Tapping into the potential of procedural generation in Caves of Qud

I’m not sure where to start talking about Caves of Qud. It presents a surreal, post-apocalyptic world to explore. In the classic roguelike style, its graphics are ASCII characters, and there is permadeath and random generation. But its permadeath is optional, and its randomness has a substantial pre-made element. Yet, it also randomly generates a history for the world, and random legends and text, like Dwarf Fortress does, and peoples its world with a lot of strange creatures and beings. You can have some really bizarre adventures in Qud’s fields, towns and caves, and it’s a lot of fun having them.

CoQ’s creators have shared a lot about things like end-to-end procedural generation, understanding bugs, and worldbuilding at conferences like GDC, Roguelike Celebration, and NarraScope, and have written in-depth about procedural generation and procedural storytelling, emergent narrative and reparative play, and procedurally generating history for Caves of Qud over the years as well.

We here add onto this legacy of documentation with the answers to a number of questions about Caves of Qud, how it’s made, where it’s been, and where it’s going.

Who are you, and what is Caves of Qud?

We're Brian Bucklew and Jason Grinblat--co-chiefs at Freehold Games, whose MO is making bizarre, baroque games that feel like overgrown gardens. We've also written and talked a lot about game architecture, generative systems, emergent narrative, etc.

Caves of Qud is a lot of things: classic roguelike, science fantasy RPG, deep simulation, far-future hallucination, maximalist mess. It's tagline is "chisel through a layer cake of thousand-year-old civilizations," and that speaks to its essence pretty well.

The copyright screen notes a date of 2007, meaning, CoQ is approaching 15 years of development! In roguelike circles there have been games that have been going longer (NetHack famously goes back to the '80s), but it's still a venerable age for a computer game still in development. How did you get started, and how'd you keep it going for all that time?

Grinblat: Around the mid aughts Brian and I were playing a lot of Dwarf Fortress and ADOM and toying around with a roguelike engine. Simultaneously we were working on a TTRPG with a labyrinthine, geological, far-future setting. At some point these projects fused into one.

A lot went right for this or any project to survive as long as it has. Sometimes it's a slog. I recently read thecatamites' Large Game Manifesto and the line "wander as a little guy through the temple of your own poor decisions" hit me like a brick. But we've managed to grow a bright, warm, passionate player community over the years and it's been crucial to keeping the momentum going.

Caves of Qud appears to take a lot of inspiration from the early TSR product Gamma World, one of the great RPG settings. Do you have any experience with it, or other post-apocalyptic TTRPGs? What makes that kind of heavily-mutated future setting useful?

Bucklew: My family was deeply evangelical when I was growing up in the '80s and '90s. Dungeons & Dragons was verboten due to the cult panic of the era. They didn’t translate that same taboo to Gamma World 3rd Edition, however, so it ended up being the first role playing game I bought as a kid.

Due in part to that twist of fate, a group of my friends, including Jason, ended up playing a lot of Gamma World and other non-D&D RPGs like Rifts. Those early game sessions had a huge impact on us and served as a springboard into much of the work we do today.

To move on to presentation, Caves of Qud takes a lot of its presentation from classic (and some more recent) roguelikes, with monospace character-style graphics and a complex keyboard interface. Do you feel this holds the game back, or does it improve the play experience?

Caves of Qud leans heavily on the simple, abstract presentation of its content. It’s an extremely surreal world with a huge variety of content, some of which is sensible and some of which is less so. Trying to present the most surreal of the content in a completely rendered and concrete manner would be both incredibly expensive and, I think, less effective than simply having the player imagine that content for themselves.

While it would be possible to create a game like Caves of Qud with a more illustrative style, I think it would be a fundamentally different game, as well as a necessarily more constrained one.

For a game with font-based graphics and a strong classic roguelike presentation, CoQ has amazing atmosphere, much of it conveyed by terrific writing. Its mutants feel really different and interesting. I'm reminded of Dwarf Fortress's animal people and forgotten beasts. What are you proud of, regarding how Caves of Qud animates all these weird beings? Is there something you wish it did better?

Grinblat: Any kind of future fiction is necessarily oriented toward its past. But we wanted the setting's present to have richness, to be more than a collection of arrow-hints pointing toward the conditions that produced it. The weird specificity of the animal factions, the diverse cultures of the speciated humanoids (snapjaws, goatfolk, hindren, svardym), little oblique sayings about mutations — I'm proud of how these things piece together into a living world.

As for what I wish we did better — there's an opportunity to find deeper couplings between process and culture. Whenever I watch a nature documentary I have a habit of imagining what cultural ritual might form out of a behavior; how would a priest describe this as a sacred rite? One that comes to mind is brunsvigia tumbling across the grasslands, dead, but sowing the fields with its seeds. Caves of Qud cares so much about both process and culture; I feel like we could be doing more to explore how sapient beings might spin meaning out of their material surroundings, their bodies, their relationships with the earth.

Qud is a mix of pre-made and procedural content. Towns are mostly the same, but dungeons and wilderness, as well as some of the game's history and lore, are generated afresh each game. Why did you go with this approach, instead of explicitly designing and writing more of the game world?

Bucklew: Though we would later spend a lot of time unpacking the rationale for why it worked, a lot of our early design decisions for Caves of Qud were made based more on cargo-culting in the things we intuitively enjoyed from the games we had played before.

The roguelikes that had the biggest impact on me at the time were Omega and ADOM, both of which had static areas that were wrapped in procedural worlds. Using this approach for Caves of Qud meant we had a static backbone to deliver some of the deep world building and lore that both Jason and I are big fans of.

The procedural elements served the other initial goal, which was to make something we could play ourselves. The game was initially simply a personal amusement for us, something we both wanted to play but didn’t exist, a role playing game set in a sci-fantasy world. So, the procedural elements meant we could play our own game without the experience being a mere review of a bunch of hand-crafted content.

CoQ has always had a somewhat more relaxed approach to restarting than most games that take so much inspiration from classic roguelikes, with the option to disable permadeath. Recently added were two more relaxed ways to play, Roleplay and (especially) Wander modes. I like the opportunity to play around in Qud's weird and evocative world without having to worry about losing. Was there a lot of call for these modes? What have player reactions been like?

Grinblat: Reaction has been very positive! There wasn't a lot of call for the new modes explicitly, but many players have used the debug option to save/load for years, and for a while we've tacitly recognized it as a valid way to play. As for Wander mode, it started as an experiment: what if we turned off creature hostility in most situations? The result is dreamlike in how it morphs the play experience. The world is much more responsive to itself; you get to watch creature behavior play out emergently as if you weren't there. Almost like you're haunting the game.

To me, no mode is strictly better than another. Permadeath serves an important role in encouraging a sort of horizontal exploration of the game space, but that comes with a cost. Roleplay mode lets you carry a character through to the endpoints of what is essentially an RPG, without the tactical mastery that is in many ways vestigial to what the game’s become.

Ultimately we feel the game has enough dimensionality to make the trade-offs of each mode worthwhile across the whole matrix.

A quick list of fun items and ideas in Caves of Qud: normality gas, Spray-a-Brain, neutron flux (don't spill it!), cloning draught and metamorphic polygel, Schrodinger's pages, brain brine, primordial soup, liquid mechanics in general, nanocrayons, the portable beehive, and the mutations Precognition, Evil Twin, and Corrosive Gas. Each of these things would be hugely fun if it was the only cool thing in the game, but Caves of Qud has all of them, and more. What's your favorite weird and awesome thing in Qud? How do you come up with ideas like these?

Bucklew: Quite a lot of these ideas pop out of concepts pinging between the heads of a small team of people who love weird stuff. Most of the time a simpler concept is brought to the table on purpose or by accident (mishearing someone say something is a common source of the weirdest ideas). The rest of the team will push the idea around with their own personal styles of make-believe and quite often something deeply and delightfully weird will pop out at the end of those games of telephone-design.

My favorite weird thing might be the climactic event of the game’s main plot, which isn’t yet in. It was the clear result of one of the telephone-design games between team members. You’ll have to wait a bit longer to see what it is, unfortunately.

Grinblat: Right, there are many angles in. Sometimes it's an accident, sometimes it's a more fully-formed top-down design. Sometimes it's bottom-up and structural, like when the entity component architecture gives you a view into the possibility space you wouldn't have had otherwise. That's how one of my favorites came to be: animated furniture. One day we realized we could put the Brain component--the part that controls most creature behavior--on a table object. Now we have a walking, talking table. Okay.

A few more of my favorite weird things:

I feel like this should be a standard question by now, but... Caves of Qud is made with a really small team. Do you have any advice for people trying to found/make/promote their own dream projects?

Bucklew: It’s strictly possible to succeed but it’s a long road. You’re often working on these projects in your spare time with many other pressures asking for time and it can be a challenge to stay motivated and make progress. What has worked for us is to ignore the advice to make small and simple projects. Instead we bit off something vastly too ambitious. However it was something important enough to us to keep us coming back to the table when things got tough.

Pick something that lights your fire and try to do a little bit of it every day, even if it’s 30 minutes. If you don’t feel it after that time, set it down and go do something else. Often enough, you’ll find that those 30 minutes turn into a productive session once you get started. String enough of those sessions together and, in my experience, after 15 or 20 years you’ll have most of a video game on your hands!

Where is Caves of Qud going from here?

Grinblat: A real 1.0 release in the near to mid-term. That means the whole new RPG UI added, and the main quest finished. After that, porting to other platforms, and DLC: we're already brewing a real mess of ideas. After that, ????