SpatialOS: The World’s Next Operating System
SpatialOS may well be the world’s next operating system, following a lineage that runs from MS-DOS through Windows, the Web through iOS and Android.
Developed by Improbable.io, the company has serious funding, a deep talent pool, and the kind of press that a company can only dream of when their product is barely in beta.
Build Big Big Games
SpatialOS is built on a relatively simple premise: leverage the power of scalable cloud computing to manage thousands (or millions) of ‘entities’. These entities can then be displayed through a front end platform like Unity or Unreal to create massive scale persistent experiences for hundreds or thousands of users.
Which means that the easiest way to think about SpatialOS is as a platform for MMOs or games. It removes many of the limitations to scale that bedevil truly persistent, dynamic game worlds:
- How do you manage more than a few hundred concurrent players without latency?
- How do you create a persistent world with dynamic objects that change over time?
- How do you deal with issues of scale? For example, how can a user walk across a country-sized game world without the ‘stutter’ that comes from sharded servers and concurrency?
SpatialOS deals with this by launching ‘workers’. Much like Amazon or Google can throw up a new server if the traffic to your website suddenly spikes, SpatialOS launches new servers based on the level of activities in your ‘game world’ (and does so with the backing of Google’s cloud infrastructure).
What’s unique about its approach is that it doesn’t do this based on overall network traffic. Instead, it does so based on “geography”.
Imagine an island: the port has a population and ships, while the hills have a few trees. As the port gets more crowded, servers dynamically shift what ‘level’ of geography they support based on that activity.
As a user moves through the game world, they have no idea that at one point there might be a bare minimum of computing resource “behind” the game world, while at another there may be a significant swarm of computing resources that have picked up the management of the experience.
The image above shows this concept: the coloured squares represent the ‘servers’ (workers) that overlap in order to manage the dynamic space. Those squares, and the ways they overlap, change as the game world changes, or as new players join or move throughout the space.
This concept of ‘swarm’ computing resources being applied to a game world potentially unlocks the promise of Snowcrash: a truly persistent, dynamic virtual world.
And so it’s no surprise that much of the attention on SpatialOS and Improbable is on its role in powering a new generation of MMOs and online games.
The World’s Next Operating System
But the attention on games and virtual worlds, I think, overlooks the deeper implications of SpatialOS.
Technology needs a new operating system to underpin the next generation of interfaces: from augmented reality glasses and mobile AR to virtual reality and in situ digital screens. Even better would be a technology that also acts as a bridge between these systems and machine learning/artificial intelligence.
SpatialOS may be that platform.
It is built, natively, to manage concepts of space. And while this conjures up images, naturally, of 3D worlds, its implications are more profound (and cross over into topics of “placeness” that I’ve explored for over a decade). Because as our technology has become more aware of the world around us, as it has become better at simulating experiences of that world, the line is blurring between the physically real and the digitally real.
Everything, in other words, is rapidly shifting towards technologies that employ some sense of spatial awareness — whether an Internet of Things device that keeps your home warm, or augmented reality glasses that will overlay digital artefacts on physical world space.
If we view SpatialOS as solely capable of supporting alternate realities, we overlook its potential to power a paradigm in which digital and physical ‘spaces’ have very little separation.
New Paradigms for Computing: Beyond Simulation
This blurring of lines between virtual and physical space means that SpatialOS is a natural simulation engine. It has the potential to scale up to simulate entire cities or an ocean’s weather system. It is, perhaps, most limited by our current capacity to process physics (whether in game engines or more generally) or to conduct quantum scale calculations — but those are less the limits of SpatialOS.
As an OS, it is much like Windows as an OS — it’s the platform upon which highly performative tasks can be conducted, and it has always been up to the hardware to catch up to the boundaries of the operating systems.
Its power to conduct simulations becomes even more interesting when those simulations are ‘attached’ to artificial intelligence/machine learning.
Imagine a traffic simulation where the ‘rules of the road’ are rewritten in real time by, say, algorithms generated by Tensor Flow. These aren’t just simulations with a ‘start’ button — these are simulations that change dynamically on the fly.
And now imagine, further, that the simulation could send data out.
What if that simulation was used to adapt traffic signals in a real city in real time?
The implications are profound.
What’s Next
I’ve only begun to wrap my head around SpatialOS. For close to 20 years I’ve been exploring virtual worlds, simulations, and proximity/IOT technologies. With each of them, we’ve bumped up against limits that have prevented us from fully realizing what we imagine.
SpatialOS promises to remove those limits. Whether it can fully execute on its vision, and whether it can avoid the potholes ahead, will require further thought and monitoring.
But today I’ve seen what the world’s next operating system will look like, and I believe it looks a lot like SpatialOS.
Tagged applications, AR, game development, improbableio, spatialos, VR
Originally published at dusanwriter.net on December 19, 2017.
