this year, after reading a post by Greg Tavares, one of the WebGL architects for Chrome. He points out that as platforms go, the browser is pretty good: features like GL are well-tested and reliable, JavaScript is a great rapid prototyping language, and cross-platform support is pretty decent. If your needs are relatively humble, or you're already well-versed in web development, why not use something like Electron, paired with Three.js or another framework as your "engine?"

I've been meaning to test out Electron anyway, so I spent some time tonight trying to answer that question. The result is You Don't Know Electron. It's a simple party-style trivia game, with a twist: instead of crowding around a keyboard or using controllers, players scan a QR code to use their phone as a "buzzer." Everything is in sync, so when the first person presses the correct answer, everyone else's device will switch to a "get ready for the next question" screen almost instantly.

Behind the scenes, the app opens up a native-looking window on the host computer for displaying questions and scores, but it also spawns a hybrid HTTP/WebSocket server for the mobile clients. The result is a little slapdash, but surprisingly responsive in practice. UI is all done through a super-simple Angular thin client, with most of the state living in the persistent Node server. And while the look of it isn't anything fancy, it's also only the result about about 3 hours of work — a little extra polish, and I legitimately think it could be a fun party game. Feel free to clone it and try for yourself.

If you know Node, Electron turns out to be an surprisingly pleasant development experience. I'm no indie game dev, but based on tonight's experiment, I'd seriously consider it if I wanted to start a new project. For casual games, WebGL (or even 2D canvas) and HTML are a solid foundation, especially since Electron's browser window is up-to-date with features like flexbox and web components. It's no Unity, but it is free and built on well-documented standards.

The fact that it's web-based under the hood opens up a whole range of exciting possibilities around little custom servers: phones or tablets could be used to show hidden information to each player (a hand of cards, personal inventory, secret messages, etc.), or additional computers could instantly turn a single player session into multiplayer. All with no installation friction: just connect to the host with a browser and start playing.

(In fact, I've had a long-standing idea for an asymmetric "virtual board game" of Alien vs. Nostromo crew that would work perfectly with this setup. If only I had the time...)

Meanwhile, we're seriously considering this at the Seattle Times as a way to package up our tools for newsroom users. We're not the first people to do it — NPR's Lunchbox bundles up some of their social tools — but since our workflow is almost entirely Node-based, we're in a much better place to take advantage of everything Electron can do (NPR basically just bundles the static output of their Flask apps). Don't be surprised if you see more Electron-based tooling news from us in the future!

December 10, 2013

Filed under: gaming»design

Policing Procedurals

Even if I'm sticking with Steam for most of my gaming, our new PS4 did get me interested in Warframe, the free-to-play shooter that's available on Playstation and PC. I don't normally care for free-to-play, if only because I feel guilty for never buying anything, but I liked the central conceit of Warframe: procedurally-generated levels and highly-mobile Mass Effect-style combat. In retrospect, I probably should have been more skeptical. To understand why, we have to look back at how shooters have been built over the last twenty years.

There was a time, way before Halo and before franchises like Battlefield ran the earth, when one of the main selling points of a first-person shooter was the quantity of unique weapons that it brought to the table--an actual arms race, peaking with Duke3D which (for all its flaws) had some clever joke guns to go with the ubiquitous pistol/shotgun/chaingun trio. I'm not saying this was a better time, or that they were better games, but there was definitely a sense that the genre was about creative destruction, in the same way that fighting games are about combo systems and special attacks.

Then id Software built a monster for competitive deathmatch: Quake and its successors had an incredibly bland set of weapons, because in "serious" multiplayer the goal is to streamline everything except moving and shooting. This was the second refinement of shooter design, and it focused on the levels themselves, but as topology instead of as setting. Players concentrated on learning the levels so that they could plot a path that would keep them supplied, while denying pickups to the other players. A good Quake or Unreal Tournament player knew the game's weapons and how to aim, but more importantly they knew where to go, and when. Navigation became the mark of quality for a deathmatch bot.

Since then, these tendencies have mellowed as the possibility of more complex interactions and narratives has become available. Environments are built more for realism and story, weapons are more traditional and not usually why you buy the game. Which brings us to Warframe, which has basically none of these things. There's hardly any story, the "levels" are randomly generated from a series of tilesets, and the weapons are part of the free-to-play grind: either buy a new gun with real money, or spend a lot of time crafting one within the game's economy. Unlike the Mass Effect and Gears of War titles it resembles, there's no explicit cover system, but players do have a much wider range of movement options than a typical shooter: there are slides, flips, and wall runs available through various key combos.

If Warframe's computer-generated levels were any good, this would be a different post. Good levels would give players a way to put their acrobatic movement skills to good use, rewarding people who have learned the parkour system and can improvise in response to new environments. But the random generator mostly builds long, boring hallways connecting wide-open, pre-designed rooms, none of which require any particular skill outside of strafing and taking cover behind walls. Since players can't learn the levels and their flow, they can't optimize or get better at moving through them. And since new weapons require an investment of serious cash or time, almost everyone's using the same rifle and the same melee weapon, which means you never see anything that makes you want to spend any money or time in the first place.

The irony of this problem is that someone already got the formula right for doing procedural FPS games, and they did it by almost exactly reversing Warframe's formula. Borderlands has hand-built levels and enemy placement, combined with randomized weapon generation: each gun consists of components assembled onto a set of base bodies, which vary by "manufacturer" with certain preset tendencies and aesthetics. For example, Mariwan guns always inflict status effects (poison, fire, etc), while Jacobs weapons are Western-themed and can often fire as fast as the player can mash the button. Within those simple parameters, however, the results when you pull the trigger can vary wildly.

The result is a game that combines the two old-school driving forces of FPS design — clever level design and weapon variety — with the collector's urge that powers massive multiplayer games (Borderlands even borrows the color-coded quality markings from World of Warcraft, making it easy to evaluate a weapon drop in an instant). The innovation is not proceduralism — games like Diablo have long offered that — but figuring out how to balance it with the formula for a replayable and rewarding shooter. As someone who almost totally lacks a collection instinct, but loves the classic FPS genre, Borderlands 2 hits the sweet spot with remarkably few missteps (it's also surprisingly smart and funny, which is a welcome change).

I don't think procedural generation is impossible for first-person games — indeed, I think it's likely to have a bright future, particularly as web games mature and optimize for delivery size — but it illustrates just how difficult the challenge is likely to be for anyone who attempts it. For all that people talk about the genre as if it's just a collection of bro-heavy manshooters, there is undeniably a huge amount of craft that goes into the fundamental mechanics. As Kevin Cloud notes in Dan Pinchbeck's analysis of Doom (now 20 years old!),

Every genre has its real strengths, but in a shooter... if running and shooting is not fun, doesn’t feel natural, doesn’t feel visceral and powerful, then I think you are going to lose out.
Movement in FPS games is not just about how the player transitions from point A to point B, but about all the obstacles and decisions that make up that route. As such, building procedural content is not impossible, but it needs to provide good options for cover, paths for moving between pickups, and unexpected chances to either ambush enemies or be ambushed. Warframe may do this one day, but right now it's failing miserably. In the process, it's showing how little the developers have really thought about how the game they're building fits into the traditions of its genre.

May 2, 2013

Filed under: gaming»design

POV

There's a common complaint about the Bioshock games, which is that they're not very good shooters. People writing about Bioshock Infinite tend to mention this, saying that the story is interesting and the writing is sharp but the actual game is poor. And this is true: it's not a very good first-person shooter, and it's arguably much worse than its predecessors. But this implication of most of these comments, from Kotaku's essay on its violence to Brainy Gamer's naming it the "apotheosis of FPS, is that Infinite is bad in many ways because it's a first-person shooter--that it's shackled to its point of view. In doing so, it has become a sort of stand-in for the whole genre, from Call of Duty to Halo.

I sympathize with the people who feel like the game's violence is incoherent (it is), and who are sick of the whole console-inspired manshooting genre. But I love shooters, and it bugs me a little to see them saddled with the burden of everything that's wrong with American media.

Set aside Infinite's themes and its apparent belief that the best superpower is the ability to literally generate plot holes--when we say that it's not a good FPS, what does that means? What is it, mechanically, that separates the two? I'm not a designer, but as a avid FPS player, there are basically three rules that Infinite breaks.

First of all, the enemy progression can't be just about "bigger lifebars." A good shooter increases difficulty by forcing players to change their patterns because they're not able to rely on the same rote strategy. Halo, for all its flaws, gets this right: few of its enemies are actually "tough," but each of them has a different method of avoiding damage, and a different weapon style. By throwing in different combinations, players are forced to change up their tactics for each encounter, or even at multiple points during the encounter. Almost all of Infinite's enemies, on the other hand, are the same walking tanks, with similar (dim-witted) behaviors and hitscan weaponry. I never had to change my approach, only the amount of ammo I used.

Along those lines, weapons need strengths and weaknesses. Each one should have a situation where they feel thrillingly powerful, as well as a larger set of situations where they're relatively useless. This doesn't have to conflict with a limited inventory--I loved Crysis 2's sniper rifle, spending the entire game sneaking between cover positions in stealth mode, but it was always paired with a strong close-in gun for when I was overrun. A good game forces you to change weapons for reasons other than "out of ammunition." Infinite's close-range weapons feel identical, and its sniper rifle is rarely useful, since a single shot alerts everyone to your position.

Finally, every fight cannot simply be about shooting. Most shooters are actually about navigating space and territory, and the shooting becomes a way of altering the priorities for movement. Do you take cover, or dodge in the open? Do you need more range, or need to close on an enemy? The original Bioshock made the interplay between the environment and your abilities one of its most compelling features: electrifying pools of water, setting fire to flammable objects, flinging scenery around with telekinesis. But at the very least, you need an objective from time to time with more complexity than "kill everything," both as a player and in terms of narrative.

Bioshock Infinite has, in all seriousness, no period I can remember when my objective was not reduced to "kill everything." Combined with a bland arsenal and blander enemies, this makes it a tedious game, but it also puts it at complete odds with its characters. The writing in Infinite is unusually good for a shooter, but it's hard not to notice that Elizabeth freaks out (rightfully) during one of Booker's murderous rampages, comes to a cheery acceptance with it a few minutes later, and then spends the rest of the game tossing helpful items to you under fire. That's writing that makes both the narrative and the mechanics worse, by drawing attention to the worst parts of both.

It's not the only shooter with those flaws--people just had higher expectations for it. The average FPS is badly written, and it's a favorite genre for warmongering propaganda pieces. But that's true of many games, and yet we don't see pieces talking about the "apotheosis of platformers," or talking about RTS as though they're emblematic of wider ills just because Starcraft II is kind of a mess. And there's still interesting stuff being done in the genre: Portal and Thirty Flights of Loving come to mind. To say that FPS have reached their limits, ironically, seems like a pretty limited perspective.

Past - Present