How Lionel Davoust writes fiction books using the LYT Frameworks (Obsidian)

Lionel Davoust’s core claim is that long-form fiction becomes manageable when the writing process is treated like a system for handling “multi-dimensional” story knowledge—ideas, uncertainties, character arcs, and world rules—rather than a linear outline. He frames the central challenge as the “tesseract dilemma”: fiction asks writers to compress a four-dimensional construct (characters’ motives, moods, causes, and consequences) into the one-dimensional experience of reading sentence by sentence. The payoff, he says, is a workflow that lets story structure emerge while keeping the writer oriented amid complexity.

Davoust builds his approach around the LYT Frameworks in Obsidian, using a loop that turns raw ideas into increasingly testable story hypotheses. In his model, a “light bulb” captures an idea; refining it produces a hypothesis (the “meaning of the arrow”); and writing forces constraints to lock in over time—his example being that if a character dislikes cake in chapter one, that dislike should persist into later chapters. He also emphasizes the “red question mark”: the hard questions that block progress, such as where a scene takes place, what the characters want, and what the scene’s goal is. First drafts, he argues, are less about producing answers and more about surfacing and sequencing those questions; rewriting then becomes the work of improving the answers.

In practice, he describes how Obsidian’s atomic notes and linking make emergence feel natural. Random scene fragments, punchlines, and world details get captured as separate notes, then reorganized into “maps of content” that function like modular tables of contents for characters, arcs, and the world itself. He reports that this structure makes it hard for ideas to “stall,” because new notes keep feeding the graph, and because the unknowns are explicitly identified as parts of the system rather than floating in his head.

He also stresses constant refactoring. Because story elements influence one another across chapters, he treats the system as something that must be continuously rebalanced—like refactoring a living model rather than freezing a plan. He warns that heavy use of block references can slow refactoring; his workaround is to create new notes and embed them, since renaming and relinking is easier when the structure shifts. He uses emojis in filenames as quick “jumping points” to mark questions or ideas he wants to revisit.

Davoust’s screenshots illustrate three main graph clusters: the intricate map for the book in progress, a cast-under-construction area for characters, and a separate workspace for workshop notes and observations that later feed into the main canon. He repeatedly returns to a guiding principle: worldbuilding is only valuable insofar as it helps write the story. The system ultimately supports a later transfer step—moving published canon into a franchise-level vault with the same structure.

Beyond tooling, he positions the method as psychological relief. Compared with file-and-folder workflows that can feel messy and anxiety-inducing, the graph-based approach gives him an overview of the unknowns and lets him “pin” questions so they don’t haunt working memory. The result is a workflow designed to keep momentum—trusting emergence while maintaining an overarching purpose: write the damn thing.