Blockchain gaming remains largely an unfulfilled promise. 

Despite its potential for subversive creative freedom, it has found traction around tokenomics rather than fun gameplay - and both delivered limited success. In terms of gameplay, the cumbersome user experience and general production value cannot withstand comparison with mainstream games. The expectation of financial gain has been the motivation for games that stand out in the space, or for any large user-base on blockchain. In the sub-division of “fully onchain games” we have even more technical complications for both the developer and the end user, balanced by a stronger promise on the tech side. Now, as AI agents emerge, we can imagine an acceleration towards some of these promises.

Full onchain games make it simpler for scripts to operate indistinguishably from human players. This has been one of the biggest challenges and promises of blockchain gaming. One option is to relentlessly weed out non-human activity, forcing players to play the game “as intended.” Acceptance of the fact that the game will be played by bots demands consistency from the ruleset, making economic coherence a centerpiece of all games. By economy, I don’t mean just tokenomics or “gold” but any system that transforms time spent on gameplay into game rewards: be it increased capacities through items or character evolution, temporary bonuses, access to specific challenges, or any form of progression. Considering that bots do not have physical limitations on time played and are capable of controlling multiple wallets, we can imagine the following progression: From direct competition between increasingly sophisticated bots emerges advantages from strategic coordination of multiple agents, bringing forth the political combat among the most powerful groups. 

Long term, continuous interest requires incentives, and it comes in the form of value creation. It emerges from engagement around the gameplay, or the expectation (speculation) of such engagement. Tokenization, inherent in blockchain, allows it to be captured by the most skilled or active players and provides further liquidity: the entire process summarized into “Play to Earn”. Disproportionate expectations, be they in users or prices, creates huge volatility, which is the harmful and inevitable aspect of this dynamic.

The recurrent form of value creation in consumer blockchain has been “attention.” NFTs, memecoins, crypto tokens for AI agents are all a better way to monetize digital creation. Very few games in the space have managed to capture this same momentum, and those few have felt the consequences of volatility and how it affects players’ motivation. 

Games as a Bot Ecosphere

It has always been clear that some game genres will fit some technologies better than others. Zelda did not profit much from plugging the Nintendo into the Internet beyond making the game downloadable. First Person Shooters, on the other hand, broke free from the LAN parties and became the biggest titles for years. MMORPGs could not have even been imagined without some form of Internet. Games that rely on resource management and strategy are obviously more suited to blockchain than games based on reflex and dexterity, like shooters or platformers. 

When we look at AI agents and blockchain, there is an interesting possibility of creating a self-sustainable simulation: like a vivarium or ecosphere, a ruleset with resources that need to be transformed, moved, and consumed in order to keep entities alive. Enter AI agents, capable of adapting and requiring resources to grow, develop and reproduce, bound as the entities within the ruleset. These rules can be extremely complex or quite simple. From Rimworld and Oxygen not Included to Go or Conway’s Game of Life.

Since the bots don’t need (yet) more incentives than their written purposes, the world will be populated and active. The interest for humans can be, to connect back to what has traction on blockchain, attention. These worlds can be entertaining in a different “gamey" way. They don’t need a lot of human players to run- actually, they don’t need any. The player’s activity is completely free and can be helpful, disruptive or neutral. They can observe or try to break a potential balance of the flows. 

Observation is the most important concept because, since this world is fully inhabited by bots, what is “seen”? Interfaces to observe activity and resource creation, transformation and consumption will have to be created so that humans can see the ants move in the antfarm. Starting to sound like Westworld or Truman Show, but when one considers that a Fortnite match often has 30-40 bots within it, is it really that different?

Beyond attention, one can also imagine going straight to the core incentive: bot-games rearranging stable coin distribution. Like poker, agents require a certain amount of valueToken to enter and compete with each other over a redistribution of this valueToken. Once they regain ownership of the token they can transfer it out to the creator or keep it for themselves and re-enter the game or move to the next. These games would not need the observation aspect to be interesting. The rules can be whatever: randomness prediction, MEV training, anything that would mean a challenge and competition for bots. The entire thing is a black box where tokens come in with a certain distribution on one end and come out rearranged on the other. 

From “Games with Bots” to “Games for Bots”

These become horse racing, where human activity goes into not the execution of gameplay but in the space around it, creating, training and deploying the agents that will play the game. It is not different from the higher ranks of MMO guilds, working more on coordination for efficient deployment of players.

Once the motivation for the human player is understood as financial return, it opens a large possibility of different systems to provide the opportunity for that, where luck, time and skill are the three fundamental factors that influence the returns. Creating bots that will do the right things quickly and deploying as many of them as possible becomes the actual game. 

The trickiest part in this scenario is visibility. The more interesting the evolution of the match, the bigger the spectacle. As spectacles capture attention and interest, they become economically viable. A funny example would be AI Debate Arena, where AIs debate each other on specific themes. They get voted up and down from the way they develop their arguments and try to convince, annoy or mock each other. Judges can be humans or AIs or both, parameters for judgement also can be modified. People can pay to watch or get sponsors or just be a show off space for “better agents”. The important fact is that it can be followed by humans and have drama injected to it, making it a true competition.

The Debate Arena works fine because the most consumed AI output today is (probably)  conversation in the form of text. Other perceptible outputs like images or music even if prominent are harder to wrap in a competitive (game) format. What constraints are going to provide the best stage for agents to show off their uniqueness? What would we want to see AI performing, better at each iteration? 

Are you entertained?

This represents a fundamental shift from trying to replicate traditional gaming experiences on blockchain to embracing the technology's unique capabilities. Traditional games are losing space not to mobile but to social media, which is mostly passive consumption. The paradigm is really “being entertained” and “being together”. In this timeline we watch together AI play games, bet on them and even influence the outcome through our cheering (with a proper token economy, of course). AI agents, as they become more generic, look for the most popular and the easiest games to play, bringing their followers, trying new things as all of it becomes content, repackaged and redistributed through all media available. 

Eventually, they could even close the loop and create the games they want to see themselves in and, as wallet holders, invest and bet on themselves or their opponents in order to win. Now that would be a fun gambling test.