Recently Rainer tried to introduce the readers of Der Standard to the wonders of procedural generation with this article, but I had already plunged the further depths of this topic in last year's article for games publication GameStar. During my extensive research, I conducted a series of in-depth interviews with a few notable game designers on procedural generation in games today. In this interview, the second of a series of diminuitively nicknamed "footnotes", I talked to Jim Rossignol, game journalist extraordinaire and head designer at BIG ROBOT.
Let’s kick it off with a blunt question for the developer in you: If you would have had the resources, would you have preferred using hand-crafted terrain and landscapes for Sir, You Are Being Hunted? Or, to put it differently: Was procedural generation more of a means to an end or a central aesthetic choice?
There's a lot to be said for hand-crafted worlds, and in all honesty I would love to see a big studio approach the sort of game that we're making with a huge team of artists and level designers. That said, when these big studios do make open worlds - Skyrim, Far Cry 3 for example - they use those teams of artists to build on procedurally-generated landscapes. The results of this is often fantastic, and I think the future is probably a mix of the two approaches. Even our approach is a mixture, one limited by our small budget. There's something enormously powerful about procedural generation tools, and the true extent of its power has not yet been explored. Consequently the true potential of a mixed approach has not been explored. Exploring the depth and extent of that power is something that interests both me and our lead programmer, Tom Betts.
Big Robot’s first games showed quite different approaches to procedural generation. Fallen City did not have much of it, but it was a crucial element in both AVSEQ and Sir, You Are Being Hunted. Since you are not a programmer at heart or by profession, is the experience developing games different when the games are less “handcrafted”? Is it maybe harder to give creative input for you in this game than it was in, say, Fallen City? Do you think that designing a game heavy on procedural generation is a fundamentally different experience?
It is fundamentally different, but it requires no less input from designers. Rather than designing "the first level", we have to design the entire game at once. With Fallen City, while we were refining systems right to the end of the game, there was a lot of work in simply building the levels after we had designed the systems that would operate in them. With Sir, and to a lesser extent AVSEQ, a change in any of the rules changes the entire game. It is a very different way of operating, but not entirely different – the requirement to actually design the systems and the content remains, it's just that the process of getting to a playable game takes on a different sort of pace.
To ask more generally: In your experience, is there a different set of skills, or even a different kind of developer, needed when you design a game that makes heavy use of procedural generation? You get the impression that the closer one moves to procedural generation, the more the work resembles that of an engineer/designer/programmer, rather than that of an “artist” in the classical, Western sense of the word. Any thoughts on that?
This depends on how procedural generation is used. It might be used to create some unchanging underlying shapes - the structure of Eve Online's galaxy, for example - or it might be used to create something unique every time, like a Minecraft world. It could be said that Minecraft's developers are more like engineers than artists. However, as I mentioned previously, I think there are different recipes and these can encompass the strengths and interests of all kinds of developers. Sir, You Are Being Hunted requires constant input from animators, while relying completely on Tom's procedural generation. Something like the original Elite, though, was all engineered, and required no classically understood "art" beyond some simple vectors. See also Dwarf Fortress and the ASCII Roguelikes.
Speaking of which: One interesting thing about procedural generation is that on the one hand, it opens up possibilities for the video game that no other medium has and could thus be considered a defining aesthetic element of the medium. On the other hand, it clashes hard with the preconceived notions of what is “art” in the Ebert-line of thinking, where “authorial intent” is front and center. Any thoughts on that? Do you think that this point plays into the world of misunderstandings in the whole “are games art?”-debate? Do you think that things will change one day?
Sir, You Are Being Hunted is procedurally generated so that each world, and therefore each player's experience, is different. Does that remove authorial intent? I would say no: every player is still hunted, and will experience that same fear. Every world still satirizes British class systems, and any player can read that from the game world, no matter how differently structured that world might be. Artists use tools to express themselves in their work, and mathematical, generative systems are tools. That they might generate something that creator did not intend or anticipate is often the joy of imprecise tool-based methodologies - look at cut-ups or paint splatters.
While games with heavy use of procedural generation have been around for a long time, I still got the impression that we are witnessing a sort of renaissance lately, what with all the rogue-like crossover-genres, the endless runners on mobile etc. In case you’re inclined to agree, what are, in your opinion, the reasons for this resurgence?
Small teams, and talented programmers. Procedural generation is not a simple approach, but it is one that can produce dramatic amounts of content, which often is what small teams need to reach their goals. We need to create detailed landscapes for our games, and we can do it in minutes with our engine, compared to hours of work by hand.
Also, do you feel like there’s a fundamental qualitative difference in how procedural generation has been used lately, compared to its use in games of, say, the 80s?
That's inevitable when you look at the end technologies. Earlier games could not rely on libraries of shaders or vastly powerful processors to do endless calculations on the fly, and we now have all that at our fingertips. The potential application for them is therefore multiplied and the way it has been used is necessarily expanding into things that were inconceivable thirty years ago.
If one squints really hard, one could maybe see a schism on the horizon: In indie games, dynamic generation of elements is seen more and more often, and it is often a crucial part of the game design. In AAA titles, though, with a few exceptions that prove the rule, it is rarely used visibly; it’s more likely, for example, used for generating terrain procedurally in the development stage, which is then “corrected” by artists before it’s put into the game in a largely static way. That said, can you imagine a (more or less distant) future when procedural generation will be an even more central part of game development, even for AAA-titles? If so, what does it take to get there, in your opinion?
As mentioned previously, I think the mixed approach – using procedural generation to create the content and then have artists add the intelligent, edited pass – will come to dominate both indie and AAA approaches. It is the most fruitful process for both parties – it gives us the mathematical advantage of abundant, automated content, and the control of hand-designed worlds. I think AAA studios will lag behind in this because they are larger and less flexible - indies can move faster and iterate experimentally in a way that large studios cannot. However, content-heavy studios like Rockstar and Bethesda are already looking at these systems with a view to altering their workload.
To talk to the developer/speculative designer in you again: Procedural generation seems to be vastly more likely (especially in the AAA-sector) to show up in certain genres, while in others, it’s hardly ever present. Are there genres (existent or non-existent) which could profit, in your opinion, of embracing procedural generation?
I would love to see a big RPG studio tackle a game in a generative fashion. Imagine a Mass Effect game where pre-scripted events were seeded across a generated world! That stuff excites me enormously, and I hope those kinds of companies are able to tackle this sort of problem effectively. They may not do, of course, because their current methods make money as it is!
You once wrote that it is a pity that the often magnificent worlds created for a game are seldom, if ever, re-used as a setting or asset for other games. Could you imagine that this might change one day thanks to procedural generation? Like, it’s already common practice to use middleware like SpeedTree to create certain assets procedurally; could you imagine that one day, a developer could just buy the technology behind, for example, Fuel’s procedural terrain generation to make a different game with it? Or, to ask more directly: are there any plans to license the British country-side generator, which powers Sir, You Are Being Hunted, or even to give it away for free to modders?
Absolutely, and I hope this happens. We do have some plans to explore licensing our tech, but it will take a lot of cleaning up to get there. And I would totally love to buy the FUEL tech. What a world! Procedural middleware is already out there, and being bought up. There was a city-generator that appeared a couple of years ago that vanished with a few weeks - I understand it was acquired by a big engine company. I wonder which one!
To finish with a general question: What role will procedural generation play for game development in, say, the next 10 years? (To be more specific: Do you think that the dream of the “completely automatically generated” game is likely to come true one day? Would you say that it is even a dream worth pursuing?)
There are a couple of issues here, and one I think is based purely around commercial pressure. As the fidelity of games grows, the teams of artists required for the "by hand" approach will grow and grow. People are saying that it has already hit its logical maximum. This will mean that more attention will be given the generative and procedural approaches toward content. Once the really big players are forced to turn their attention to it, then I think we will see the biggest strides made.
A completely automatically generated game is certainly possible, but at that stage you are looking at AI as well as generative technology. There are people working on this very notion at Goldsmiths College in London. In the next ten years we'll see some more ambitious uses, commercial and private, and we'll see someone do something really astounding. I hope it's us!
Good luck then and thanks for your time!
You can find the first interview in the series here, where I talked in length with Chris Park of Arcen Games. Check back later for interviews with Pwnee Studios (Cloudberry Kingdom), Jake Solomon (XCOM: Enemy Unkown) and Michael Cook, who is developing an Artificial Intelligence which makes games (and one day will probably rule us all).