Procedurely Generated & Survival Text Adventures

What are people’s opinions on the idea of a partially procedurally generated Text adventure game?

The game would have a storyline and would have goals and objectives, as well as created areas, however, there would be parts where the game would be generated as it is being played, creating a new experience and new challenges every time.

For instance, one idea, but not the only, maybe part of the game has a desert that you need to move through. Instead of building it as a maze, you have it as a fun section of the game where you can explore as much as you like, however you like; but once a certain # of turns has passed in the desert, or once you get a certain item (e.g. a map) that has a certain % of appearing every time you move directions, you find the other end of the desert.

I’m thinking about building a survival text-adventure game with some random elements, as well as a storyline and two manually-built (traditional room-style) areas of the game, where the beginning and end-game of the game happens.

Any ideas or suggestions would be very much appreciated! Thank you in advance.

This approach absolutely can bear fruit, but for the procedural sections to feel rich and varied, bear in mind you will end up doing as much writing work as if you had handcrafted each area.


I would be super interested in theory! I say in theory because, well, new experimental things are interesting but it’s hard to do them well.

In particular there are some design challenges in combining preset narrative and procgen stuff. Procgen stuff is often designed to maximize replay value–it’s new every time in some ways, and you also become gradually familiar with the systems, but you never have to repeat yourself exactly.

Handcrafted narrative does not reward replay in the same way. It’s the same thing over and over again. So there’s a potential for a mismatch here–if the narrative is lots more interesting than the procedural part, people will play it through once and miss the new experience every time. If the procedural stuff is the really interesting part, people will potentially get tired of the narrative parts every time they replay–like a big cutscene.

So it’s important to think carefully about what you want out of the procedurally generated stuff and how it fits into the game! One thing I found in a non-text game was an exploration game where the randomization meant you had to explore for real, instead of learning where everything was for subsequent playthroughs.

OTOH the way you describe things with the manual sections at the beginning and end could work–it might be like nethack where some key stages, including the end, are preset maps (or chosen from a few possibilities). Then the prewritten parts are kind of a reward for victory–though you’d also have to make an entire interesting procedurally generated game, which is very tough!

Some things worth looking at would be Rogue of the Multiverse, which has a very strong narrative with some randomized sections. (I think the consensus is that nobody replays it to get more of the randomized parts.) ADOM is a classic roguelike (not a text adventure) with a strong narrative, as I understand–I’ve never played it much partly because I barely play standard roguelikes that aren’t Brogue anymore. Joey Jones’ and Mevlin Rangasamy’s Calm maybe doesn’t have procedural generation per se but it seems to me like it has a lot of that kind of flavor mixed with narrative? You might also check out Flexible Survival, which is a mind-bogglingly ambitious RPG done in Inform (I should also mention that it is an extremely explicit postapocalyptic furry pornography game). There’s also Legerdemain which is a game in a roguelike style with preset maps and a very strong narrative–at least I think it has a very strong narrative, I never got that far in it.

Another thing you can do is have some procgen stuff that provides flavor in an otherwise conventional IF game. Chris Conley’s room in Cragne Manor is a great example of that. (Runs away cackling before you figure out how long it takes to get to Chris Conley’s room in Cragne Manor.)

1 Like

Procedurally generated games can totally work. Writing for them takes some adjustment perhaps up an abstraction layer. If you think of how some modern boardgames procedurally generate the configuration of a story where the player(s) are welcome to infer plot connections. Mysterious Island is a good example of a boardgame - you know the game and you know the rules, but the challenge is taking advantage of or being flummoxed by how the environmental layout falls. Betrayal At House on the Hill is another - that could be thought of as a generated environment with specifically-authored scenarios that can be “used up” essentially, but dictate how a plot plays out in the environment as it is generated.

I was also fascinated by a boardgame that had a campaign module where the playthrough actually permanently customized the version of the base game the players wind up with after the campaign. It involved tearing up cards permanently, writing on cards and board spaces, giving them a permanent “history” and placing stickers on other pieces to alter their default base stats.

1 Like