Setting sail in Bossa’s mindblowing Worlds Adrift, a fully physical and persistent world that’s poised to change the MMOG’s tack forever
Scouring its official YouTube channel, it’s hard to grasp what Worlds Adrift is, never mind what it might represent. During our first hour at Bossa Studios, our interest is aroused, but a flying-camera tour around Adrift’s shattered, sky-bound islands fails to epitomise the titanic undertaking, in conjunction with tech startup Improbable, that Bossa is reluctant to label an MMOG. The innovations aren’t subtle, but rather so enormous as to demand a shift in perspective before their implications can be appreciated. As co-founder Henrique Olifiers and game designer Luke Williams mess about in the clouds, excitement starts to burn as if seeing Minecraft for the first time, and suddenly it seems a monstrous error of judgement that Bossa hasn’t shouted and screamed about these features in its marketing. But the team is worried. Worlds Adrift is such a dramatic departure from Bossa’s previous games, including Surgeon Simulator and I Am Bread, that anyone not watching live might reasonably call it faked.
Worlds Adrift is a vast, persistent physics simulation of a like never seen in singleplayer games, let alone an online world, and the technology it’s built on threatens to change not just how we play but how every online game is developed.
Procedurally generated islands lace a billowing cloudscape, and from these – in line with many games looking to cash in on the construction and crafting boom – players can gather resources and process them into items, building airships to carry them onwards to new lands and opportunities. To begin with, it’s wonderfully serene. Worlds Adrift can host any number of players, pulling new islands from a queue to maintain a balance of isolation, cooperation and conflict. Olifiers hopes you’ll encounter other crews with the same frequency as paths cross in Journey – a clear influence alongside Skies Of Arcadia and The Legend Of Zelda: The Wind Waker.
“The setting itself is common enough, isn’t it?” Williams says. “But the argument I use is Halo: people are always like, ‘Yeah, you’re on this huge halo!’ Well, no. The halo is a skybox and I’m just on terrain. A lot of these games use floating islands as a setting – they don’t make full use of it.”
Bossa’s commitment to realising the broken remnants of a planet plundered for a levitating mineral has called for technical ingenuity. Clouds are not textures but volumetric shaders with strategic value, able to conceal fleeing ships as they manoeuvre through the fog. For this to work, each cloud must be identical for every player. Achieving that synchronisation meant two months of work, but Bossa has been approached by two companies hoping to license the technology for the likes of flight simulators.
Worlds Adrift is littered with history of the people who used to live here (several novels’ worth, apparently) and by photographing and documenting your travels, you gain knowledge of new schematics to improve your engines, your cannons or perhaps your hull, the better to penetrate Adrift’s storm walls, which serve as progression gates. There’s no levelling, only the pursuit of new tech, knowledge and the EVE-like freedom to decide your own objectives based on what you know. Here, Worlds Adrift steps back in time to an age when insight into an online world was shared by players and not walkthroughs.
“These storm walls change in thickness throughout,” Williams tells us, “so if you find where one’s particularly thin, that’s valuable information that’s not easily transferred over a wiki or something. You’re drawing your own maps, Phantom Hourglass style, and your place in the world is completely relative to what you’ve drawn. The players are generating valuable knowledge that’s actually very local.”
Each player’s unique perspective on the world is the first hint that Worlds Adrift is more than an airship-enabled survival sim. This knowledge, recorded on a ship’s mapping table, is permanent, existing until modified by a player or Bossa shuts down Worlds Adrift. It can be shared, sold, stolen or defaced, and the same is true of every last object: each tree, each branch you fell, each cannonball fired, and each wreck you abandon will lie there – identical to every observer – until disturbed.
“While we have the backstory of the world and all the history the players will be discovering,” Williams says, “the presentday [game]world doesn’t have anything unless the players put it in. So the stories of the crashed vessels and the camps that you find – they’re all player placed. If this ship were to crash and all these parts were to fly off or be part of a wreckage, you’d be able to find that and go through it all and maybe find their map and all the notes of what they’ve done; you’ll find pictures of where they’ve been, and so there’s this strange storytelling that’s real history.”
Stranger still, the first thing you do in Worlds Adrift is relearn how to move – and how to move other things. WASD still works, of course, but walking isn’t your go-to method of traversal, for even Spider-Man would envy the freedom of 3D motion offered by your grappling hook. The rope, linking player to chosen pivot point, is a physical object: swing past a tree and it will catch in the branches, changing your course and momentum. Or, if you like, you could stand on the ground, weaving your rope around many trees to create a cat’s cradle. You don’t have to – there’s almost certainly no point in it, and Bossa didn’t code it so that you’d be able to do so. You just can, because all of Worlds Adrift’s objects are physical entities. If it exists, you can interact with it as far as the laws of physics permit.
Williams tries and fails to pull an engine up a hill, because it’s heavier than he is. With producer Herb Liu’s assistance, it begins to inch forward. The application of two anti-grav gizmos have it floating like a helium balloon. We then watch as Olifiers chops down a tree as he pleases, shearing a branch and then the lower trunk, toppling it with a tug of his grapple.
“If you roll one of these logs down a hill, you’re going to kill the player,” he says. “In the same way, if you take one of these logs and rope it to your ship, you can use it as a pendulum to hit another ship. You can do whatever you want with these things; they’re physical objects. They serve as harvesting [resources], fine, but that’s not where it ends; that’s where it starts.” Liu obligingly tries to crush himself with the engine.
‘MMOG’ is a poor label for Worlds Adrift, the concept of which is as different from World Of Warcraft as Super Mario 64 is to Zork. Entrenched ideas of levelling and questing break down when you realise a dreadnought can be scuppered by flying above it and raining junk on the pilot (although the prudent may have built one or more backup helms below deck).
“It is literally impossible to balance because of one small aspect,” Olifiers states. “Because it’s all physical, there is no established strategy to do anything. There is no such thing as the best way of engaging in a battle or doing anything. There’s no best way of chopping down a tree. It’s a selfbalancing system, because someone will find a way to do things and someone else will find a way to top it infinitely.”
Worlds Adrift is for the discerning sky pirate: if you want to see the enemy smashed for no other reason than your own satisfaction, you’ll want to disable their antigrav cores; or maybe you want their ship for your own, in which case you’ll need to leave it unharmed as you take out their respawners. The capricious could enter from beneath, leave a photograph as evidence of their presence and vanish into the night.
“It’s not traditional MMOG looting – it’s actually looting,” Williams says as he cuts an engine free. “I have detached it from the ship, strapped some stuff to it so it wouldn’t fall, and I’ve now escaped with it – I could take it back to camp or maybe use the ship’s harpoon and drag it with me.”
A lifelong videogame player accustomed to idiosyncratic reality-fudging will find that this physical world is not altogether intuitive. You have to learn how to problem-solve again. For example, if you needed to punch through a hull with tough armour plating, the normal solution might be to acquire a ‘Superior Cannon of Piercing’. In Worlds Adrift, a heavier cannonball will serve just as well. It’s a shift from game logic to real logic.
In addition, Worlds Adrift promises an ecosystem of the sort that MMOGs have claimed to be implementing for years, in which animals and plants boast complex, interdependent AI and lifecycles that go far beyond spawn, fight, die.
“They don’t spawn,” Olifiers tell us. “If a creature is in a mating season, it has to find a gender-specific creature compatible with it. In the same way, they occupy an island, and if the player is just arriving to that island for the first time, there will be a balance there; but if you start to chop down all the trees and gather everything, the creatures will get hungry and be much more aggressive towards the player. The bottom line is: players could wipe out life from an island. We have systems to reintroduce life later on, such as flying creatures that migrate from island to island, or they take seeds with them or they poop something and it grows back.”
This dynamic paradise is done justice by its prodigious draw distance. We’re far out into the sky and Williams indicates a moving pixel in the distance. Zooming in with a camera (telescopes are in development) reveals a fully rendered Liu jumping and swinging from an island that looks for all the world like it’s metres away.
We have an appointment with Herman Narula, CEO of Improbable, whose technology underpins Worlds Adrift’s feats of physics and permanence. Olifiers, having spent the morning in a heightened state of excitement, is now beaming at having convinced his audience that what they’re seeing is real despite reading like science fiction. The man we’re due to meet makes him seem dour by comparison. Narula flits breathlessly between adoration for the latest build (which he often breaks off to try), an impromptu lecture on the science behind it, and planning new features with Olifiers. “I wonder if we can do water?” they mutter. “There’s an optimisation or two we could do to actually get it to work.”
With drive like this, it’s little wonder that Improbable talked $20 million out of venture capital firm Andreessen Horowitz. Early in developing its tech, Improbable reached out to developers brave enough to play guinea pig, and discovered everything it hoped for in Bossa and Worlds Adrift.
“It’s the greatest metaphor for player freedom,” Narula says. “It’s explicitly in its bones: soaring through the clouds on wings. It is the absolute opposite of the constraints and boundaries people normally associate with online games. The notion of a world where every object you encounter will be interactable will be the best kind of statement for Improbable and our tech, which is about making everything come to life.”
“We are very disappointed by the evolution of online games in general,” Olifiers adds. “Where every single other area of gameplay and game design has been evolving a lot in the past ten years, MMOGs are pretty much the same. Can you imagine players who grew up playing Minecraft, where they have the freedom to do whatever the hell they want… It’s impossible for a player like that to be put into an amusement park.”
The solution, as Narula describes it, is a rethink from first principles about how an online world is ruled by its servers. In a traditional MMOG, any given region of space is ruled by a server, handling everything from player position to chat. Clients act and the server reacts, advancing time by one frame and pushing the result back to everyone. Lag and instability develop where there are so many clients connected to the server that it can’t process all the inputs before advancing time. EVE copes by slowing time; other games ensure what the server has to communicate is incredibly simple, allowing for mere scraps of logic unrelated to the players.
“We don’t do this. This is bad. This isn’t the best way to build a modern distributed system,” Narula says. “Imagine if we presented to the developer a totally abstract world with no game loop, no time step – everything happens simultaneously. All entities really are is a little bit of state – like their health, their position or their rotation – and some behaviour. At its core, Improbable’s system understands this abstract representation of entities and it cheats. Behind the scenes, we’re able to run this on lots of machines because these entities are fundamentally individual processes.
“But actually this isn’t that great. It’s just this big abstract world. There’s no physics there – there’s no awesome game engine doing navigation and other stuff. And those game engines, they need game loops, so what we do is we have a separate layer and instead of one mammoth, intelligent server engine that understands the whole world, we treat engines like Unity as dumb little workers. We make them authoritative over a subset of these entities in the world. So the entity says, ‘Hey! I got rotation; I got position – I want a physics engine to do my physics!’”
The largest deployments tested by Improbable incorporated 200 servers with 700 instances of Unity running across them managing a world 160 square kilometres – an area the size of Israel. Using bots, Bossa launched a 300-ship battle involving 1,000 crew, and the game rumbled on unscathed. At this point, when server technology is no longer a limiting factor, design considerations take over. Thousands of cannonballs flying from all directions is unplayable chaos that would kill client-side hardware long before the servers started sweating. Now, designers are tasked with creating organic worlds that bring people together in a manner and volume suited to the intended tone.
The results of the division of labour are more profound than the ability to include true physics in an online game. Since the assignment of work is in flux, engines popping up and going down as needed, hosting the world is cheaper and more stable than the single-server equivalent. The failure of one engine can’t scupper the game, which Narula demonstrates for us through a clean and legible browser interface: he kills an engine responsible for a collection of items scattered about a visual representation of Worlds Adrift’s islands, and its work is instantly distributed between those that remain. At Bossa, the process is attended by one man in place of a dev ops team.
The ‘how’ of the distribution is hidden from the designer – that’s Improbable’s job. The programmer codes behaviour in Unity, using the Improbable API, and lets it fly – a new, live feature that didn’t require downtime. Worlds Adrift’s hang glider was implemented after a single day’s game jamming. More significant still, Improbable does away with the need for the multi-year development of networking code for a game to sit on: any game built with Improbable is inherently multiplayer, exposing the costliest genre to indie developers. Additional engines will be integrated with the Improbable tech as it develops.
Improbable and Bossa, acutely aware of the value in their offering, are employing shock and awe tactics, which, if successful, could cement their place at the forefront of online games for the decade to come. As Improbable’s lab rat, Bossa has had to solve a great many problems of its own, such as the global physics in Worlds Adrift, contributing the results back to Improbable’s code libraries. Dean Hall’s mysterious work in-progress, Ion, also uses Improbable and already incorporates code written by Bossa. However, the enormity of their joint ambitions is made clear in the revelation that Worlds Adrift will be forkable in its entirety: any new developer can press a button and generate a fresh copy of Worlds Adrift, running on Improbable’s servers, to edit however they please. Anything Bossa can do, you can do.
“To be clear here,” Narula says, “the modders or the forkers of the game will be able to support deployments of big sizes, just like Bossa can. It’s a service that they’re also able to use. If their version of the game is more popular, then people will be on that! And Bossa have a deal which means they’ll do well too. The cost will be proportional to the scale of the world, which is happening organically depending on how successful their game is.”
Narula presents his approach to online gaming with such elegance as to make it seem impossible that this has never been attempted. It’s easy to forget that it has taken the combined brainpower of former members of Google, Amazon, Goldman Sachs and Cambridge University to develop the technology, a feat he didn’t see happening within the game industry itself.
“The reality is that the infrastructure we had to build didn’t exist, and a lot of stuff we’ve had to pretty much invent ourselves,” Narula tells us. “The game industry wouldn’t have created an investment hole that would have attracted the non-gaming technical talent necessary to make it happen, because the problem sits between gaming and outside of gaming.”
Improbable’s approach to distributed systems is a solution to a general problem: many industries, from healthcare to utilities, need realtime, interactive simulations to run across multiple networks. Improbable splits work that was previously indivisible. Given demand and obvious investor interest, it begs the question of why Improbable is working with games at all. In collaborating with Bossa, Improbable hopes to showcase its invention in settings that provoke wonder – tactile game worlds that convey the potential of a new technology in ways boardroom PowerPoints can’t. Worlds Adrift, which allows you to soar with friends in a ship of your collective design, sifting wreckage, and making history, looks set to do just that.