Indie Studios & Content-Generating Algorithims

Games today rely on a lot more than the quality of the design ideas alone. There are many smart, enthusiastic and savvy indie game developers out there that have ideas worth making into games, but in many cases the games that end up being made have a hard time competing with those from big studios that are built on more established game design structures. The thing setting the two archetypes of games apart, besides sales numbers, is simply the manhours and high-priced design tools that large studios can afford while smaller ones cannot.

Most of the time, it seems that indie developers simply have to cut down on development time/cost in the simplest way they can: making less content. Fewer assets, fewer gameplay mechanics, less animation, and a generally shorter game are a result of developers with small teams making the tough choices that give their game better odds of actually getting to release intact.

A couple of the indie games that have made it big over the past few years have bucked this trend somewhat, and with teams that were (when the games first started getting popular) as small as technically possible (i.e. solo developers), but somehow ending up with more content in the game than any one person could ever rationally consume. Could this signal a new front in indie game development (or even in triple-A development)? I can’t say.

If you haven’t guessed by this point, the games I’m referring to are mainly Minecraft and Dwarf Fortress, though I’m sure many other indie games that fit this mold are in development currently.

Is it easier for an indie developer to design a game around randomly-generated content rather than building it entirely by hand? Probably not. To begin with, it’s not like Minecraft (for example) doesn’t require a lot of handcrafting. For now, “random” content-generation in games is decidedly limited. Textures aren’t randomly generated, 3D models aren’t randomly generated, AI isn’t randomly generated (though some AI systems in games are majorly impressive in that they can adapt and act organically, they are still very handbuilt), gameplay mechanics aren’t randomly generated, and so on.

Not to mention, the actual work of making code that can build a game world dynamically/randomly is difficult and non-intuitive. It’s time consuming to give a level designer whatever software program they use and say, “make me a world”, but at least you can be assured that they know how to do it and given enough time and money can make a big, interesting game world. World generation, on the other hand, is still mainly uncharted territory, and games like Minecraft use algorithms (which are better than any I’ve ever made) that aren’t remotely as capable as human level designers.

Perhaps the irony in hoping that random asset generation could be the saviour of small-team development (I’m not trying to insinuate that the indie development scene needs a saviour, either) is that the work required to make a game that is truly built from algorithms alone would in fact require a large team of programmers.

Really though, all this is just what I’ve gathered and come to conclusions about. Am I missing something? Do any indie developers (or just enthusiasts like me) want to weigh in on the whether random world/asset generation can ever or will ever replace human design?

5 Comments
5 Comments
Posted by Fripplebubby

Games today rely on a lot more than the quality of the design ideas alone. There are many smart, enthusiastic and savvy indie game developers out there that have ideas worth making into games, but in many cases the games that end up being made have a hard time competing with those from big studios that are built on more established game design structures. The thing setting the two archetypes of games apart, besides sales numbers, is simply the manhours and high-priced design tools that large studios can afford while smaller ones cannot.

Most of the time, it seems that indie developers simply have to cut down on development time/cost in the simplest way they can: making less content. Fewer assets, fewer gameplay mechanics, less animation, and a generally shorter game are a result of developers with small teams making the tough choices that give their game better odds of actually getting to release intact.

A couple of the indie games that have made it big over the past few years have bucked this trend somewhat, and with teams that were (when the games first started getting popular) as small as technically possible (i.e. solo developers), but somehow ending up with more content in the game than any one person could ever rationally consume. Could this signal a new front in indie game development (or even in triple-A development)? I can’t say.

If you haven’t guessed by this point, the games I’m referring to are mainly Minecraft and Dwarf Fortress, though I’m sure many other indie games that fit this mold are in development currently.

Is it easier for an indie developer to design a game around randomly-generated content rather than building it entirely by hand? Probably not. To begin with, it’s not like Minecraft (for example) doesn’t require a lot of handcrafting. For now, “random” content-generation in games is decidedly limited. Textures aren’t randomly generated, 3D models aren’t randomly generated, AI isn’t randomly generated (though some AI systems in games are majorly impressive in that they can adapt and act organically, they are still very handbuilt), gameplay mechanics aren’t randomly generated, and so on.

Not to mention, the actual work of making code that can build a game world dynamically/randomly is difficult and non-intuitive. It’s time consuming to give a level designer whatever software program they use and say, “make me a world”, but at least you can be assured that they know how to do it and given enough time and money can make a big, interesting game world. World generation, on the other hand, is still mainly uncharted territory, and games like Minecraft use algorithms (which are better than any I’ve ever made) that aren’t remotely as capable as human level designers.

Perhaps the irony in hoping that random asset generation could be the saviour of small-team development (I’m not trying to insinuate that the indie development scene needs a saviour, either) is that the work required to make a game that is truly built from algorithms alone would in fact require a large team of programmers.

Really though, all this is just what I’ve gathered and come to conclusions about. Am I missing something? Do any indie developers (or just enthusiasts like me) want to weigh in on the whether random world/asset generation can ever or will ever replace human design?

Posted by Ubersmake

I think it really depends on the game. For example, I can't imagine a non-random Nethack. A lot of the fun is seeing how far down you can make it, and what dungeons get generated and how that random generation might help or hinder you.

On the other hand, I can't imagine a totally generated Journey. Or a totally randomized Super Meat Boy. Those games were designed with a particular experience in mind, and are so focused in their implementation that you don't really have a lot of room to add random elements (though you can add them, like random item or enemy placement).

If you're going to build a sandbox for people to build stuff in, I think random generation is the way to go. Same thing with sandboxes that are there to provide some sort of high-score challenge or bragging rights. That way, no two cities will ever end up the same way, and no two journeys into the deepest dungeons will ever have the same story. If you're going to build a sandbox to explore, where there are things to be found, then it has to be created. Because the developer is trying to provide a certain experience in the latter case, which was the case for games like Dear Esther and Journey.

"It depends" is a hell of a cop-out answer, but I think it makes sense for this. It depends on the experience the developer is trying to provide, and in the case of Minecraft and Dwarf Fortress, and for older games like Nethack, and even games like SimCity, this means randomly generating the world so that every experience can be different.

Posted by granderojo

The more I learn about video game development the more I learn that at it's core it's a craft. So what you're asking is, can we build a processes to lighten the load on making a craft? I would say yes in due time, its happened in other industries.

Posted by ShaggE

I can't imagine it being THAT much easier. Most content generators use pre-built chunks, and those chunks have to be designed in a way that ensures the resulting level is playable.

Edited by Fripplebubby

@Ubersmake: What you're saying is in many ways the way I view the industry currently, in that the games that utilize random generation are effectively designed around that component. The way I view it is this:

At its core, all level design in games is based on decisions, which are the product of some (perhaps not clearly defined) parameters. I didn't actually play Journey, but from what I know of it I would guess the (oversimplified) parameters of the level design would be: rolling desert with sparse, aesthetically pleasing structures here and there wherein the player solves proto-puzzles to advance to the next area (if I'm offbase with my guesswork, just pretend that Journey is the game that I described for the sake of argument, as it's all very arbitrary).

The level designers for Journey would have chosen throughout their design process where to put the models and structures, the exact design of the puzzles, and the path the player takes throughout the game entirely based upon the parameters listed above. If some sort of program could be made that could perfectly adhere to those parameters, I suppose that it would be technically possible to have an infinite Journey. Don't get me wrong, it would be ridiculously, mind-numbingly, soul crushingly difficult to make a program intricate enough to handle that, but the idea fascinates me.

@thabigred: Sure, if a program could be made that would just pump out textures and models nonstop that would go a long way towards leveling the playing field between large studios and small, but I don't know if such a tool is some sort of middle ground between what we have now and what I detailed above or another challenge entirely. Probably the latter.

To address the craft point: I'm sure a similar idea to the one I'm talking about has been brought up a million times since the invention of the computer in regards to other mediums, namely books and music, and so far it hasn't worked out so much there (well, the music front is sort of interesting but not prevalent on any scale). I really can't say how much of a game is artistic choice and how much is calculated rationality, or if there is even a difference between the two. It makes me think.