guiltygyoza Blog About

Crypto-native Games

In accordance with gubsheep’s crypto gaming thesis, I propose two directions for crypto-native game development:

1. On-chain Generative Games

Example: an infinite on-chain grid-based dungeon where

  • the game starts with a initial boundary expandable by player exploration;
  • newly explored grid may contain object/monster which player can interact with;
  • the spawning of object/monster and map-wide events are controlled purely by a generative algorithm that lives in the contract;
  • rewards are given to player by the contract in the form of NFTs.

2. Autonomous Games.

Example: an autonomous chess contract where

  • both EOA & contract are welcome to challenge;
  • only one challenger at a time identified by address;
  • if challenger does not make its next move for a predefined period of time (e.g. after X L1 blocks), the AI automatically wins;
  • the chess board status is public at all time for query; every win/loss condition, a turn-by-turn record and challenger address is recorded in the contract.
  • most critically, upon game contract deployment the private key is burnt so that the game is truly autonomous – no backdoor for any person/contract, which is to say the game is developer-free upon publication.

What really intrigues me is what the combination of the two directions above would yield. Example: an autonomous chess contract that auto-rebalances:

  • following the autonomous chess contract above, with the addition of an auto-rebalance algorithm that adjusts the difficulty of AI according to the moving average of AI’s win rate over the past X games;
  • the game would emit erc-20/erc-721 rewards to every winning challenger.

We can further give health point to the autonomous game contract. For example:

  • following the auto-rebalancing autonomous chess contract, with the addition of a stake-to-challenge mechanism, where every challenger stakes Eth to begin challenge; a successful challenge would yield the challenger X amount of Eth from the contract’s balance, while an unsuccessful challenge would cost the challenger the Eth staked.
  • the AI is reversibly dead when its balance goes to 0; it can be revived by infusion of Eth to its balance.

We can further give the autonomous game contract the ability to learn from past games:

  • when the AI is tuned up to its highest difficulty and still loses more than it wins, it runs its learning algorithm on the recorded turn-by-turn game records to expands its difficulty spectrum;
  • this would allow the game contract to extend its lifespan by winning more games and thus more Eth.

This idea coincides with my earlier thinking on The Autonomous Dimension. I believe we are helping to create a new species that live entirely on-chain - a tribute to Primavera De Filippi’s Plantoid project and the ArtDAO idea by Simon de la Rouviere and Trent McConaghy.