On-chain Collaboration World Building: Sentences Prototyping System

A notable feature of the on-chain game design field is its high emphasis on collective decision-making. To explore this characteristic, we developed Sentences, a collaborative world-building tool based on on-chain technology. The core structure of the system revolves around the original structure with only additional branching narratives, utilizing MUD construction to provide a lightweight environment for collective narrative creation. In Sentences, each new game launch generates a brand new world, and players enrich this world by gradually adding legends.

Design Motivation

In autonomous world spaces, a common task for new projects is to conduct collective "world building", which means creating a shared lore system that determines how players experience this world. This activity is often informal and unstructured, but it can also be supported through organized prompts and exercises that help world builders shape the structure and coherence of their worlds.

These exercises themselves constitute a world. They establish an information system among the narrators, which is a set of rules that promote the development of narratives. These rules are the world itself, creating an open space of reality for the birth of new ideas. The choice of exercises inevitably influences the narratives created: a set of questions may assume a potential space for answers; a prompt in one direction may limit exploration in another direction.

Therefore, Sentences can be viewed as a "world-building world", a primitive sandbox for collaborative knowledge generation. Its design intention is to serve as a laboratory for testing new ideas, and thus it must be concise and limited. In the world of Sentences, there are no other forms of interaction or contribution besides expanding the narrative. The world is constructed linearly but can branch out in multiple directions from the initial seed.

Core Mechanism

When players load the Sentences client, if there is no ongoing world, the system will provide the option to generate a new world. Once generated, players will receive a prompt guiding them to use core mechanics to develop the story. For example:

"This society values nature the most" "This civilization completely relies on horizontal" "This group is built on money."

Once the world is generated, it will enter an active state. The author has a fixed time (set to 20 blocks, at regular intervals) to propose new supplementary content for the story. After the time ends, it enters the second phase, where participants will vote on their favorite proposals. After the voting ends, the most popular proposals will be incorporated into the story, and then the process will start again.

If no proposals are made within the specified time, the world will "perish". This perish is not permanent: the world simply archives other already perished worlds. Since all proposals and votes are recorded on-chain, each world retains a complete history of possible development directions, and these "perished" branches represent a series of possible parallel universes.

Technical Implementation

The state of the narrative is managed by two interconnected systems: one responsible for generating new narratives, and the other handling time, voting, and proposing new content.

The initialization of the new story consists of sentences randomly selected from a preset prompt list. These prompts are generated using a simple replacement syntax script. Future versions may replicate this script in the contract to provide more diverse starting points.

After the story is initialized, the initial prompt is added to the chain, and a new proposal period begins. This cycle lasts for a certain number of blocks, defined by variables, and is set at the beginning of each new proposal period on the active story.

During the proposal period, participants can suggest new narrative entries and vote on expansions. Proposals and voting are processed by a specialized system, which validates these contributions based on the end time of the cycle. Each proposal points to a "parent," and successful proposals form a linked list.

At the end of the proposal period, the system calculates the number of votes for each proposal. In case of a tie, one of the proposals is randomly selected. If no proposals are made, the story ends and is archived, and players can choose to start a new story.

Application Scenarios

Sentences are best used in meetings with 10-50 participants, who may or may not know each other, but gather together in the spirit of creating prototype legends. For example, events can be hosted on social platforms, with the goal of generating five new worlds within an hour.

As an independent tool, Sentences may not be very exciting; it feels more like a tool rather than a complete world. However, as a modular component, the narrative generation mechanism of Sentences can integrate well into richer role-playing games, dynamically weaving the game structure throughout the gameplay. Its voting mechanism can also be used to support the inclusion of generative output in player-generated prompts.

A limitation of the current version is the specificity of the initial prompts. While this can be adjusted for different deployments, further expanding the game's prompt generation mechanism to align more with the openness of the narrative section would be an interesting direction for development.

Source of Inspiration

The design inspiration for Sentences comes from text-based games, improvisation, legends, and tabletop role-playing games. During the development process, we referenced specific projects such as Max Kreminski's Epitaph (a fantasy narrative generation game) and Kate Compton's Tracery alternative grammar tool. These projects use simple randomness and modular approaches to create complex branching narratives, and we are particularly interested in the potential applications of these primitives in collective scenarios.

Although Sentences is primarily designed for world-building, it can also be applied to other constrained collaborative writing scenarios, demonstrating a wide range of application prospects.

Future Outlook

A clear direction for the expansion of Sentences is to allow the creators of the world to set specific rules that determine how new sentences are added to the narrative. For example, these rules could modify the voting system, requiring a certain number of players to participate in order to continue the narrative, or adjust the voting time to change the pace of narrative development. In this way, Sentences not only becomes a tool for world prototyping but also a tool for prototyping specific story styles and dialogue structures.

Another interesting development direction is to increase the options for generating parallel worlds from the "dead" branches of the current world. The best way to implement this mechanism may be to generate a new "New World" contract that contains pointers to the old branches, rather than generating a new world from scratch.

Through these expansions and improvements, Sentences has the potential to become a more powerful and flexible collaborative world-building tool, offering creators and game designers more possibilities.