Thinking spatially: persistent worlds


Three robots with big space on the left on grey background

At Improbable, we think SpatialOS can make a huge difference to games in three primary ways by enabling game designers to create games with seamless scale, rich simulations, and meaningful persistence. This post will look at meaningful persistence in the context of persistent game worlds.

People game for many reasons; socialising, exploring, fighting, and achieving, as the Bartle taxonomy has it. In one of its earlier forms, ‘achievement’ could be seen in the battle of gamers to get onto the high score table of the local arcade machine. Showing you were king of the sticks could only happen if game developers created systems that delivered a form of data persistence – a record between rounds (or instances) of who had collected the most points.

Think of how classics like Space Invaders or Pac-Man keep you coming back for more – with the chance to take pride of place at the top of the high score table. That table was persistent – as long as the game stayed plugged in- and changed over time to track the achievements of local players. But as games became more complex, and technologies like the Internet became increasingly available to game developers and players, the potential to offer varied and more interesting kinds of persistence grew.


The MUDs (multi-user dungeons) of the early internet allowed multiple users to be hosted in an interactive “world” that could persist for a limited amount of time. MUD1 (1978) by Roy Trubshaw and Richard Bartle (who later designed the Bartle taxonomy for MUDs like this) was a text-based game that first exhibited this basic kind of world persistence. Multiple users could “log in” to the game world before it would “reset” due to memory limits.

"At Improbable we face the challenge of designing and improving a technology (SpatialOS) that enables games to maintain and constantly update vast quantities of spatial data. But we’re also aware that developers have their own challenge – to make unmissable games."

While this created an engaging paradigm, the loss of progress between sessions placed a limit on player achievement. As the game world would soon lapse into non-existence and no record was kept of what you had done, the experience of playing MUD1 was limited too. What was the point in investing a lot of time into a game if, unlike even the arcade, nothing you did was carried over? Whatever the answer, persistence in games had started in two forms; world and data.

Multiplayer Persistence

It’s tricky to pin down why we like to play games with other people. It’s probably related to the possibilities for new experiences that happen when unpredictable human interactions interact with the rules and logic of a programmed environment.

A screenshot from a text-based multi user dungeon

Equally, it’s probably related to the human need for social connection – it adds meaning to us on a psychological (and for some, spiritual) level if our gameplay is shared, whether through multiplayer game design, competitive esports or live streaming.

Whatever the reason, higher orders of persistence require a game world that is stable and large enough to facilitate complex encounters and interactions with other players, which enriches the experience of gaming.

What does a persistent game look like?

World of Warcraft (2004) is the most famous example of data and world persistence being combined to make a phenomenally successful game, building on the examples of Ultima Online and Everquest in particular. These games boast persistent data features like a character inventory where crucial items could be kept between play sessions. Accidentally drop a key that you need for a door in a quest and the game will give you a way of reclaiming it, either by preserving it where it was dropped or respawning it in its original location. Additionally, players can level up and grow their characters – deeper progression than had previously been on offer in role-playing games.

Players take down a foe in Everquest

In terms of world persistence, these massively multiplayer online role-playing games assigned players to one of many constantly available worlds running on its own server or servers (in World of Warcraft these were called “realms”). Logging in would allow you to join a small group of other players in hub areas, with your progress carrying over from the last time you were last logged in.

The world itself even claimed to remember you. NPCs you’d met before would “recognise” you in the dialogue of future encounters, giving the sense that time was moving forward and that you had a role to play in a complex society.

However, this kind of persistence had its limits, which placed limits on the ability of game designers to make their worlds feel real too.

Enter the theme park

Unlike in the real world, which is truly persistent, you can’t permanently affect the fabric of this kind of game. You can’t knock down a building or permanently kill an NPC because other players have to be able to experience that “content” too. If a boss existed just once, it would be swarmed, each player seeking to defeat it the one and only time it appeared. Instead, multiple groups fight multiple duplicates of the same enemy in many instances, and often go back to fight the same battles again and again in search of experience or better loot.

Ultima Online screenshot

Additionally, player characters need access to replenishing resources. This leads to humorous world-breaking phenomena like the cows in RuneScape that players kill for leather and meat. So urgently are they needed by the player base that they literally respawn before your eyes, so that other players can cut them down again. It seems that if a game world is to be engagingly persistent, it needs to be “real” in the sense of “believable”.

However, there are significant barriers to a shared world where changes to the state of the world are permanent. Imagine the content demands on game developers building worlds where no action recurred. Instancing – the generation of unique instances of each boss fight or set piece so that players can tackle them when ready – is a way to make the same content enjoyable for many players many times over. Again, this has a disadvantage. Players end up experiencing so many of the key events of an MMO alone or in small raiding parties that they may realise just how little of a massively multiplayer games is actually massively multiplayer.

To simulate deeper world persistence and to foster a greater sense of player interaction and involvement, some MMOs use “world events” – occasions where whole areas of the game are changed by a cataclysmic event apparently brought on by the actions of the players. In reality, nothing has happened in-world to alter its state. Instead that part of the game has been reuploaded by the developers in a new patch, the old version switched off and the new one switched on. This is scripted persistence and it’s arguably the best these games have been able to do to make players feel like they are in a living, breathing world.

World of Warcraft screenshot

The familiar metaphor is that the modern MMO is a theme park. Players encounter each other in the concourses and queuing for the rides, but fundamentally experience every roller coaster alone, one at a time. And the ride always resets for the next rider.

This is not fundamentally a bad thing – obviously this kind of experience is still deeply engaging, as the playing hours that devotees of World of Warcraft, or RuneScape, or EVE Online will testify. But, at Improbable we would argue that a new era for game persistence and persistent virtual worlds is on the horizon. And that it will be made possible by distributed computing technologies like SpatialOS.

Would a totally (or at least very) persistent game be fun to develop and to play?

Imagine the following game world scenario:

You are fighting with other players on the blue team for control of a multi-storey building. The building is owned by the red team and is giving them a huge tactical advantage in this suburban corner of the game world – a giant ruined city that stretches for miles.

Your squad is under heavy fire from the building, but you make a flanking push across open parkland and cut down a foolish opposition player carrying C4 explosives. A teammate uses local voice chat to shout in your ear.

“You hero!” he says. You’re surprised – this is by far the nicest thing you’ve been called online in ages. “Get that C4! We can take down their block!”

In this game, you remember, C4 is a limited resource. Given the fragile buildings, there’s only a small amount of it in the entire game world. If it’s dropped, it stays where is and doesn’t despawn.

The other player calls in reinforcements, who suppress the enemy in the block whilst your squad sneaks up close and acquires the C4.

If you can hold on to it and get close enough, that C4 will take down the block with everyone inside. Given enough time and enough C4, the entire map could become ruins…

Green soldier

It’s easy to imagine examples like this of what a meaningfully persistent game could look, feel and play like. The advantages for players should be obvious too. For instance, the high levels of data persistence achievable by new technologies like SpatialOS mean that games can keep track and run an economy with logistics, finite resources and scarcity. The decisions that players need to make, like what equipment and resources to take on a raid, begin to matter more. Consequently, player achievements carry more weight because the long-term impact of success or failure carries on past a single gaming session.

"The power of meaningful persistence will be a key differentiator in the next generation of games."

The multiplayer experience will be richer too. It will take a greater degree of cooperation to complete a short-term objective, like capturing a key position, in the knowledge that it will need to be held. Long-term objectives will require planning and coordinating, bonding players like never before. A game with this much world persistence can fulfil the promise of the MMO and create highly engaged players who are heavily invested in the world of your game.

Attempts have been made to create this kind of world, with its heightened stakes, without distributed computing. The unique Planetside series of games has managed similar mechanics, but with careful compromises over object persistence and player numbers, and without resource-eating but gameplay-enhancing elements like destructibility, object physics and longer-term world persistence. By distributing the world across many servers in the cloud, SpatialOS, our distributed platform for game development, removes the need for a lot of compromising.

Planetside 2 battle screenshot

Meaningful persistence benefits studios in other ways also. Instead of constantly working on new content, developers can focus on creating engaging, rich systems for players to interact with, creating their own emergent fun and meaning.

Take a look at Worlds Adrift by Bossa Studios, for example. This game is made with Unity and powered by Improbable’s SpatialOS to achieve an enormous persistent game world. Content creation is largely user-driven via a free tool called Worlds Adrift Island Creator. This process grows the world, engages the community, and saves on development time and expense. This enables a smaller indie studio to build and operate a massively multiplayer RPG.

MMOs are almost universally challenged by players’ ability to burn through new content quickly – often more quickly than anticipated. The emergent stories created by Worlds Adrift’s persistent sandbox, coupled with Bossa’s crew and guild systems, mean that players create a lot of fun for themselves. Part of the joy of running a SpatialOS game, it seems, is the joy that developers themselves feel in seeing unexpected things happening in the games they create. For instance, imagine the emergent joy at seeing a member of your crew accidentally latch onto an intelligent AI creature that then flies away at rapid speed…

Meaningful persistence is a new (but rewarding) challenge

At Improbable we face the challenge of designing and improving a technology that enables games to maintain and constantly update vast quantities of spatial data. But we’re also aware that developers have their own challenge – to make unmissable games. We love seeing the challenges they encounter along the way. For instance, distant parts of the game world in Worlds Adrift can only be reached if players team up to build powerful skyships. Since these ships are co-owned, how persistent should they be? If you designed them to persist in the world forever, including when all their crew members had logged out, how would you stop them being stolen by other players? More to the point the point – should you? It’s an interesting thought, but perhaps the most engaging (note – not necessarily the most “fun”) games are the ones where consequences are tangible, emotional and perhaps a little painful.

As it happens, Bossa chose a middle ground solution for this particular problem by using a game mechanic called the “personal respawner”. This is a portable respawn point for your character, should you die in combat or fall overboard. Placing your personal respawner on a ship is tantamount to shared ownership or at least crew membership. Bossa decided that a ship should respawn when anyone who has a personal respawner on it logs back in. This prevents ship piracy by total strangers – if all of the crew are logged out, so is the ship. However, a mutinous crew member could log back in and destroy the personal respawners of the rest of the crew while they sleep…

Clearly, captains need to think very carefully about who they allow to place respawners on their ship! You can still get burned by unpredictable player interactions created by meaningful persistence in Worlds Adrift. Doubtless this risk/reward dynamic is part of the fun. Imagine the emergent story lines and genuine quests that come from this kind of persistence.

"Meaningful persistence benefits studios…instead of constantly working on new content, developers can focus on creating engaging, rich systems for players to interact with, creating their own emergent fun and meaning."

You could go on listing the challenges too. How do you prevent next-level griefing? How do you integrate the complex VOIP chat systems that enable the kinds of complex teamwork these games require? With the persistence of the game world essentially generating all your content for you, how do you ensure things stay fun and balanced? Some of these questions are not unique to highly persistent games, like the ones SpatialOS enables, but many are. The point in listing them is that they excite us. Improbable and our platform developers are actively working on innovative solutions to this new frontier for gaming. But it’s not just about solving problems – it’s about making great games. The power of meaningful persistence will be a key differentiator in the next generation.

We hope you find “thinking spatially” as compelling as we do. What kind of hyper persistent game world are you going to create?

More solutions. More possibilities.

Explore all IMS solutions