MatchboxDAO is a collective of developers, artists, and designers building the infrastructure of on-chain games with StarkNet.

This is the fourth article in MatchboxDAO’s series on “How to Make On-Chain Gaming Competitive”. A new article will be released Tuesday every fortnight and will cover different topics around how developers are shaping on-chain games to become more intuitive and entertaining for gamers. This can include new features, games, onboarding tools and just about anything that improves the on-chain experience in a gaming context.

It’s part of our ongoing efforts to reduce stigma around blockchain gaming and to show the benefits behind keeping these features on-chain in a trustless, accessible manner. We hope this series can drive adoption from new developers and gamers because the future of on-chain gaming will be collaborative. Join us each fortnight to learn where this space is heading and how we’re preserving integrity and most importantly, fun in gaming.

Introduction: “Living” Games

What does it mean for a game to be alive?

You know the feeling…

When a game feels tangible and consequence-rich. When it sees and reacts to you. When the game world feels like it keeps on going, even when you leave the keyboard or put down the controller for the night.

That world awaits you…

If you’re anything like me, and have spent hundreds if not thousands of hours in your favourite game worlds, you’ll know exactly what I mean. Whether it’s GTA or Minecraft, Pokémon or Zelda, all of these generational titles transform you into wanderers in a foreign land.

But how do they do it?

What Makes a Game?

After talking to some of the smartest game devs I know, I think I’ve found the answer:

All of these games are marvels of Immersion and Interactivity

From reactive worlds to inspiring mechanics, rich lore to clever characters, the best games are case studies in player immersion. And without exception, they all feature design choices that create a sense of consistency and fairness to their game worlds, as well as deep interactivity with the players.

• GTA boasts one of the most life-like, complex worlds in gaming. But it’s consistent physics and interactive AI NPCs allow for a truly high degree of player experimentation.

• Minecraft’s procedurally generated worlds perfectly balances the need for variety while capturing a sense of consistency in player interaction.

• The latest Zelda, finally, captures the delight of giving players a wide open world and the tools to play with remarkably robust in-game physics.

Today, however, we wanted to zero-in on the next great level-up to both player immersion and interactivity — Artificial Intelligence.

Beware: Living Agents in Your Game

To us, “Other Agency” is the single most effective implementation of both immersivity and reactivity.

Other Agency* — the feeling that other “intelligent” agents inhabit your game.*

When intelligent AI’s coexist in your game, players get the feeling that there is a complex and somewhat hard-to-predict “Other,” behind the game mechanics they’ve become accustomed to.

This elevates the degree to which your games “react” to you, as well as acts as a renewed commitment to the consistency of the game world by reinforcing the foundational rules. The more consistent your game worlds are, the more immersed your players are.

Titles like the ones we previously mentioned have all incorporated impressive AI agents, and importantly, these agents can both take on active (literal stand-ins for other players/agents) or passive (reactive, setting-changing) roles. In fact, the two are frequently paired for especially compelling encounters.

The good news is, thanks to rapidly advancing zk-prover technology, we’re now at the precipice of bringing that same upgrade — with a distinct web3 twist — to on-chain games…

Quick recap

The process of bringing your AI on-chain borrows heavily from zk-rollups themselves: Pump an AI model’s architecture and inputs through an efficient zk-proof system — off-chain — on-chain. This allows us to:

• Take advantage of the undeniable efficiency of centralised compute while…

• Forcing the central operator to surrender control of the model

When every use of a given model is verified, the consequence of this system is clear: the accessible use of verified AI inference, which is both trustless and powerful.

In other words, what Modulus Labs is researching is AI that’s fair.

Trustless AI for On-Chain Games

ZK-AI brings machine intelligence up to the web3 standard of security. Which is to say, AI can now be used in on-chain games without discarding the trustless ethos of the chain.

Trustless intelligence also adds an entirely new dimension of interactivity to your games that goes beyond the ways that AI already can.

Whether it’s chatbot generated dialogue or galactic governments controlled by AI agents, the players can now confirm that the models are used without tampering.

The implications of this are significant — especially when dollars are at stake.

An intriguing use-case is brought up by Loaf — imagine a massive in-game economy with player and AI merchants. Each merchant trades with another constantly, effectively forming an in-game AMM. In this world, AI’s trade with each other as well as players, always self-optimising and autonomous.

Fairness is paramount here, and when guaranteed, creates a whole new standard for total interactivity for human players (since the markets are always open).

But that’s just the beginning. With verifiable AI, game designers can now create entirely new forms of gameplay mechanics:

• Players can form alliances with AI agents that they’ve contributed to, either by architecture or input

• Or, as we’ll demonstrate with our POC, players can compete together against a wildly intelligent AI NPC

• And most tantalizing of all, AIs can now control and self-optimize important gameplay parameters, dynamically

Key to the paradigm of trustless intelligence, of course, is that this is absolutely unique to web3 games. This kind of verifiable intelligence has never before existed, and new ways of constructing games around fair AIs are sure to emerge.

Conclusion: The State of On-Chain Games

So what does the chain add to the greater gaming tapestry? How does the chain ultimately change a games’ ability to be immersive and interactive?

The good news is, the same kind of brilliant-game-making-traits we first identified are showing up here in web3 as well. In fact, they’ve already taken on a distinctive web3 flavour of their own.

• Ambitious game devs like the folks behind Realms are creating eternal games — ever-lasting games that rely on the chain to guarantee consistent global state. This kind of game world must be built on the exceedingly robust protections of a decentralized network.

• Games like “Isaac,” on the other hand, rely on player coordination at a truly massive scale (players must mine and deploy resources together to save their endangered planet). The chain, in this case, provides a foundation of known fairness (since global state is protected), while codifying each player interaction as substantive, irreversible decisions.

To us, trustless AI is the next logical level-up, a new tool that the zk paradigm uniquely enables. And like the clever designs of great games past, verifiable intelligence will both increase an on-chain games’ ability to immerse its players, as well as its awareness of the players.

Article written by:

Don’t miss upcoming articles: