Panel Recap: What’s next to watch in On-Chain Gaming? | ETHShanghai2023

Mask Network
14 min readJul 21, 2023

At the ETHShanghai2023 summit, we invited guests to discuss "what's next to watch in on-chain gaming." Guests include Andy, core dev of MUD, Loaf, core contributor of Realm & Dojo, Tarrence, core contributor of Cartridge & Dojo, Guiltygyoza, co-founder of Topology, and Yijia, co-founder of Curio.

Host: Each new Noble computing paradigm and technology expands the horizon for gaming development.

In 2020, there was a radical experimental game called Dark Forest that leveraged zero knowledge in sync with on-chain gaming mechanisms. This led to the birth of some of the very first on-chain gaming with incomplete information, which is also openly extendable.

In 2021, we saw the rise of the Loot project, a mind game that rapidly became a cornerstone of a base-layered IP. It inspired countless games and derivatives to incorporate its gameplay and allure in a permissionless manner. We also saw a flourishing community grow out of the Loot project. Some of my colleagues, such as Taylor and Casper, were deeply involved in the Loot project from day one. They wrote an amazing 10,000-word analysis about fully Unchained gaming. Fast forward to this year, 2023, and we are witnessing a burgeoning industry of fully On-chain games and autonomous worlds.

We are excited to have core developers and contributors from various projects who will share their unique insights and perspectives on this topic.


  • Andy, Core dev of MUD
  • Loaf, Core contributor of Realm & Dojo
  • Tarrence, Core contributor of Cartridge & Dojo
  • Guiltygyoza, Co-founder of Topology
  • Yijia, Co-founder of Curio

Q: What you’re working out and what excites you most about this fairly new industry?

Guiltygyoza: I’m the co-founder of Topology. we are working on a fully on-chain game that borrows inspiration from Street Fighter, Tekken, and so on. Hopefully, we’ll bring it out to the public by uh July 21st Paris.

Terence: I am one of the co-founders of Cartridge and a core contributor to Dojo. I got involved in the on-chain gaming space about two years ago, and start building a fully on-chain game on top of Starknet today. Past year I primarily focused on building out Dojo which is an engine for on-chain games on top of Starknet.

Yijia: I’m a co-founder of Curio. we are building revolutionary game crypto native name mechanisms based on crypto native infra that we are both leveraging the existing as well as building our custom gaming-specific modules. We’re building our first game it’s a strategy game borrowing inspiration from typical strategy games like Civil Life, and Age of Empires but it’s also wildly different. It incorporates the social building blocks that we named Treaty that we tested earlier this year and we can all look forward to it.

Loaf: My journey started in on-chain gaming pretty much a week after Loot launched when I’m one of the core contributors of Bibliotheca DAO and the dojo engine. We tried to build a mini strategy game on Arbitrum at first then but then we quickly realized that we wanted a bit more compute and bigger design space so we quickly shifted to Starknet. we’re kind of experimenting with a bunch of different games right now from Grand Evergreen strategy games that are going to persist forever and more shorter ephemeral-style roguelike games. My day is consumed with experimenting with these games and experimenting with the dojo Engine with Terence

Andy: I mainly come from game development actually so I was like crypto pilled by Dark Forest and it was like my first time writing smart contracts was Dark Forest actually so very strange introduction to crypto like I could not write a standard web3 contract I’ve only written games so it is fun interacting with people and I don’t know what they’re talking about.

Q: I think it’s really hard to not discuss the exception of an on-chain game without talking about Dark Forest from a very thin idea into this radical experiment to explore blockchain scalability and usability problems. Questions for Andy, can you walk us through this Evolution journey and maybe offer us some insight into the transition from Dark Forest as a conceptual testing ground?

Andy: I joined Dark Forest in late 2021 and helped run their last round. The founder of Lattice (where I work as a core Dev) had already left Dark Forest to work on their own project, and they had forked the Dark Forest code base to start their own on-chain game. We both came to the same conclusion: game design ideas take too long to come to fruition, and the feedback cycle is not fast enough. After shipping the last round of Dark Forest, we were exhausted. We had a good engagement, and even the DAOs building on top of it had full-time engineers. We needed a better way, and that’s when Ludens called me to tell me about Mud(an on-chain game engine) I had a prototype of an on-chain game in two weeks using Mud before that it take months to add some small features to Dark Forest. From then on, all I’ve been doing is prototyping and building games. We want to create a complete vertical slice with hardware, software, and Dev tooling working seamlessly together. That’s the vision of Lattice and where we want to take on-chain gaming.

Q: My next question is for the Starknet folks, can you give us some insights into why do you decide to choose Starknet as your blockchain platform I’m really curious about your team working dynamic.

Loaf: We started building Realms, the OG game, in 2021. It was incredibly difficult to build because there were so many different parts to it we need to build it from the ground up, and they all had to work seamlessly together and it took forever and so that’s kind of what we did originally. It did work but it was clunky and painful. The blockchain presented issues with latency. We eventually found Cairo VM, an alternate VM that is still compatible with EVM but has more design scope. We created a group in October to collaborate with other Stark net teams and build a generalized engine in Cairo called Dojo which is now evolving. We are trying to hack things altogether.

Tarrence: I’ll cover the specific properties of Starknet that make it particularly interesting for on-chain games. My journey with Starknet began when I discovered the roll-your-own project, built on top of Starknet by Parama. The roll-your-own engine was the first attempt at building a permissionly extensible on-chained universe where people could easily incorporate new mechanics into the games. It opened my eyes to the power of Starknet, even though it was too ambitious for its time. Now, Starknet is just entering an inflection point where it becomes quite interesting for on-chain games. zk Roll-Ups enable exponential increases in the amount of computing that can be put on-chain. Recursive proofs allow you to almost infinitely increase compute into a constant-size proof that can be verified. This enables interesting architectures around scalability, called fractal scaling, and client-side proving that allow for introducing private information into games easily. Starknet uses State diffs to checkpoint state to the layers below it, which allows for temporal compression. We have some cool stuff coming soon, such as volition support, which is going to be great for games. It is a way to choose different data availability solutions for your game based on the properties that you want. There is also a project coming soon that enables the native execution of programs driven in Cairo, which will allow for executing Cairo logic or smart contracts at a similar performance to native languages like Rust. These are exciting developments in the space and one of the reasons why we think Starknet is a great platform for on-chain games.

Guiltygyoza: Topology has been primarily focused on game design, which is why we haven’t contributed as much to the Dojo project. Recently, we’ve been exploring game design patterns and constraints, which have helped lower the barrier for people to create blockchain-based games and applications. We’re now moving into the exciting phase of designing and iterating games with the toolsets we’ve created.

Since October 2021, we’ve been running physics experiments to create rigid body collisions and detect overlaps. We’ve also created tech demos with 2D rigid body physics simulations, such as a billiard board game and a planetary simulation game called Isaac. We’ve faced challenges with project complexity and TPS issues, but we’ve made significant progress.

Starknet is a unique space that allows unorthodox ideas to develop without peer pressure. It’s a great platform for fully on-chain games and compute-intensive applications. While it competes with other layer twos and roll-ups on TPS, it can find a niche in the market for verifiable games.

Q: I want to ask about Yijia and your project Curio. How do you differentiate your approach from Dojo and MUD? What is your growth strategy?

Yijia: I agree with Guiltygyoza. Game design is the foundation of the game’s engine and infrastructure. But currently, our space has not seen a powerful game. Dark Forest was exciting, and it motivated my co-founder and me to start in late 2021. The game was community-driven, and even mod-heavy games like EVE Online didn’t quite match in terms of ratio.

Our interest in Dark Forest was primarily due to its community engagement, where players were invested in shaping the virtual worlds they played in. Despite there being over 55 plugins at its peak, serving over a thousand players, we discovered that most of these plugins were on the front end and client, which is not unique to online games. However, we realized that on-chain games are possible with Dark Forest, especially in the MMO strategy genre.

Next, we plan to empower players to shape the game’s mechanics and aesthetics. This concept aligns with the idea of an autonomous world, where players can shape the game’s fundamentals. Our first product, launched in March or April 2022, is a virtual world where each item is like a 2D Minecraft. Every block is a smart contract that performs a specific function. For instance, you can create a door that only whitelisted players can enter. However, we must maintain a balance between the level of freedom we allow players and the game’s core, including the narrative and shared experience. Consequently, we created a shared ideology and narrative for the world. In July, we built our first MMO strategy game. It resembles the old Empire game, where you have to capture castles with your ships and land. Within three weeks, we had more than 100000 transactions, and we continue to focus on delivering what players want.

To create an on-chain game, a structured back end and shared data layer are needed. We built our own ECS to cater to the game we wanted to make. We added crypto-native gameplay components to make it more engaging. Instead of physical composability, we allowed players to create agreements enforced by the blockchain and game. We created a smart contract called Treaty which allowed players to sign and create their own contracts. People seem to enjoy creating their own agreements, which we didn’t have to consider when building the game.

There was one leading Alliance towards the end. But the night before the game ended, a player borrowed troops from a non-allied large player to raid the center for a Battle Royal strategy. This Treaty mechanism changed the game significantly. It’s a unique feature that is not present in other strategy games like Web2, Civilization, and Paradox games. We’re all looking for crypto-native mechanisms that can drive mass adoption.

After Treaty, we realized that blockchain couldn’t support game tick natively, causing long wait times for battle actions. This led to a performance bottleneck, requiring a block for repeating actions. While focused on smart contracts, we’ve taken a radical approach by adding game-specific modules to the EVM to OP stack directly, including an insurance indexer and precompiles that offload computing from on-chain to in-chain. With Keystone, released last month, we can solve the next big problem of on-chain games, which is mechanisms are interesting but too few players online at the same time with long transaction wait time. This is crucial for most games, unlike Dark Forest, where game actions take 15 minutes to several hours.

Q: The concept of the autonomous world encapsulates some of the essences of blockchain technology and uncovers the illusionary nature of freedom in traditional Web2 games. It is a powerful philosophical concept that resonated with me in the same way as the term Metaverse. However, Metaverse has become a meaningless marketing term over time. Can you explain what the autonomous world means in your career? Do you see it as a marketing or philosophical term? How can we prevent it from becoming an overused term like Metaverse?

Andy: I believe all of us on the call here share a vision of an autonomous world. However, the term is becoming overused as a marketing term and losing its meaning. We’re all taking baby steps towards this dream by experimenting and launching new projects. Dark Forest had some properties of an autonomous world, but it had a whitelist that limited its scale. We are not just raising money and doing research, but also doing playtests and seeking inspiration from other industries.

Personally, as a gamer, I get bored if I’m not actively playing a game. At Lattice, we play tests every Friday and I play the game every day with my brother, the lead game designer. We talk about the future we want to create based on our experience playing online games. Our dream is to create a game that we can play forever and that won’t get shut down. We believe there are many people who share this vision.

Old-school RuneScape is a perfect example. It came out in 2004, and everyone loved it. However, the game eventually shut down as the developers wanted to focus on a more profitable direction. This caused an uproar in the community, and players stopped playing. Eventually, the developers caved and re-released the old game alongside a new version. This is a rare example of a company listening to its community. We want to bring this essence to the forefront and give people a platform to explore their creativity in this way. Ultimately, we want to create a game that people will play and enjoy.

Yijia: I agree with Andy but want to emphasize that the underlying notion of autonomous worlds is the community that drives the game forward, and that’s something that Web3 games, and especially fully on-chain games have the ability to do. Achieving this goal may involve governance or allowing players to turn parameters in the game and launch mods that are still part of the larger community. However, some people pursue the notion of persistence without considering what players actually need. Instead, we should focus on creating social contracts and useful content to engage players. While persistence is important, game companies need to release new content to retain players. We should ask ourselves what players really need and focus on building good products rather than just using vague marketing terms like “autonomous worlds.”

Guiltygyoza: I don’t have much else to add, I also say I don’t think we can prevent hyping and minification from happening. But I think as a builder, we just have to ship. So let’s build something fun.

Tarrence: The term “autonomous world” should not be used interchangeably with a fully on-chain game. It is a subset of fully on-chain games with specific properties. Persistence alone does not make an interesting game. What’s important is the ability for any participant to propagate properties over time in an autonomous world. The expansion should be permissionless and organic, starting from a fundamental award-driven foundation. Interesting games and mechanics will emerge on top of that foundation, serving as a basis for experimentation.

I think there’s probably nothing we can do to prevent people from using it as a marketing term, it’s the sad reality of doing interesting things. I’m sure we’ll see it abused a lot in the future.

Loaf: I agree that there may not be much we can do about it. To me, autonomous worlds are similar to L1 chains in that they require a high level of persistence. If you want to interact with this world, you need to ensure that the code will run indefinitely. We are all building on Ethereum, which is designed to execute code no matter what, and thousands of nodes back it up. Therefore, I think autonomous worlds need to have this level of persistence at a fundamental level. We are experimenting with different design cycles for our game, but we have found that persistent games are inherently complex. To deal with this complexity, we start at a primitive layer with basic rules and let an equilibrium form within the world. From there, complexity can grow. The natural world also started in chaos and formed an equilibrium. A true autonomous world should have similar properties, starting with a minimal level of physics and expanding from there. This is an exciting idea, and it’s great to see everyone participating in this cool experiment every week. This is currently the most exciting part of crypto.

Q: I have two questions to ask. Firstly, on-chain gamers seem to have a deeper connection with autonomous world games due to intensive ownership of in-game assets and achievements as well as motivation to master the game to secure high-value assets and prestigious titles. What’s the one thing you have observed from gamers that meets your expectations? Secondly, how do you see AI’s role in the autonomous world in the future?

Yijia: For the first question. We were pleasantly surprised when we launched Treaty with smart contract components. The success of this feature boosted our confidence in making Treaty and social building blocks a core part of our games, rather than just an add-on. We will continue to explore new crypto-native interactions and dynamics.

And for the second question.AI has many potential use cases, such as reducing the barrier to creating user-generated content in genetic content-based applications or games like Zelda. Most players don’t build complicated things in games, but AI can be used to make it easier to create content using natural language instead of writing code. When AI is integrated into existing popular products, it could help bring crypto to the mainstream.

Guiltygyoza: The game we’re making is a fighting game where you make agents and strategies and I think when the front end is open source some people will come in and integrate GPT4, but now we’re not providing any machine learning stuff. We’re just keeping things simple.

Andy: We created Opcraft, a Minecraft clone where each block built was a transaction and players could claim territory with diamonds. After a few days, someone mined 80,000 diamonds, claimed the entire world, and declared dictatorship over it. They launched a smart contract where players had to pledge allegiance in order to play. This was an unintended but inspiring moment for us. What will happen when the game becomes more permanent? It leaves room for imagination.

Loaf: Emerging behavior is the core of what we want to achieve. Games can surprise us with the unpredictable creations of players and developers. Zelda is the perfect example, with endless possibilities and no clear objective. This is the excitement we aim to evoke — building blocks and allowing something to emerge.

There is potential for designing agents to play games using AI, even before GPT. The generative AI technology is exciting as it lowers the barrier to content and client creation, providing a new interface for games. With everything on-chain, multiple clients can interact with the same game, making it more exciting.

Terence: One surprising thing is the open ecosystem that attracts many talented individuals who participate in the community and contribute to the project out of passion.

Many agree that on-chain games are the ideal platform for user-generated content. Combining AI and blockchain technology can make it even more accessible, with built-in payment rails and infrastructure. It’s a powerful intersection of two new and exciting technologies.