Some Notes on Procedurality research

Picture by John Fowler.

A recent article by Chris Priestman at the excellent indie game website Indie Statik points out that “procedurality” has been increasingly used as a buzzword in the gaming community. It’s not always clear what that word really means, much like the subjective concepts of “smarter AI” and “better graphics”.

The article takes as a starting point the impressive soon to be released No Man’s Sky (Hello Games), a game that claims to have “every atom procedural”. Since the discussion around it brought up some good points, I’ll take this opportunity to post some of my own observations and notes on the subject.

Read another more recent post on this topic here.

My research is about procedural creation in a broader sense, though games are an important part of it. Read more about my study using the blue top-right menu.

Read my tweets on No Man’s Sky here.

Procedural Creativity

A common misconception regarding procedural creation is that the designer or creator has less authorial control over the final product or experience. Maybe this is still true in most cases, specially in the game industry. But this shouldn’t be mistaken as an intrinsic quality of this particular creative approach.

Although it is a fact that in procedural creation there is less direct control over the experience, on the other hand there is an increase in control on a meta-creative level. So a designer can’t choose exactly where to place a tree, for example, but he can define how the distribution of trees work in the game world. It’s a different kind of control, with it’s own qualities and limitations.

In Zephyr’s reply to Priestman’s article he writes:

[W]hy something procedurally generated could have less meaning than something “tailor made”? (…) I think procedural authorship is giving up some things (as an author, I mean) to access other things : you don’t lose control over your creation, you just gain ways to control it differently.1

I believe most of the confusion around procedural creation comes from the fact that this is still a relatively new approach to expressive and creative thinking. Of course, artists have explored programming since the invention of the digital computer, but there is still no clear common grammar through which to talk about these practices. This leads to a great deal of borrowing from and overlapping with traditional art forms such as Cinema, Literature and the Visual Arts.

Process Intensity

I think it’s important to note that procedurality isn’t a concept closed in itself. There are many different aspects to consider, such as it’s level of intensity and what elements it affects in a certain context.2

This relates to game designer Chris Crawford’s concept of “process intensity”. The idea is that there is no absolute procedurality, since in a fundamental level all dynamic processes and behaviors are made up of static elements.3

Priestman’s article mentions Elite (Acornsoft, 1984), a game that in principle could generate trillions of different galaxies for the player to explore. But since this impressive number didn’t actually translate to impressive improvements on gameplay, the designers ended up selecting “only” a few thousands of the galaxies for the final product.

On the other hand, interactive narrative Façade (Michael Mateas & Andrew Stern, 2005) seems to do a much better job in integrating its procedural system to the game mechanics and player experience.

Elite (Acornsoft, 1984; left) and Façade (Mateas and Andrew, 2005).

Elite (Acornsoft, 1984; left) and Façade (Mateas and Andrew, 2005).


There seems to be a fundamental misunderstanding about the very nature of procedural creation. It’s true that many times this practice is used solely as a technique to save time (and money). For example, by eliminating the need for artists to create every single asset of a game. But this is just the tip of the iceberg compared to the creative and expressive potential of this approach.

I think that to say that proceduraly generated content is “repetitive”, or lacking narrative structure, is missing the point. After all, this approach has little to do with the linear and static nature of traditional art practices. One could just as well criticize the movie Scarface (1983) on the basis of it not allowing the viewer to control the main character, for example.

Which makes it ironic to me when games such as Gone Home (The Fullbright Company, 2013) and The Stanley Parable (Davey Wreden, 2011) are mentioned in a discussion about procedural content. These are both great titles, worthy of all (or at least most of) the praise they have been receiving. But the fact is that they are both actually very linear and static, with mostly predefined plots, delivered through prewritten or prerecorded text and dialogue.4

Again, these are great games. I am just questioning if they are really suited to be used as arguments in a discussion about the potential for procedural content.5

Lose Control

Note: the following is a more subjective analysis, very open for debate. Please bear with me.

Maybe these experiences heavily based on procedural content are supposed to be mostly repetitive and homogeneous. The greatest example of that being “real life”, where No Man’s Sky‘s promise of “every atom procedural” is a given (in the sense that “everything” emerges from a series of simple rules).6

But even though this “real life” experience, lacking preauthored (or “tailored”) content, tends to be uneventful (as compared to a movie or game), people still seem to enjoy it. It is of course a kind of enjoyment that is very different from the one we get from a song, a movie or a videogame (considered in a traditional sense) – but it is a valid and very unique kind of enjoyment.

This comment by Ed Key in Priestman’s article sums it up pretty well:

I find patches of wilderness (even in vacant city lots) to be infinitely richer and more evocative than any garden.

Some believe that this “unknown” (or the “mistery”, as put by Zephyr) is one of life’s greatest qualities. To be able to explore this in an expressive and artistic manner through digital media is to me a very interesting prospect.

Photograph by John Fowler.

Photograph by John Fowler.

A couple of paragraphs above I wrote that procedural experiences are “mostly repetitive” and “tends to be uneventful”. The words marked in bold are important. As we all know, exciting, interesting and beautiful things do happen in life, usually when we least expect them. They happen in games too, even those that don’t rely so much on procedural systems.7

Sure, a developer can control the simulation in order to maximize these moments, thus leading to a supposedly “more exciting” experience. I think that’s fine – in a way, this is what happens in all “sandbox” games.8

But I’d also love to see more artists and designers truly exploring the full potential of the meta-creative approach. This means abdicating control over specific situations or instances, to focus on creating a “machine that creates the art” (Sol Lewitt, 1967), or a machine from which the game emerges.9

In the future maybe more people – creators and consumers – will be interested in exploring the nuances in character behaviors and relationships, in “patches of wilderness” and natural rock formations; and maybe find them as exciting as an epic prewritten scene or dialogue, or a carefully crafted environment.


Quick update (2013/12/12) with some information and a quote from this interview with No Man’s Sky‘s creators.

  • They explain how they think they reached the artist’s initial (non-procedural, authored) vision via procedural non-authored content.
  • And how the “every atom procedural” approach is a way of ensuring consistency in the world (colors, for example).
  • “We are designing a set of rules, not designing a game (…)”
  • “If there’s a crashed ship, it’s there because a ship has crashed. (…)”

Finally, this quote:

[T]he moment I loved Minecraft was [when] I just dug down and then there was a cave underneath. It’s the simplest thing in the world, but no one can deny the feeling you get when that happens.

You know that they’re underneath everywhere, and that everyone is having the same experience, but there’s something… ‘Hey guys, check it out.’

It’s probably the best example of a game which is randomly generated but really the game doesn’t feel random. It’s not about the landscapes in a weird way, but it really adds to the game that they’re procedural, which is what we’re going for.

[Posted 2013/12/09, updated 2013/12/12. Added link to tweets on 2015/11/18.]

  1. This quote includes part of the author’s post and part of one of his comments. Slightly edited.
  2. In my masters dissertation (previous research) I suggest a methodology of analysis for these procedural creations (post).
  3. By the way, Crawford was one of the first to talk about the expressive potential of digital media and games. See The Dragon Speech (1992).
  4. Façade, mentioned earlier, also uses prerecorded dialogue, but in a much greater level of granularity. NPC behavior emerges from a complex artificial intelligence algorithm, as well as a rich narrative system based on the concept of “story beats” (Mateas and Stern have many articles describing the system).
  5. There are also titles like Heavy Rain (Quantic Dream, 2010), which in my opinion is a very bad game, and a very good example of misuse of a medium. And even this has been involved in discussions about procedural content (there is a post in my personal blog about this here).
  6. This is, of course, my point of view. If you believe in stuff like destiny, god or a “greater narrative” guiding us all, I totally respect that.
  7. Youtube is full of cool crazy emergent situations in games (not to mention the glitches).
  8. Something similar happens in generative art too, when the artist selects certain instances of the algorithm’s output that best represents his vision.
  9. This is a call for more. Thankfully, there are (and there has been) lots of great artists, designers and creators exploring this approach. There are many examples in this very blog, as well as in my masters dissertation (although it’s in portuguese, I believe the references are easy to find).
See more researchblog.