Thèse Substrats Computationnels pour la Multimédia Interactive H/F - Institut Polytechnique de Paris Télécom Paris

Établissement : Institut Polytechnique de Paris Télécom Paris
École doctorale : Ecole Doctorale de l'Institut Polytechnique de Paris
Laboratoire de recherche : Laboratoire de Traitement et Communication de l'Information
Direction de la thèse : James EAGAN ORCID 0000000271077035
Début de la thèse : 2026-10-01
Date limite de candidature : 2026-04-15T23:59:59Les logiciels d'aujourd'hui sont conçus avec la notion que, pour transformer le comportement de ses logiciels, il faut programmer - que ça soit en écrivant du code, en utilisant un outil de type programmation visuelle (c.f., No code, blocks, flot de données), ou en s'appuyant sur une IA. Ce projet vise à déverrouiller l'espace entre les deux en s'appuyant sur les principes suivants : 1/ un expert de domaine est capable de s'exprimer à travers des représentations notationnelles complexes de son domaine d'expertise, 2/ les substrats interactifs[@mackay25] et l'interaction instrumentelle[@beaudouin-lafon00] forment un socle théorique qui permettrait la création de versions interactives de ces représentations, et 3/ des utilisateurs experts de domaine peuvent exprimer de la computation à travers ces substrats interactifs sans écrire du code.

Pour orienter cette exploration, ce projet se concentra sur le domaine de développement de jeux vidéos. Les jeux vidéos allient de l'expertise dans plusieurs domaines, avec des outils pointus sur ces sous-domaines. La thèse se déroulera en trois phases : dans un premier temps, il s'occupera d'une analyse approfondie des besoins utilisateur, des formalismes utilisés dans le domaine et des façons d'exprimer les besoins interactifs aux programmeurs. Dans un deuxième temps, il appliquera l'interaction instrumentale et la notion de substrat interactif pour en créer des substrats et des instruments capables d'exprimer les besoins interactifs identifiés dans la première partie. Enfin, dans un troisième temps, la thèse évaluera les techniques proposées afin de non seulement identifier leur aptitude à satisfaire les objectifs des utilisateurs mais aussi de façon qualitative pour mieux avancer les connaissances sur l'application de ce type de stratégie.

In many disciplines, domain experts develop specialized notations, representations, and tools to reflect their deeply-honed expertise and understanding of their particular application context. Adapting current digital tools to these contexts, however, requires programming-such as by working with a programmer, a no-code tool, or relying on an AI-companion. As such, current digital tools implicitly create a divide in which one is either a \_user\_ or a \_programmer.\_ This project aims to democratize computation and empower users by transforming the way that we build software. It will explore the space between user and programmer and to propose and evaluate flexible abstractions that lie in this in-between space.

The goal of this project is to explore and propose alternative interactive abstractions for which computational extensibility becomes a byproduct of use. Most existing approaches attempt to address this problem by making programming easier, so that non-programmers can in effect become programmers. This work aims to take a different approach, by exploring ways that non-programmers can nonetheless express computational concepts through their domain contexts.

Many specialized disciplines develop domain specific notations, representations, and tools that allow experts to express complex ideas without relying on general purpose formal languages. Musicians over centuries have developed complex notations to capture the rich space of musical expression. Mathematicians have relied on various notations to describe abstract concepts since well before the dawn of computing, and choreographers capture movement through rhythm and space through Labanotation[guest05].

To ground our research, we will focus on the field of game development. This discipline is particularly interesting because it combines various forms of domain expertise: art, narrative, design, and sound. Moreover, it has a rich ecosystem of tools to support the creation of assets in these domains; the binding of assets into interactive behavior is typically mediated by game engines and their computational representations. In practice, this places the expression of game logic triggers, state changes, progression, and interaction rules within notations that carry programming concepts, even when they are presented visually.

Existing tools that aim to reduce the barriers to entry, such as Ren'Py (renpy.org), RPG Maker (rpgmakerweb.com), Scratch[@maloney10], and node-based scripting environments, constrain genres or translate programming into visual form. They still require creators to adopt computational abstractions that may not align with their domain thinking. As a result, toolchains concentrate agency over game behavior in the hands of those fluent in the engine's computational abstractions; others contribute assets but must negotiate for changes in behavior[@blow04].

Empirical work on game development practice highlights the central role of tools and technical constraints in shaping creative outcomes[@berg-marklund19]. This suggests an opportunity to explore alternative interactive abstractions that allow artists, writers, and designers to directly express behavior in terms that emerge from their own representational practices, rather than through a translation into general-purpose computational form. Video game development thus offers a promising test bed for investigating how instrumental interaction and interactive substrates might support computational extensibility as a byproduct of use, rather than as an explicit act of programming.

To achieve these goals, this Ph.D. will focus on exploring new interactive metaphors and concepts to build such interactive systems, supporting the creation of new interactive tools in ways that fundamentally transform the role of code and computation in the authoring of interactive systems. It will draw on the under-explored but promising principles of instrumental interaction [@beaudouin-lafon00] and of interactive substrates [@mackay25], and build on Klokmose & Eagan's prior work on Webstrates [@klokmose15] and MyWebstrates [@klokmose24].

Diplômé·e de l'IP Paris, master Interaction Graphique et Design (IGD) de la mention informatique

Conception centrée-utilisateur

Programmation orientée-objet et de systèmes interactifs

Développement web