Based on our experiences from ongoing collaborations with computational engineers over the course of six years and two interdisciplinary research projects, with this article we suggest that the building of collaborations between anthropology and computational sciences that alter disciplinary boundaries and bridge epistemic differences can be accomplished through three levels of engagement: a shared research project, becoming involved in each other's theoretical universes, and crafting physical spaces for shared intellectual practice. Taking an empirical point of departure in our colleagues’ attempt to cross the methodological and epistemic divide between engineering and anthropology through game theory, we introduce how the distinction between the ideal, the real and the actual serves not only as a model of our collaborations but also as a generalisable model for future collaborations across anthropology and computational sciences.