Daejeon, July 27–31
Storytelling is an essential aspect of the human experience that allows us to share memories, transmit knowledge, and shape our collective understanding of the world through language, symbol, and structure (Gottschall 2012). Traditionally, the storyteller was an individual with an understanding of cultural nuance and intention, and the audience was fellow humans in a shared narrative space. With the rise of algorithmic narrativity, storytelling is no longer limited to human agency (Rettberg / Rettberg 2025; Sharples / Pérez y Pérez 2022). Computational Narrative Systems (CNS) do not just support narrative production; they are becoming storytellers in their own right, generating stories, characters, plots, and even emotional arcs based on probabilistic logic and encoded cultural assumptions (Li et al. 2024; Pérez y Pérez / Sharples 2001). As a result, we are witnessing the emergence of non-human storytelling for humans (Borges et al. 2024; Coenen et al. 2021; Li et al. 2025).
Narrative Nubs is a collaborative research project focused on reimplementing existing CNSs to preserve their essence while minimizing their complexity (Montfort / Kusnick 2026). These reimplementations, known as “nubs”, are meant to support a range of exploratory activities, including research, education, and creative practice. This poster proposes a future visual interface – Visual Narrative Nubs – to exhibit the history of CNSs and provide a tangible “playground” for researchers, pupils, and the interested public (Kusnick 2026).
Figure 1: Grid-view homepage of Visual Narrative Nubs, envisioning two examples of small multiples to visually compare nubs.
Visualization techniques can help answer questions arising from this narratological shift For example: “How can authors maintain creative control and aesthetic intention when sharing authorship with algorithmic agents?” Or “How can computed stories be evaluated so that they stay understandable, interesting and engaging for humans?”
Distant Reading (Moretti 2013) methodologies and visual tools have enabled large-scale analysis of text corpora (Alharbi / Laramee 2019; Jänicke et al. 2017; 2015) or Distant Viewing (Arnold / Tilton 2019). With nowadays common trained machine learning models the provenance of the sources as well as copyrights are getting disguised and the contents and styles mixed so that we are observing similar processes in the generation of content (e.g. LLMs trained on text) and narratives (e.g. CNSs trained on existing narratives) – thus entering an era of Distant Writing (Floridi 2025).
This poster presents a project investigating whether we can extend this new paradigm to a visual authoring process itself: Algorithmic storytelling informed by macro-level narrative structures and audience feedback loops via visual interfaces?
To ground this question in practical, historical, and experimental terms, we will engage with the reimplementation of existing CNSs, canonical e.g. MEXICA (Pérez y Pérez / Sharples 2001), or less known e.g. TAILOR (Smith / Witten 1991) ones. By reconstructing and critically comparing CNSs, we will explore how different models conceptualize plot, character, and creativity (Montfort / Kusnick 2026). We will develop a comprehensive visual dashboard to compare the systems’ input parameters, outputs, author/AI interactions, and narrative trajectories to both critique and inspire new forms of storytelling that are algorithmically assisted but culturally grounded.
Figure 2: Multifaceted visual interface for the TAILOR nub (Smith / Witten 1991) detailing examples for possible visualizations.
This project investigates how AI-generated stories can be better understood, authored, and critiqued by integrating visualizations with existing CNSs. While contemporary generative text systems offer ease and speed in storytelling, their outputs are often rather simple, homogeneous, and detached from cultural nuance or structural awareness. To counter this, the project seeks to connect current practices with historical, contemporary, and future CNSs, offering a foundation for both innovation and critical reflection. By reimplementing legacy systems, we can remember and understand the development of architectural principles of algorithmic storytelling.
Therefore, a core research objective is to develop and evaluate interactive visual interfaces that support both the analysis and creation of computational narrative systems as well as their machine-generated stories.
Through a newly composed overarching design space and metrics, the different narrative approaches will become comparable (see Figure 1). With these evaluation measures at hand, a comprehensive visualization-based interface can be developed (see Figure 2).
This dashboard will make visible otherwise hidden characteristics and patterns of the different systems, and, through their reimplemented functionality, they become tangible through a “playground” that can serve research, domain experts, and interested lay users to understand, challenge, and reshape algorithmic storytelling.