Hello and welcome to Dark Tunnels, a newsletter dedicated to exploring the emerging ecosystem of fully onchain games. Dark Tunnels is published by Always Scheming, a research consultancy focused on the future of interactive entertainment.

The following essay is Issue #3 of Dark Tunnels. Enjoy!

Gaming has always pushed the limits of technology. Whether it was The Legend of Zelda’s use of non-volatile memory save states in NES cartridges, DOOM's real-time rendering and 3D graphics pushing the limits of available PC hardware, or Crysis setting a new standard for fidelity with its advanced visuals and physics simulations, gaming has a long and storied history of driving technical improvements while navigating creative constraints.

Though the early experiments in fully onchain gaming are a far cry from the titles mentioned above, they are still representative of this historical dynamic. The unique combination of low stakes and high technical complexity found in gaming makes these environments ripe for innovation.

I find the following quote to be particularly instructive in elucidating this dynamic:

“If we accept that games are a leading indicator of new technologies, and that the most interesting applications of new technologies will lean into truly new affordances rather than incremental improvements, then it follows that the bleeding edge of crypto application design in the next few years will be found in crypto-native games.”

Gubsheep, “The Strongest Crypto Gaming Thesis”

In other words, we should be paying attention to fully onchain games not because they are amazing interactive entertainment experiences today, but because they will incubate the breakthrough applications of the blockchain-enabled technologies of tomorrow.

Despite the recent momentum surrounding fully onchain gaming, decentralized games have actually been inextricably linked to the development of the blockchain from its earliest incarnations:

• The original 0.1.0 Bitcoin code contained snippets of a peer-to-peer poker game.

• Satoshi Dice, a “blockchain-based betting game,” has been operating on the Bitcoin network since 2012.

• Huntercoin, perhaps the first truly decentralized game, started as an “experimental prototype” in 2013, forked from Namecoin (a protocol that was itself forked from Bitcoin). [2]

Just as these early endeavors spurred new development and investment in blockchain tooling, wallet design, and scaling, so too have the fully onchain games of the last few years kicked off their own wave of interest and innovation. The most frequently-cited example of this is the seminal grand strategy game Dark Forest.

At its core, Dark Forest is a fully onchain massively multiplayer “4X” game of space conquest and exploration, similar to Galcon. Players start with a single planet in a shared universe, expanding outwards by claiming other planets, gathering resources, upgrading infrastructure, and attempting to dominate opposing players.

A hallmark of fully onchain games is that they hold all game logic and game state onchain for anyone to see. This implies that all players in a given instance of Dark Forest should be privy to complete information: every player, planet location, and move would be instantly revealed to all participants. However, through use of zero-knowledge proofs (ZKPs), Dark Forest enables elements of incomplete information, such as “fog of war” (represented in the image below by the gray areas of unexplored space). In fact, zero-knowledge technology is so ingrained in the DNA of Dark Forest that the game’s URL to this day is “zkga.me.”

Of course, fog of war isn’t exactly a new concept in game design, nor is it necessarily required for a game to be considered fully onchain. So why should game developers care about zero-knowledge proofs?

Put simply, a zero-knowledge proof is a means to cryptographically prove that a given statement is true without providing any additional information. Apart from revealing that the statement is true, the “prover” is able to verify certain hidden information without divulging any further details.

This allows for private information to occur off-chain, while keeping the proof onchain. In other words, I can verifiably prove that the actions I took off-chain did occur as I say they did, without revealing exactly what those actions were.

Through its use of ZKPs, the Dark Forest team has shown that incomplete information games can be run fully onchain. The implications for gaming are immense; ZKPs open up an entire category of games for use onchain.

Previous incarnations of fully onchain games had been limited to so-called “complete information” games, or those where all players know the full state of the game. A simple example of a complete information game is chess: all players can see the locations of all game pieces at any given time, the sequence of play is known by all, and the payoff functions are also common knowledge. Other examples of complete information games include checkers, go, or tic-tac-toe.

However, there are many instances where incomplete information is required to play the game. Any game with a fog of war utilizes incomplete information, as do many card games (poker, Magic: The Gathering), board games (Battleship, Guess Who), and countless others. The asymmetry of information in these games – where a given player or set of players may know more or less about the state of the game world than other players – allows for a richer set of strategies, trade-offs, and social dynamics.

ZK tech has many applications beyond gaming, too. For example, proving that a given wallet belongs to a human and not a bot without revealing that person’s identity, or perhaps cryptographically verifying some form of status (e.g. number of Twitter followers, professional credentials, location within a given legal jurisdiction, etc.). ZKPs could even be used to verify the integrity of machine learning applications, including AI models.

There is much to be said about this set of technologies, and one could easily fall down the rabbit hole of ZK techno-babble acronyms (SNARKs and STARKs and PLONKs and 

A similar story is set to play out in the field of account abstraction, where recent developments are quickly unlocking meaningful improvements to web3 UX. [4] Also known as “smart accounts”, account abstraction essentially changes basic Ethereum addresses into programmable smart contracts. Whereas a typical address relies on case-by-case approvals for each transaction, a smart account may contain all manner of predefined conditions and logic.

The most straightforward use case for this is in eliminating the need to sign every single transaction – an absolutely dreadful user experience for games without account abstraction. This is particularly important if web3 games are ever going to attract non-web3 users. Gamers don’t want to deal with wallets, and early leaders in “web2.5” gaming like Sorare have shown the power in removing wallets from the user onboarding flow.

Other use cases include defining preset spending limits, social recovery (in case of lost keys, theft, or death of the user), sponsored transactions (contracts pay gas fees on behalf of users), and custom access control (sub-accounts, transaction allowlists, etc.), among many others.

Early implementations of account abstraction have already made their way to onchain games, too. The technology is currently being developed most prominently within the Starknet ecosystem (Cartridge, Argent, Realms), but account abstraction is possible on all EVM-compatible chains. [5]

Account abstraction will go a long way towards addressing concerns around the accessibility of fully onchain games. As more games continue to adopt this standard and prove out the impacts of streamlined UX on web3 game adoption, a new precedent will be set for reimagining how user interactions can occur within decentralized applications. These advancements have the potential to reshape not only gaming experiences but also financial applications, identity management systems, decentralized governance models, and perhaps other areas yet unexplored.

If you’ve read the introduction to Always Scheming, you’ll know that I’m a big proponent of Chris Dixon’s The Next Big Thing Will Start Out Looking Like a Toy. In that piece, Dixon writes:

“To distinguish toys that are disruptive from toys that will remain just toys, you need to look at products as processes. Obviously, products get better inasmuch as the designer adds features, but this is a relatively weak force. Much more powerful are external forces: microchips getting cheaper, bandwidth becoming ubiquitous, mobile devices getting smarter, etc. For a product to be disruptive it needs to be designed to ride these changes up the utility curve.”

It should come as no surprise that I believe fully onchain games are similarly poised to benefit from these sorts of external forces. In many ways, that is the underlying thesis of this entire essay.

Even beyond the trends discussed above, one could point to improvements in blockchain scaling, the growth of community-led game development, or even the ubiquity of tools like Discord and Telegram as forces external to onchain gaming that will help to propel the format up the adoption curve. However, the one trend that may prove to be most impactful to the development of fully onchain gaming is the ongoing proliferation and rapid improvement of generative AI (“GenAI”) applications.

Many commentators (myself included) have written at length about the democratizing impacts that GenAI will have on the game development process. Teams will become more efficient, prototyping will accelerate, and processes that once took weeks or months will be compressed into a matter of hours. This applies to nearly all aspects of the games industry.

However, there are a few areas where fully onchain gaming is poised to capture potentially outsized benefits. For one, generative AI can help to ease the burden of content creation when bringing autonomous worlds to life. Given the vast design space available, as well as the open and community-led nature of their development, fully onchain games are well-suited to the use of GenAI. We have already seen this play out in projects like Realms: Adventurers, where the BibliothecaDAO team has built a custom AI model that allows players to create their own characters for use in onchain experiences.

Furthermore, the clientless nature of fully onchain gaming is a natural fit for the endless creativity enabled by GenAI tools. Players are free to bring their own art, UIs, and plugins to these worlds, and will be better equipped to create these assets for themselves through the use of generative AI. This enables near-limitless personalization of onchain experiences. While you might be a fan of the current intergalactic themes of Dark Forest, others might prefer a Wild West experience, a nautical interface, or a dystopian cyberpunk frontend.

Generative AI will also be particularly impactful in growing the pool of early builders, as non-technical players will be further empowered to participate through use of tools like ChatGPT, Replit’s Ghostwriter, or Github’s Copilot. By combining the infrastructure standards being developed within projects like MUD or 

I expect that this increase in accessibility will be further complemented by efforts from existing developers in the onchain gaming space to lower the barriers to entry for aspiring contributors. Low- and no-code solutions seem like a natural conclusion to this trend, and the early seeds of this can already be seen in initiatives like Curio’s templated Treaties or MatchboxDAO’s sample cars for 0xMonaco Season 2. Playmint’s forthcoming project, 

The gaming industry will always be at the forefront of innovation, continually pushing the limits of available technologies in service of creating novel interactive entertainment experiences. Fully onchain games are no exception.

Whether the broader games industry believes in the commercial prospects of onchain gaming or not is of little consequence. Builders need to build. Technical constraints always lead to creative solutions. Sometimes that leads to silly results, like Doom Eternal on a fridge, but other times it leads to 

The point here is that creators are going to create, regardless of the technical limitations. While the constraints of building maximally on a blockchain may seem cumbersome to many of today’s rank and file game developers, technologies such as those outlined above are poised to expand the boundaries of the onchain gaming design space in ways that will only lead to further innovation.

Thanks for reading.

1. This essay does not discuss the emerging group of fully onchain game engines. If you’d like to delve into the differences between MUD, Dojo, the Argus World Engine, and Curio’s Keystone, I highly recommend this deep dive. I also touched on these briefly in The Inevitability of the Format.

2. It’s worth noting that members of the Huntercoin team are still operating in the onchain gaming space today. They are now known as XAYA and have built their own blockchain, as well as games like Soccerverse and Taurion.

3. For a deep dive on all things zero-knowledge, I recommend exploring a16z Crypto’s Zero-Knowledge Canon.

4. “Account abstraction” can be a bit of a catch-all term covering a number of different use cases, some of which are outlined in this essay. For a deeper dive into the ins and outs of account abstraction, check out writing from

Special thanks to baz at Tonk, Brian at Pocket Gems, Martin at RockawayX, Neeraj at Moonstream, and everyone in the Decentralised Gaming Association Discord for the thoughtful feedback and discussion that helped to shape this piece.

Hello and welcome to Dark Tunnels, a newsletter dedicated to exploring the emerging ecosystem of fully onchain games. Dark Tunnels is published by Always Scheming, a research consultancy focused on the future of interactive entertainment.

The following essay is Issue #3 of Dark Tunnels. Enjoy!

Gaming has always pushed the limits of technology. Whether it was The Legend of Zelda’s use of non-volatile memory save states in NES cartridges, DOOM's real-time rendering and 3D graphics pushing the limits of available PC hardware, or Crysis setting a new standard for fidelity with its advanced visuals and physics simulations, gaming has a long and storied history of driving technical improvements while navigating creative constraints.

Though the early experiments in fully onchain gaming are a far cry from the titles mentioned above, they are still representative of this historical dynamic. The unique combination of low stakes and high technical complexity found in gaming makes these environments ripe for innovation.

I find the following quote to be particularly instructive in elucidating this dynamic:

“If we accept that games are a leading indicator of new technologies, and that the most interesting applications of new technologies will lean into truly new affordances rather than incremental improvements, then it follows that the bleeding edge of crypto application design in the next few years will be found in crypto-native games.”

Gubsheep, “The Strongest Crypto Gaming Thesis”

In other words, we should be paying attention to fully onchain games not because they are amazing interactive entertainment experiences today, but because they will incubate the breakthrough applications of the blockchain-enabled technologies of tomorrow.

Despite the recent momentum surrounding fully onchain gaming, decentralized games have actually been inextricably linked to the development of the blockchain from its earliest incarnations:

• The original 0.1.0 Bitcoin code contained snippets of a peer-to-peer poker game.

• Satoshi Dice, a “blockchain-based betting game,” has been operating on the Bitcoin network since 2012.

• Huntercoin, perhaps the first truly decentralized game, started as an “experimental prototype” in 2013, forked from Namecoin (a protocol that was itself forked from Bitcoin). [2]

Just as these early endeavors spurred new development and investment in blockchain tooling, wallet design, and scaling, so too have the fully onchain games of the last few years kicked off their own wave of interest and innovation. The most frequently-cited example of this is the seminal grand strategy game Dark Forest.

At its core, Dark Forest is a fully onchain massively multiplayer “4X” game of space conquest and exploration, similar to Galcon. Players start with a single planet in a shared universe, expanding outwards by claiming other planets, gathering resources, upgrading infrastructure, and attempting to dominate opposing players.

A hallmark of fully onchain games is that they hold all game logic and game state onchain for anyone to see. This implies that all players in a given instance of Dark Forest should be privy to complete information: every player, planet location, and move would be instantly revealed to all participants. However, through use of zero-knowledge proofs (ZKPs), Dark Forest enables elements of incomplete information, such as “fog of war” (represented in the image below by the gray areas of unexplored space). In fact, zero-knowledge technology is so ingrained in the DNA of Dark Forest that the game’s URL to this day is “zkga.me.”

Of course, fog of war isn’t exactly a new concept in game design, nor is it necessarily required for a game to be considered fully onchain. So why should game developers care about zero-knowledge proofs?

Put simply, a zero-knowledge proof is a means to cryptographically prove that a given statement is true without providing any additional information. Apart from revealing that the statement is true, the “prover” is able to verify certain hidden information without divulging any further details.

This allows for private information to occur off-chain, while keeping the proof onchain. In other words, I can verifiably prove that the actions I took off-chain did occur as I say they did, without revealing exactly what those actions were.

Through its use of ZKPs, the Dark Forest team has shown that incomplete information games can be run fully onchain. The implications for gaming are immense; ZKPs open up an entire category of games for use onchain.

Previous incarnations of fully onchain games had been limited to so-called “complete information” games, or those where all players know the full state of the game. A simple example of a complete information game is chess: all players can see the locations of all game pieces at any given time, the sequence of play is known by all, and the payoff functions are also common knowledge. Other examples of complete information games include checkers, go, or tic-tac-toe.

However, there are many instances where incomplete information is required to play the game. Any game with a fog of war utilizes incomplete information, as do many card games (poker, Magic: The Gathering), board games (Battleship, Guess Who), and countless others. The asymmetry of information in these games – where a given player or set of players may know more or less about the state of the game world than other players – allows for a richer set of strategies, trade-offs, and social dynamics.

ZK tech has many applications beyond gaming, too. For example, proving that a given wallet belongs to a human and not a bot without revealing that person’s identity, or perhaps cryptographically verifying some form of status (e.g. number of Twitter followers, professional credentials, location within a given legal jurisdiction, etc.). ZKPs could even be used to verify the integrity of machine learning applications, including AI models.

There is much to be said about this set of technologies, and one could easily fall down the rabbit hole of ZK techno-babble acronyms (SNARKs and STARKs and PLONKs and 

A similar story is set to play out in the field of account abstraction, where recent developments are quickly unlocking meaningful improvements to web3 UX. [4] Also known as “smart accounts”, account abstraction essentially changes basic Ethereum addresses into programmable smart contracts. Whereas a typical address relies on case-by-case approvals for each transaction, a smart account may contain all manner of predefined conditions and logic.

The most straightforward use case for this is in eliminating the need to sign every single transaction – an absolutely dreadful user experience for games without account abstraction. This is particularly important if web3 games are ever going to attract non-web3 users. Gamers don’t want to deal with wallets, and early leaders in “web2.5” gaming like Sorare have shown the power in removing wallets from the user onboarding flow.

Other use cases include defining preset spending limits, social recovery (in case of lost keys, theft, or death of the user), sponsored transactions (contracts pay gas fees on behalf of users), and custom access control (sub-accounts, transaction allowlists, etc.), among many others.

Early implementations of account abstraction have already made their way to onchain games, too. The technology is currently being developed most prominently within the Starknet ecosystem (Cartridge, Argent, Realms), but account abstraction is possible on all EVM-compatible chains. [5]

Account abstraction will go a long way towards addressing concerns around the accessibility of fully onchain games. As more games continue to adopt this standard and prove out the impacts of streamlined UX on web3 game adoption, a new precedent will be set for reimagining how user interactions can occur within decentralized applications. These advancements have the potential to reshape not only gaming experiences but also financial applications, identity management systems, decentralized governance models, and perhaps other areas yet unexplored.

If you’ve read the introduction to Always Scheming, you’ll know that I’m a big proponent of Chris Dixon’s The Next Big Thing Will Start Out Looking Like a Toy. In that piece, Dixon writes:

“To distinguish toys that are disruptive from toys that will remain just toys, you need to look at products as processes. Obviously, products get better inasmuch as the designer adds features, but this is a relatively weak force. Much more powerful are external forces: microchips getting cheaper, bandwidth becoming ubiquitous, mobile devices getting smarter, etc. For a product to be disruptive it needs to be designed to ride these changes up the utility curve.”

It should come as no surprise that I believe fully onchain games are similarly poised to benefit from these sorts of external forces. In many ways, that is the underlying thesis of this entire essay.

Even beyond the trends discussed above, one could point to improvements in blockchain scaling, the growth of community-led game development, or even the ubiquity of tools like Discord and Telegram as forces external to onchain gaming that will help to propel the format up the adoption curve. However, the one trend that may prove to be most impactful to the development of fully onchain gaming is the ongoing proliferation and rapid improvement of generative AI (“GenAI”) applications.

Many commentators (myself included) have written at length about the democratizing impacts that GenAI will have on the game development process. Teams will become more efficient, prototyping will accelerate, and processes that once took weeks or months will be compressed into a matter of hours. This applies to nearly all aspects of the games industry.

However, there are a few areas where fully onchain gaming is poised to capture potentially outsized benefits. For one, generative AI can help to ease the burden of content creation when bringing autonomous worlds to life. Given the vast design space available, as well as the open and community-led nature of their development, fully onchain games are well-suited to the use of GenAI. We have already seen this play out in projects like Realms: Adventurers, where the BibliothecaDAO team has built a custom AI model that allows players to create their own characters for use in onchain experiences.

Furthermore, the clientless nature of fully onchain gaming is a natural fit for the endless creativity enabled by GenAI tools. Players are free to bring their own art, UIs, and plugins to these worlds, and will be better equipped to create these assets for themselves through the use of generative AI. This enables near-limitless personalization of onchain experiences. While you might be a fan of the current intergalactic themes of Dark Forest, others might prefer a Wild West experience, a nautical interface, or a dystopian cyberpunk frontend.

Generative AI will also be particularly impactful in growing the pool of early builders, as non-technical players will be further empowered to participate through use of tools like ChatGPT, Replit’s Ghostwriter, or Github’s Copilot. By combining the infrastructure standards being developed within projects like MUD or 

I expect that this increase in accessibility will be further complemented by efforts from existing developers in the onchain gaming space to lower the barriers to entry for aspiring contributors. Low- and no-code solutions seem like a natural conclusion to this trend, and the early seeds of this can already be seen in initiatives like Curio’s templated Treaties or MatchboxDAO’s sample cars for 0xMonaco Season 2. Playmint’s forthcoming project, 

The gaming industry will always be at the forefront of innovation, continually pushing the limits of available technologies in service of creating novel interactive entertainment experiences. Fully onchain games are no exception.

Whether the broader games industry believes in the commercial prospects of onchain gaming or not is of little consequence. Builders need to build. Technical constraints always lead to creative solutions. Sometimes that leads to silly results, like Doom Eternal on a fridge, but other times it leads to 

The point here is that creators are going to create, regardless of the technical limitations. While the constraints of building maximally on a blockchain may seem cumbersome to many of today’s rank and file game developers, technologies such as those outlined above are poised to expand the boundaries of the onchain gaming design space in ways that will only lead to further innovation.

Thanks for reading.

1. This essay does not discuss the emerging group of fully onchain game engines. If you’d like to delve into the differences between MUD, Dojo, the Argus World Engine, and Curio’s Keystone, I highly recommend this deep dive. I also touched on these briefly in The Inevitability of the Format.

2. It’s worth noting that members of the Huntercoin team are still operating in the onchain gaming space today. They are now known as XAYA and have built their own blockchain, as well as games like Soccerverse and Taurion.

3. For a deep dive on all things zero-knowledge, I recommend exploring a16z Crypto’s Zero-Knowledge Canon.

4. “Account abstraction” can be a bit of a catch-all term covering a number of different use cases, some of which are outlined in this essay. For a deeper dive into the ins and outs of account abstraction, check out writing from

Special thanks to baz at Tonk, Brian at Pocket Gems, Martin at RockawayX, Neeraj at Moonstream, and everyone in the Decentralised Gaming Association Discord for the thoughtful feedback and discussion that helped to shape this piece.