This weekend I got a chance to attend Roguelike Celebration: a small conference put on in SF dealing with the design and history of roguelike games. This was a fantastic experience and I wanted to jot down a few notes from it. All of the talks were recorded so if you’re looking to watch them yourself, which I highly recommend, the first stage stream archive is here and the second stage is here.
Permadeth shouldn’t be a painful thing
This came up in numerous presentations. For me, the interesting part was that it wasn’t seen as a difficulty thing by the developers. Instead, it was seen as a way to make choices more weighty. Lots of developers looked for ways to make past games affect future game worlds. The dwarf fortress developers, specifically, presented this as a logical extension of the top scores lists that were in the arcade games that they grew up with.
One great example of this is in Binding Of Isaac, a game inspired by rogue-likes. Each play-though of that game has a pretty good chance of unlocking additional item pickups/characters. Enough “wins” will also make the game harder/longer, which is a great way of adjusting difficulty/not overwhelming players.
Spatial navigation is the hardest thing to make accessible for people without sight
Alexei Pepers gave a fantastic talk on the research she did on making rogue-likes accessible.
— Alexei Pepers (@ampepers) September 18, 2016
Lots of great a11y points here that are not game-specific. If you make interactive products and this stuff matters to your work (it probably does) I urge you to listen to the talk. One of the interesting points made in this talk was that the text-based nature of these games make them much easier to adapt to an audio interface, which ties in nicely into…
Sacrificing graphical fidelity allows developers to create incredibly deep game-play interactions
In their talk Zach and Tarn Adam, creators of Dwarf Fortress, brought up a bug that was causing cats to throw up. The flow goes something like this: a dwarf throws beer on the ground, a cat walks over the beer and gets it on it’s paws, the cat licks the beer off it’s paws, some bad math (the actual bug) causes the game engine to think the cat just drank an entire cup of beer, at which point the math that calculates the BAC for the cat decides that the cat drank too much and should throw up.
This is an incredible amount of game systems all interacting with each other! And it would never be possible with AAA games today. Modeling this behavior would be completely infeasible if each one of those steps needed animations/models/sounds.
The cutting edge of procedural generation involves a human component
I got to witness a play-though of the Bad News game. This was probably the most interesting part of the presentation for me, personally. The “engine” of the bad news game is composed of man and machine. The closest thing I can compare it to would be centaur chess.
A quick summary: the player is a mortician responsible for notifying a next of kin of a death in an unfamiliar procedurally generated town. The town and all of it’s inhabitants are generated (with backstories, personalities, desires, and relationships) for each game. The inhabitants are played by an actor who talks with the player face to face and has a display telling them the relevant information about the person they are playing and what they know about the topic they are discussing with the player. The player goes from location to location asking people questions and tracking down the next of kin.
The part of this I fixated on was how the responsibilities of maintaining a coherent game world were divided between the human and mechanical parts of the engine. The machine handled what everyone in the town knew and where they were. The humans handled the in-person interaction and the construction of the narrative. As the player plays, another person queries the database of the roughly 300 people living in this town to try to create a compelling narrative that the actor can lean on to provide a cohesive experience.
There is probably an entire blog post in here about how the human mind craves a narrative and how we tend to create them out of nothing to make sense of information in a way that’s wholly different from what a computer does. (Why haven’t we created narrative-based databases?) Buuut… this post is already pretty long.