Gaming algorithms is the future of gaming

In January 2016, an article in Wired appeared under the headline Games of the future will be developed by algorithms, not humans.

It cited how producing high-quality games had become overly expensive and labor-intensive. The 2014 release, Destiny, cost about half a billion dollars to develop. The argument ran that by using a technique called “programmatic generation”, games could take on more depth and variety. This was thought to be a huge bonus for independent developers with much smaller budgets compared to blockbuster studios.

Programmatic generation is effectively the process whereby a game, or at least a detail within the game, is developed by computers following a set of instructions to amplify what has already been built. It isn’t actually all that new to gaming. In the 1980s, classic titles such as Elite used algorithms to create their environments, planets, or asteroids with necessarily limited graphic quality.

Minecraft: Rapid scale achieved through algos

Even modern titles have quite routinely used programmatic generation to save time on some secondary aspects in a game. This could involve imagining the design of weapons, how plant life is represented, or the way blood spurts from the faces of slain enemies.

Minecraft was the first of the modern wave of popular games to use programmatic generation to create entire worlds, thus allowing it to provide immense and rapid scale to its game at a very low cost. It helps that Minecraft thrives on having an anti-modern look and feel with simplistic block-like structures. Then game concepts like No Man’s Sky, created by Hello Games, in Surrey, England. It used algorithms to create an entire universe.

No Man’s Sky claims to feature an astonishing 18,446,744,073,709,551,616 individual planets for players to explore. Presumably, the vast majority have not yet been visited. Now that algorithms can function so beautifully, you could argue that it’s just as important for game developers to push resources towards their algos as much as towards core game creation.

Personalized stories

Importantly, they function not only in terms of environments but also in creating a personalized story for individual players. Rimworld has what is described as an “AI storyteller” creating events and mini-narratives for the player while Moon Hunters use the technique to create “myths” — some of which are depicted in artwork, about the player’s achievements.

While SuperOne is keen not to give too much away about how algorithms will work in its own game, the possibilities are, if not endless, then at least multi-stranded. And given that it seeks to entertain one million players upon launch in late December 2021, it will certainly use programmatic generation to ensure there are almost as many different questions as there are planets in No Man’s Sky.

SuperOne’s own developers can write questions on various themes, and ask the algorithms to provide the depth. For example, if there was a question that asked you to identify a particular football club’s badge you might create a dozen such versions of the question yourself, and then write an algorithm that allowed machines to source their own club badges and pick ones that might be mistakenly identified as the correct ones.

Algorithms also lack human intuition, which means extensive testing and modification. A particular problem, for example, has been identified with waterfalls in Minecraft. A major challenge is building software that is good at analyzing and understanding what it has created, and this is where algorithmic development moves into the realm of machine learning.

They are, however, an extremely important tool for any game developer and will be central to the SuperOne gaming experience.