dmstfctn – Fango 1000

Fango 1000 is a fully on-chain emergent narrative game where players compete to tell the best stories about AI. In the game, set in a society similar to the Middle Ages, a trove of one thousand chronicles has been recovered by a powerful Institution – each chronicle detailing how humans from a future society relentlessly train AI systems for mundane tasks. Players join one of three factions with conflicting traditions and views on the implications of these chronicles. These factions compete against each other for control over the narrative of the coming role of AI in human societies by trying to tell the most memorable stories using only data sourced from the chronicles. How they tell their story, however, is up to them. The best stories will last forever, the worst stories will be burnt.

The data from each chronicle is generated by players of a separate AI training game called Godmode Epochs, developed by dmstfctn and played thousand of times since its release in 2023. Each play-through of Godmode Epochs constitutes a factual dataset to be used by players of Fango 1000 as raw material for their stories. The data is provided to a blockchain to be used in the game, and all players’ actions from the game are stored on-chain as a transactions between wallets, alongside the game rules and mechanics which are stored on-chain as smart contracts.

Fango 1000 is a toy implementation of the ideas hypothesised in the essay Large Lore Models (0xPARC, 2023) written by dmstfctn, Eva Jäger and Alasdair Milne. The essay proposes the basic configuration of a tool, a Large Lore Model (LLoreM), able to facilitate collective and emergent narratives in fully on-chain simulations. The LLoreM would link a permanent record of actions generated within an on-chain simulation or game to the stories of its users or players. The function of this design is to maximise the affordances of the blockchain’s immutability while preserving the inherent mutability of human narrative, and its contestability through multiple perspectives. The LLoreM is also advanced as an alternative to automated narrative generation systems like GPT, emphasising the creative potential of human co-operation and collectivity.

Lascia un commento

Il tuo indirizzo email non sarà pubblicato. I campi obbligatori sono contrassegnati *