Creating a Game With Pen, Paper and AI

Hi! Can you tell us who you are and what you do?

Hello, it’s Osebo here. I’m working as a designer at Castle—a mobile game engine and social experience platform where people can play lots of games, create their own using our simple yet powerful editor, and socialize with other users inside the app.

Aside from working at Castle, my partner and I also made couples.game. It’s a fun little chain-guessing game we update every day. Try it out for yourself and see if it’s as fun for us as it is for you!

What game are you playing these days?

It ranges. I'll break this down into mediums. I'm a fan of Zach Gage, by the way. My partner and I love crosswords and different kinds of word puzzle games. During the holidays we actually went to a grocery store and picked up a bunch of magazine puzzle games that I'm not sure if people play as much as back in the day, but I think they are still quite relevant.

I mentioned Zach Gage because I bought his crossword books, and we've been working through those. To be honest, I'm not actually the best crossword player, but I've been trying to get better. My partner, she smokes me at them so I'm getting my reps in so I can get a bit better and hopefully complete them without having to get hints as much.

On console, I have a Switch and I've also been playing the PS5 a bit at some of my friends' houses, but mainly just play the Switch. And I've been playing a game from Bill Basso called Animal Well, and it is kind of like a survival puzzle platformer that I've been. I haven't played it through fully yet, but there are a lot of secrets and the art style is incredible. So that's been a fun one to work through. I've also been playing just a lot of roguelikes, and I would say that's probably my favorite genre. So I've been playing a bit of Vampire Survivors, which is a roguelite horror kind of casual game.

And then on mobile, I'd be lying if I said I wasn't playing Balatro, but I am. I'm definitely a fan of localthunk. I appreciate the pseudonymous anonymity in the way they've been releasing their games. And I think it's interesting that apparently they were kind of working this Zamboni job in college and grinding through making the game over a few years. So it's quite humbling to see that one. He released it solo while also learning sound design and art and programming all along the way and creating probably one of the best games of 2024.

But I actually listened to a podcast that he was featured in and he mentioned the game Luck Be a Landlord. It's kind of like a slot machine roguelike that has this mechanic where you essentially roll the slot machine to get coins, and the coins that you get after spinning you use to pay your rent. And I live in New York, I'm also facing some landlord issues right now, so the game has felt quite relevant to me.

I don't know if you all have played Slice & Dice as well. The Eggplant podcast has a pretty good deep dive on that game, but it's kind of like a dice battle roguelike that I've really appreciated. I was playing that earlier in the year. But I'd say that Balatro definitely kind of took precedence a bit more very quickly.

Also, two others that I have to mention, I do play asynchronous chess.com, and I definitely spend a lot of time playing that. And in New York, there's a lot of like chess culture that has really picked up in like a social context, which is really interesting. It's kind of sexy now to have chess kind of meetup events where some of them there's a focus on chess, some of them less so. There was a chess group that collaborated with Margiela, which is interesting.

One other medium I have to talk about is VR and Horizon Worlds. I can't say that I play all the experiences in Horizon Worlds and I'm biased, but Super Rumble has been a lot of fun to play. It's like a Battle Royale-style game where you essentially have an assortment of weapons that you can pick up and power-ups to essentially win the round based on a point system. So that's been a really fun one.

How did you get into gaming?

So I'm from D.C., Maryland and I was introduced to games initially through my father. We don't have a close relationship, but growing up he got me a PlayStation, the original. And I would always go to friends’ places and play the Nintendo 64. Some of my earliest memories were with those two consoles. I would play games like Frogger and we played Tekken often. He loved Tekken. So we would just spend a lot of time playing that game having, you know, the versions that persisted through the original PlayStation, PlayStation 2, 3 and so forth.

But I also got into computer games, I would spend a lot of time playing RollerCoaster Tycoon  and things like that. But also I think where it really got deep for me was playing games like RuneScape and World of Warcraft. I remember in school we had a club where our science teacher would take us to the wilderness in RuneScape where we'd go and do stuff. He was relatively high leveled and had a full dragon set and whatnot and would go around looting and helping us level up as individuals. I remember that was a game that was pretty formative for me. 

I actually remember vividly how I stopped playing. I stopped playing because as someone who was naive, I was standing in the bank and someone who was a scammer said, “Hey, if you say your password out loud, then it will show up as stars.” Me, having worked on my account for a while, said my password and within seconds I was logged out. And I remember chasing them down. I leveled up a new account and chased them down, screaming at them to give me my old account back. And I think I said a few things that were not very nice and ended up getting banned or that account at least. So that was kind of the end of my RuneScape journey.

I also did play D&D a bit and it was, you know, it's within the same realm of like the other games that I mentioned, RuneScape and World of Warcraft. But D&D is an interesting one because there's a story around this where I had a D&D book that I think my mom donated. But someone who worked at an agency that I ended up collaborating with messaged me on Facebook independently with my name written out in the book. And it was interesting because I didn't realize that this book had been sold off. So someone's messaging me saying, “Hey, you're asking to return this D&D book. Is this your book?” And I was like no, but it very clearly was my name. And as we were talking about before, I don't think there's anyone else who really has my name. So they're like, “Are you sure?” And I was like, “Yes.”

But then I think five or six years later I'd message them again saying “Hey, that was totally my book.” And they're like, “Wow, you really came back for it that long after you'd been thinking about it.” But yeah, those are the games that were most formative for me and I think especially World of Warcraft and RuneScape probably were the seeds that kind of planted my interest in social gaming.

How did you end up learning design and programming?

So I went to school for Fine Arts and Information Systems. I actually started with Computer Science but then I switched because I wasn't able to also pick up the Fine Arts. I wanted to do both. At the time I went to school in D.C. and the school didn't have a concrete kind of interaction design kind of focus, which is what I was ultimately interested in. What was cool about the Fine Arts focus was I was able to kind of choose an assortment of classes as long as they constituted art. So I just chose a bunch of Computer Animation, Illustration and Graphic Design courses. So that's where I got that foundation.

And then my Information Systems focus also helped me pick up a lot of programming-related classes. So the two worked together super well. And at the time I had actually started my work working at a local agency that was in D.C. and that's where I really was able to focus on the craft. With that kind of environment, you work with a ton of different clients, they all have different needs. You're doing everything from branding to editorial to web design to mobile design. So it was a great way for me to just go really deep and kind of refine my design talent.

While that environment was great, I actually quickly learned that I like to just kind of focus on one particular area and just go in deep. So I ended up working at a company called Splash in New York, where they kind of make event web pages or they have a tool that allows you to make event web pages. So I was a designer there and I enjoyed working on those products. But one person who ended up joining Splash had actually previously worked at Tumblr. And that's when I was just hearing their experience working on it, I realized, ”Oh, that's the type of work that I am kind of most interested in, working on social systems!”

Working at Tumblr

So, yeah, earlier on I ended up working at Tumblr and it was an experience that was quite formative. I worked on an assortment of things from the way people post on Tumblr, like the camera and post composer kind of messaging features and whatnot. This was right after the Yahoo! acquisition. It was an interesting time because the lineage of Tumblr is interesting because Yahoo had been bought by Verizon Media Group, which then rebranded to Oath. And I remember there was a time where they were asking people to write down their oath, which was, you know, Tumblr is a group of people who are kind of scene kid-esque, and it was not, I guess, of interest to many people. But yeah, then they were like, “Hey, we are going to pull back our cloak and you know, we're a Verizon Media Group again.” And yeah, so that was an interesting journey at the company I had left at that point. But then they also sold the company for, I think around $2 million to Automatic, who is now the current owner of it. And God knows how that's going. But yeah, I started working on social networking stuff at Tumblr, but that kind of led to me then joining Facebook.

Working at Facebook

I actually started off working on pages. I was working on a redesign of the Facebook page and other stuff that was compelling at the time, like how pages kind of connect to the rest of the Facebook app ecosystem, like Instagram and whatnot. But at the same time, I was very gung-ho about VR. I worked on some social VR kind of experiments at Tumblr, specifically WebVR, and they were very exploratory. It wasn't something that the company was probably going to really prioritize, but it was a fun way to kind of figure out what these social systems could be in more of a spatial context.

And it kind of continued on when I joined Facebook because at Facebook there was a group in Oculus that was focusing on essentially how you can create VR animations. There was like a whole scene of virtual animation happening. So Google kind of had its own version called Tilt Brush in addition to another tool called Google Blocks. Tilt Brush was open source. Google Blocks was discontinued. And there's another company called Gravity Sketch that focuses on something similar, but more so for product design. So Nike uses it for footwear as well as Puma, and there are automotive companies that use it for Ford and whatnot for more conceptual design. But those types of tools are really interesting to me because they focus on this theory of people creating within the medium that the product would be experienced in.

So you're creating for VR, you should be able to design in VR or you're creating a real-world product, you should be able to see it as quickly as possible in 3D. And I think there are 2D tools that have the same thesis, and Castle is one of them. If you're designing 3D games for your phone. you should be able to design it on your phone. There was a group, specifically, this person named Inigo Quilez, who was brilliant. He started this project called Shadertoy that he's been upkeeping for a while. But he had built Quill during a hackathon where he worked with a guy named Goro Fujita, who essentially creates these co-illustrations every single day. He's a prolific animator. They had partnered together and Inigo made a tool that Goro could then use to make these really compelling VR animations. And when I saw this, I was very blown away, being able to just like instantly see an animation in VR and also like edit it and create it, play it back.

And I was able to get access to this tool that allowed these animations because they kind of. They lived inside Quill. You could export them as FBX animations, which I was able to do. They had a lot of geometry but tools allowed them to render quickly in Unity. So I created a kind of social VR museum of virtual animations, pulling them from Quill, Tilt Brush, Google Blocks and some other tools. 

So, long story short, I joined what became the Oculus Media team. Initially, I actually wanted to work on Quill, but at the same time, a friend and coworker, Matt Schaefer, had joined, who was an incredible VR animator. He also came from architecture. A truly talented person had just joined, and there was room for me to contribute. But I felt that there was something more compelling in another part of the org, which was related but different, which was Oculus Venues. I ended up joining as the design lead for Oculus Venues, and there's a lot that I can dig into there.

Tell us more about your work in Castle

To me, Castle is a cognition gym. I felt that a lot of the tools, specifically, like, they're not even tools, but a lot of games that we see nowadays are kind of things that you play in a mindless context. And right now, in a world where LLMs are increasingly kind of taking away our autonomy to think, I feel that we are now more than ever in need of more cognition gyms. And I think that a lot of these daily games are ones that encourage thought.

So Castle is not a daily game, but it is a place where you can create tons of different games. The tools that Castle currently provides, the editor, is one where a lot of people, people primarily in Gen Z and Gen Alpha are using these tools to create games. And right now there's no kind of LLM, so we have kind of our own abstracted programming system that allows people to create scripts and you can also write LUA, but it teaches you some of those basics.

In addition to using art and sound, there's a DAW in Castle to create games. So it felt like something that really strengthened the mind and allowed us to help people think. So I see cognition gyms as something really important now, and Castle is something that fits into that category of tooling.

The Castle team

Castle is small. It's about nine of us. And yeah, I'm a designer. Charlie Cheever is the CEO and there's also David Cole, who's an incredible designer, co-founder and head of design of Castle. He was one of the people who brought me on. Our mission is… well, there are a lot of interpretations, but going back to like the cognition gym, ideally we're making people more cognitively active and capable. We don't want people to be cognitively scrawny. We want people to create games and ones that are both enriching and thought-provoking and really speak to the cultural zeitgeist.

The team is made up of David Luka, he is another person who actually started working on Castle primarily through making games on it. He was actually recruited because he was making incredible games on Castle and eventually, we ended up bringing him on and then we have a few other programmers who work on building the engine, work on building our social systems.

What's also compelling about the design team on Castle is that we all have chops in terms of how to program. So a lot of us, and when I say a lot of us, I mean all of us do directly contribute to the code base to some extent, while also making content in Castle and designing them as well and tools like Figma and whatnot.

What kinds of games give the best Castle experience in your opinion?

So right now on Castle, there are millions of games and people have the tools and autonomy to really make anything. And what's interesting is that the type of content that you see on Castle usually equates to social trends or just trends in gaming. You'll see a lot of FnF games like Friday Night Funkin’ and Flappy Bird-esque games. We have different kits that you can use as a starting point to make different games. So we have a roguelike kit, we have a golf kit, etc. Those types of games are pretty popular on the platform.

So we've created those kits as a way to kind of see the type of content that we want to see. But we also kind of instilled this genre system as a way to create different segmentations of interest groups on the platform. So we have a genre for people who are interested in creating art, we have a genre for people who are interested in creating horror games and so forth. We use that as a way to see the types of content that we want to see. But I ultimately think that personalization is super important because we actually ran this experiment where we were like, “Oh, we know the types of games that are best for our community. So let's make sure those games are the first ones that you see as you swipe through the feed.” And there are three canned ones or programmed ones that we added, those are hard-coded. But after those, we gave people these games that we thought were great. And very quickly we actually saw that it didn't do too well with our retention, which made us realize that a lot of our community does want to see these games that are more like “brain rot”.

I don't know how familiar you all are with that term, but it was like the Oxford Dictionary's Word of the Year last year. For 2023 it was “rizz”. But you can kind of qualify it as this low-effort kind of content that is popular amongst Gen Z and Gen Alpha that is related to meme ticks. But really it's something that I see as a reflection of internet culture today. So I think on Castle you'll see a mix of that type of content. Then quote, unquote, the “good stuff” that I would say is as someone who makes games or is looking for certain types of games, you'll be able to find that puzzle platformer, roguelike or golf game and so forth.

What are the current statistics between users and creators?

There are more players than there are creators, but creators, in terms of time spent, are dominating. People love creating on Castle and there are more systems that we're thinking about that will turn players into creators through collaboration. And I think I mentioned we have this function that allows you to remix decks or games that people create, which makes it easier for people to kind of get into that creator mode within Castle.

Can you tell us more about Castle’s monetization?

This year we've been really focused on first making the creator experience one that is quite potent, one that allows people to create compelling games. Over the last few months and going into this year, we've been focusing on the player experience a bit more. But right now you can buy these things called Bricks, and they are like the in-game currency of Castle and you can use bricks to boost your deck so more people can see it, but you can also gift Bricks to other people. You can imagine in the future people will be able to use Bricks to buy things in the editor. So you can like if there's game objects that you can use or scripts that you can use or sound that you need to use for your game, you'll be able to use Bricks to buy those things and you can imagine them used in a lot of other related kind of contexts.

How and why did you create couples.game?

We prototyped via pen and paper and I think a lot of people equate prototyping with software, but sometimes you can just start making something concrete without the use of computers. We were on the way to Jamaica for a friend's wedding and it was a friend that I used to work with at Oculus. We were on the plane and my partner showed me this trend that was kind of happening on TikTok, these word chains, but it actually originated with the game show Chain Reaction. And we spent the whole flight just kind of writing them down and really tweaking the mechanics to figure out, “Okay, should it be like four words, five words, six words, seven words?” Seven felt too hard, but five was too easy and six created this dynamic where the third and fourth words kind of felt like stragglers that you had to work towards.

And then we were prototyping this live system where we would pass the paper back and forth, and she would say, “Oh, you got it wrong.” And then, like, fill out the letters that were right and then say, “Oh, but you lost a life. Like, you didn't get the full puzzle right.” And, yeah, we played on the flight to the wedding and on the flight back, and immediately after, I was like, “Okay, I want to quickly build this so we can both play.” And when I sent it to her, I think she was pretty impressed at how quickly I was able to make it. Also with the caveat that I did use Cursor a bit, which is a tool, or IDE with an LLM inside, which really expedited the process of building this game.

And I think that we are in an era where people will be using these types of tools to make games of this nature more often. But she felt relatively good about the way it felt. I actually added a point system that she was originally talking about, But I sent it to a few other people shortly after, and they didn't fully understand the point system. So we ended up taking that out and just focusing on having three lives, and the three lives equated to three missing couples or pairs of words that you're trying to fill out.

And after that, we continued to text to people. It was one of the first things that I worked on where my mom was really excited about playing, and she was sending emails about her score, and other people were also doing it. She plays couples.game often. And other people would be like, “Oh, I sent this to my grandma!” And it was just a really good sign that it was something that was worth sharing more broadly.

So I tweeted about it, and I also realized that on TikTok and YouTube, some people started posting about it daily. And I'm sure maybe you all are familiar with this, but the idea of word talk, which is kind of this community of people who just share daily games on TikTok. But there was someone named Brian who was posting about it pretty consistently, and I joined his Discord and saw that people were frequently sharing their scores, which was really incredible to see. And yeah, it's something that I realized in retrospect, given this trend on TikTok, it made sense that people within this community on TikTok would really eat this up.

It even made its way to Elizabeth S, who was kind of the person who coined the emoji grid that led to Wordle going viral that Josh Wardle implemented into his game. So it's also cool seeing her play it. She even posted on Bluesky saying, “If I was paid for every time someone uses the emoji grid, I'd be rich.” Which is definitely true. And I wish she could be paid for it, but I'm also okay with her playing and continuing to evangelize the game.

How do you stock up on puzzles to put out?

So I've been thinking a lot about the idea of a new newspaper game. Newspaper games are just things that people would play in the newspaper and I think of a new one as one that exists online like a daily game. And I think that a new newspaper game should ideally be playable by everyone, including the creator. And that now could be very possible in most contexts through LLMs. And Wordle is one example where he has a massive list of words and it's impossible for him to anticipate what the word will be given the gameplay. But that is not true for a lot of games. 

So for couples.game, to make it so I could play, we started by making a word list that I manually inputted. If you've seen the word list, you probably know what the answer will be. So what I ended up doing was using an LLM to create a document and list that I would not look at and it would generate hundreds of these. So that's allowed me to play the game in perpetuity and it's something that I think we'll see with crosswords. I also tweeted recently about how we're talking about AGI, but LLMs still can't create a good crossword puzzle. It can solve them, but it still can't create them, given there's so much nuance with crossword puzzles. But I think we will get to that point and maybe people have different opinions about this. I am excited for that point.

But yeah, I think these sorts of games should be playable by the creator. I think that they should be fun alone but better together. For example, local multiplayer is something that is not as prevalent now, given console games are not as popular because people are playing on their phones. But the fact that at my Thanksgiving table and everyone is looking at the same puzzle and kind of guessing what it will be, that same behavior is on TikTok where someone will be playing couples.game and people are in the comments guessing what the words will be. That to me is also an example of local multiplayer in an asynchronous sense.

Couple.game’s intended design and flow

What I did was I came up with a ton of them and just kept giving them examples. So maybe what I gave influenced the pattern you’d grasp instinctually over time. But I will say that it was able to create some very nuanced versions. I will say that people have also given feedback that actually literally yesterday someone sent an email about puzzle #65, in which the answer was questionable and it certainly was.

So from that I did have to go back and just kind of double-check that everything was okay. But for the most part, I think it's been pretty smooth. And there are new models that are coming out that are more intelligent and I think it'll just make it easier for us to create more puzzles of this sort. I'll also say that we're getting to a point where people have also mentioned that you can guess both compound words and phrases. And that's also been a tricky one with getting the rules correct, where ideally you have one singular goal and the answer to the puzzle is related to that goal. And having both compound words and phrases makes that a bit tricky. But I do think that it adds a little bit more nuance to the game. It's a controversial take, but we have gotten feedback around that too.

How did you market the game?

A lot of it has just been pretty organic. Posting it on socials, texting it to people, Bluesky. Also, X's going through all sorts of things and I would love to spend more time on Bluesky. But I’ve been on X for years, so I have more of a network there obviously. However, Bluesky seems to be the place where more people actually like sharing couples.game and other word games. So for any word game creators, I would highly recommend checking out Bluesky because there's a pretty early community who are really interested in this sort of stuff. And when I say early community, I mean it's the same behaviors you see on X, but they're moving over to Bluesky, which is great to see.

Thoughts about creating an official community

I thought about setting up a Discord at one point, but it felt too mature to do for what it was. It was very much because we even had agents reach out saying, “Hey, we love to create content around this game. We'd love to talk to your team.” And I guess if you call me and my partner a team, then that's our team. But Discord felt like it was in another realm. I also noticed that Brian, one of the people who was posting the game, has this Discord called Brian's Beach Party. And I noticed that a lot of people will share the game results in that Discord.

So kind of like there was something interesting about that dynamic already existing. And us creating our own Discord obviously will help us grow our community. We have an email signup kind of thing. So eventually, if we get to the point where we want to do something like that, then we can. But there’s something just fun about Brian having his Discord where everyone is playing and I'm just a part of that. And hopefully, there are others like it. There's one other Discord I found too, but it's not as active as Brian. So there's something fun about just seeing it exist on other networks instead of really focusing on having our own. Not saying that we don't want to do that, as I think that could happen eventually, but we just want to take baby steps.

What are the current traffic statistics of the game?

I’ve set up basic Google Analytics and right now it's in the thousands. But I haven't really checked it as deeply. I've been pretty interested in creating more daily games that kind of link to couples.game, if that makes sense. I like the idea of looking at existing communities of games that people are playing and kind of creating remixes of those. So I've been experimenting with that, but also kind of tapping into platforms that increase the distribution of games like this. So looking at things like Discord activities, I really love the work that people like Echo Chess have done. And then even just like building it within other platforms that are related will help with broadening the network around it. So with Castle, we've been thinking a lot about how to enable daily games in it.

Do you plan to monetize the game?

Definitely not thinking about monetization right now. I think it's something that if it felt right, then it would be abundantly clearer that we'd do something like that. But I think we’re just more interested in making fun games that our community can play. And I'm also afraid about focusing on monetization, kind of sacrificing some of the quality of what the game offers. Right now it feels like something that you go to a lot of game sites and it's just littered with ads and we definitely don't want to be at that point. So as I was kind of talking about before, we are interested in making it something that can kind of exist more on your phone and use some of the levers that phone systems have, like notifications and social.

What are your future plans?

I've been thinking a lot about what makes sense in the future sense. I'm very interested in making more games that connect to couples. However, it's hard to figure out, like, are people just really interested in couples.game or are they interested in more games that are connected to it? But with that said, I don't want it to feel like more than it already is. I'm very interested in improving the system and community that we have around couples.game right now. There are very clear things that we can do. I'm actually pretty interested in how it could maybe be more social, maybe even be on mobile, like an app or something. Because then people have said, “Oh, I want to be able to know if my friend played without having to message them.” Like, that's a pretty interesting asynchronous use case. There are other things like leaderboards and I think a lot of games in this space don't take advantage of pretty basic mobile tech that is around. So just given some of my past experiences, I think there's a lot of opportunity to maybe just focus on improving couples.game in those contexts and maybe add additional games. But we'll see.

Can you give a brief rundown of your process for non-designers out there who want to follow your path?

Well, what I’m about to say is I feel like is specifically for word games, daily word games. I think for other types of games this workflow will probably be different. But for daily kind of word games, here’s what I’d say:

1. Starting with paper and pencil is a great idea. Just play with someone else and see if there's anything interesting or fun in the mechanic that you're thinking about. And then if it was maybe two or three years ago, I probably would have said to start using Figma and mocking stuff up. But now realistically, I’d just say:

2. Try starting with good tools. I just immediately started working in Cursor and then tried to get the basic mechanics that I have on pen and paper working. Then design in Figma, tweaked certain things because it's not easy to make small adjustments. I mean, you can in tools like Cursor, but it's a little bit faster to just do it in Figma. So I'm kind of going back and forth between Figma and Cursor to tweak the game and then what I do then is I deploy the game to Vercel. Vercel is really great because it's instant hosting and they also have analytics that you can embed in your game, kind of like Google Analytics. And you can buy a domain on Vercel and connect it and everything just kind of works smoothly. They also have Next.js, so if you want to have something stored in a server, then it handles that pretty quickly for you as well. And once it's deployed on Vercel, then often I'll just start:

3. Try the game out with a test group. I have a bunch of friends who also are working on all sorts of side projects and whatnot, and I'll end up just texting them and asking them if it's of interest at all.

That's generally my workflow. I think we're now at a stage where we can focus a lot more on polish because it's very fast to make games like this, especially things that don't require backends as much. And it's important to just get it into people's hands. Like pen and paper is one way to start to figure out if there's anything interesting. And then the second you have something that's playable, quickly give it to people so they can try the game too. I've noticed too, in that workflow, like I'll ask people not to share the game and then, you know, they end up texting it to someone anyway.

Where can we find you to learn more about you and your projects?

I'm Osebo on all platforms, but the first character is a zero, so it’s like, 0sebo. Also, I've been trying to write a lot more, so I secretly, maybe not so secretly, started a blog and it's just Osebo’s Blog.