Independent game developers are enjoying the endless possibilities (and failures) that procedurally-generated content presents

by Paul Dean


Independent game developers are embracing randomness. They’re shrugging their shoulders, throwing their hands up and admitting their games have a life of their own. They don’t know what these games are going to do and, most of all, they don’t mind one bit. Sinister as it may sound, they’ve handed control over to the machines, but it really isn’t so bad.

It works a little bit like this: Imagine you’re sitting down to play a game of solitaire, any of the many variations that exist. You shuffle the deck and deal yourself a tableau of cards. The arrangement is random and, because of this, all kinds of things might happen in your game. It might be an easy one or a difficult one, perhaps even impossible. You laid the cards out according to predetermined rules, a particular procedure that defines the limits of the game, but otherwise their combination is unique and you play the game because you don’t know what’s coming next. Every time you shuffle and lay out the cards, your game will be different and this appeals to you.

Now, imagine you’re meticulously machine-coding a space simulator that you might call, say, Elite. The limits of your early ‘80s hardware means you can’t store all the data you want to include on your star systems, planets and people, so instead you plump for another option. You instead tell your game how it can randomly generate its own data. You give it a procedure to follow that will allow it to build its own planets and people, on the fly, in new combinations every time it is played. You’ve taught the game how to procedurally-generate its content, how to endlessly recombine its elements, and you hope this appeals to its players. It’s not so far removed from teaching it how to reshuffle a deck of cards.

Elite was far from the first game to do this. The infamous dungeon-crawler Rogue, for example, reshaped its underground labyrinths time and again, and did its successor NetHack. It was something of a trick of the times and the constraints imposed by fledgling computer technology actually ended up creating endless possibilities for gamers, as early developers would code instructions for procedurally-generating any and all kinds of content. But the technique began to fall out of favour as computers increased in power and developers began to realise the complex worlds that their predecessors could only dream of. The detailed, scripted and finely-crafted environments of games from Doom to Bioshock to Mass Effect were all deliberate affairs, built by hand for a particular purpose or to offer a specific challenge, often by large teams who would labour over everything from the layout to the lighting. There were far fewer rolls of the dice, little was left to chance.

Procedurally-generated content continued to lurk in the shadows, appearing from time to time. It created random maps in Civilization, governed traffic flows in the Grand Theft Auto games, built the dungeons of Diablo and shaped the difficult, arbitrary and randomly-created racetracks of F-Zero X’s highest levels, but in an era of ever more scripted games it was increasingly sidelined.


Now, however, procedural generation is seeing a resurgence in the indie games scene, among a community who, unlike the programmers who have inspired them, are constrained not by their hardware, but instead by team sizes and money. “Content is bad,” declared Introversion Software in a GameCareerGuide development feature, because “the costs of creating game content, in terms of both time and money, are increasing at a tremendous rate.” Developing a reliable method of procedural-content generation would leave much more time to focus on other elements of the game, such as its engine and core mechanics. Many indie developers had the same idea. Why design and sculpt all your levels, missions or challenges when you can get your game to do the work for you?

“I think procedurally-generated content’s firmly back in fashion,” says Ian Hardingham of Mode 7 Games, the team behind Frozen Synapse, “It cuts down on the number of people you need to make a game. It’s one of the greatest things computers can give us in games: Our own private designer sitting in our computer ready to give us new content whenever we want.” Frozen Synapse procedurally-generates its player-versus-player arenas. When it shuffles its deck, it deals out different combinations of buildings and obstacles, meaning players will battle across many different combat arenas. The benefits are twofold, because not only has nobody had to spend time pre-building these countless arenas, but it also prevents players anticipating or trying to exploit arenas they’ve played in before.

“Procedural generation can be a lifesaver for small teams that may not have the resources to hire huge numbers because the algorithms do all the heavy lifting,” agrees Danny Day of QCF Design, developing the bite-sized Rogue-like Desktop Dungeons. “The other big benefit to procedurally generating content is that, provided you do it right, you have an almost unending fountain of stuff for players to get into. Yes, a meticulously designed map or environment is often breathtaking, but it will never be able to consistently stun or amaze veteran players the same way that procedural content can. Building a really great generation system is tricky too, but for many devs it’s an easier problem to solve than finding, paying and communicating what the game needs…”

This is just a small extract from the full feature in Issue Two. To read the entire article, click here to purchase the complete digital edition of Continue for just $2.99/£1.99/€2.25.


Comments are closed.