Join me on my journey of programming music for generated dream worlds of Tea Garden. I’ll talk about my approach and process, and we will listen to a bunch of procedural music. I use pico8 for this project and will also talk about how technical restrictions affect my decisions.

Procedural music is a fresh territory that is explored very little. So I hope this talk will provide you with some ideas to make cool generators of your own.

When the Baltimore Orioles are both a baseball team and a species of bird, online-trained image generating algorithms like DALL-E tend to get confused. Through funny images of a baseball team buying huge bags of oranges at the grocery store, or birds flying to Florida in an airplane, I explore what happens when AI has seen your phrase in another sense online.

Peglin's Early Access release was a massive success, almost entirely due to the incredible streamers and YouTubers that focus on Roguelike games. This talk will be a focused post-mortem on how our relationships with streamers were the cornerstone to Peglin's success, how Roguelike developers and Roguelike streamers have formed the perfect conditions to work together as a healthy industry, and what developers can do to better work with streamers for their own roguelike games.

Creating a modding system for a Roguelike is not for the faint of heart, but it's truly worthwhile. Players love modding systems, and they usually improve engagement and replayability. After working on Transcendence, and studying other modding systems, I've got some tips and tricks for getting the most out of them.

Last year I created an encyclopedic guide for the deeply complex post-apocalyptic roguelike, Cataclysm: Dark Days Ahead, called The Hitchhiker's Guide to the Cataclysm. This talk is about how I went about making it, what the challenges and successes have been, and the reasons why I made it. So-called “spoilers” are sometimes frowned upon or thought of as a bit dirty in the roguelike community, but our relationship with them is more complex than that. Spoilers help the player to immerse themselves, especially in simulationist games like Cataclysm. And often, spoilers are the only plausible way to win. I'll talk about the line I walk with the Guide between being a complete resource for the game's data and inadvertently giving away "surprise" content in the game. I’ll also talk about the incredible open-source community that works to build and maintain the game, and why, personally, building the Guide while caring for my newborn daughter helped me to reassess my relationship with creativity and productivity.

I've made a lot of puzzle generators - the ones that make the puzzles for Knotwords, Good Sudoku, and Spelltower, to name a few. It turns out there are more than a few similarities between creating a procedurally generated puzzle, and a procedurally generated roguelike. Once myself and Zach Gage had that insight, we began to treat the problem of creating a puzzle generator for a new problem like creating the player of a roguelike. This oppositional approach to puzzle design allowed us to create ways of assessing the difficulty of a puzzle, a framework for hints and tutorialization, and a more human approach to procedural generation.

Meaningful player choice is essential to games of all genres, and this short talk provides an overview of the major choice types that appear in games: mutually exclusive actions, risk versus reward, now versus later, resource trade, and dilemma.

To help illustrate each choice type, this talk highlights examples from popular games in the genre such as NetHack, Brogue, Forays into Norrendrin, Dungeon Crawl Stone Soup, and the original Rogue. This talk also touches on choice-driven elements I added to my own reimagining of Rogue and suggests ideas for choices developers might experiment with in their projects.

Ultimately, this talk aims to give developers a vocabulary for designing choices that can help make their roguelikes more challenging, dramatic, and engaging.

This talk follows the journey of Grimoire Groves, a non-violent rogue-lite dungeon crawler. The talk explores which implications a non-violent concept has on game design, the stunning challenges of creating an audio-visual language for such a game, and why we even came up with such a concept. Be prepared for cute and colorful visuals, a quirky game, a lot of love for roguelite games, and tough game design questions.

In procgen games, we often make simulated histories, or when the player dies, have them wake up in the world a generation later as another character. My struggle was how to make world changes feel meaningful. During the deep isolation of the pandemic, I came across a model used in therapy called Internal Family Systems, a technique to view oneself as a being of several parts, each of which have their own needs.

I’ve translated this notion of parts into an AI system called Concerns, where each being in the world is a body with a collection of concerns with individual weights. From this talk, my hope is we can expand our idea of how we can make simulated histories feel playable, and the implementation of AI behaviour in immersive sims, aiming for empathizable and interesting instead of optimal.

Causal graphs are models that describe possible causal relationships between things. They're typically used during inference: given some source of data, how can we explain the patterns we see? This talk is about doing the opposite. Instead of using them to learn about data, we'll generate it.

This is useful because a natural-seeming choice when procedurally generating data leads to something really strange. Imagine you roll some attributes independently, say hometown and favorite food. If you look at the results in Seattle, the favorite foods end up the same as for Tokyo. There is no relationship between the attributes. In the real world scientific studies have to be careful so their results aren't undermined by selection bias. If you only survey middle aged women in Nebraska your results might not generalize. A causal graph can introduce these kinds of relationships into generated content.

The talk will be a more industry/practical look at a paper I am going to present at the Foundation of Digital Games 2022 this summer. The paper details an approach and a interface that allows novice designers to control an established graph grammar for dungeon generation. I am planning to demo the tool I built, and discuss how it can be used as both exploration and specification of the generator.

This will be a fun & gentle, but hopefully immediately useful, introduction to Monte Carlo simulations. I'll draw from my experiences using them in various design capacities, such as:

• Coding a simple sim to analyze an existing physical game (Mafia/Werewolf) to understand its balance and improve how it's played
• Writing a bot to play a huge, sprawling visual novel game (Neo Cab) to explore every nook and cranny of the story, finding dead content, balancing the in-game economy, and fixing dead ends we didn't know we had.
• Combining real user research data from thousands of playerson an in-development trivia game (Cat Burglar) with Monte Carlo techniques to translate a loose creative goal into a fine-tuned, robust difficulty curve that works for potentially millions of players at different skill levels.

I will discuss how to populate a procedurally-generated roguelike with 3D art, when you don't have millions to spend on 3D artists or years to code advanced techniques (this is not No Man's Sky).

What content do you need and where do you find it? How do you incorporate it into your game? How do you make it look good and perform well?

My experience releasing Mech Armada using Unreal Engine will serve as a guide to find answers.

When the developer, half, forked Angband to make Sil, he made major changes to the codebase to create something elegant, thematic, and fundamentally new. Although Angband has tiles, Sil was developed without much regard for the tile display system.

I worked with Quirk to create and implement a new, custom tileset for his variant, Sil-Quirk (Sil-Q). I would like to discuss the process of creating the tileset for Sil-Q, as well as my thoughts on tilesets in general: the balance of representative vs symbolic tiles, the role of visual clarity in roguelikes, the challenges of drawing with individual pixels, the visual tools that a tileset artist has to communicate information, and the lessons I learned along the way. Finally, I would like to emphasize that my experience in creating a tileset was both enjoyable and fulfilling, and I encourage anyone interested in tilesets to try their hand at making their own.

The year is 1981, over one decade before the concept of "roguelike" would come to exist. It's the dark times of the prohibition of "rogue" (at least in the computer labs of the University of Oklahoma), and a young programming enthusiast called Robert Koeneke sets on a journey to create his very own version of the game to circumvent the interdiction. His work would ultimately become foundational for a genre of computer games, inspire many developers to build upon it, and provide endless hours of fun to thousands of players all around the world. 41 years later, Robert has abandoned this earthly plane, so let's take some time to celebrate his great contribution.