Various research strands have converged around a set of closely connected ideas that make similar use of information-theoretic and computational complexity principles to account for and formally model emergence and causality across different scales; what I will call here the emergent multi-scale causality (eMSC) approaches. The framework is gaining popularity—e.g., notions such as "causal emergence" and "synergistic information" in science, and some prominent versions of ontic structural realism in philosophy. However, it has been noted by numerous authors that the methodological and metaphysical considerations that explicitly and implicitly underlie the modelling decisions of the framework remain unclear. In this context of concerns, after characterizing the core principles of eMSC, this article argues that the realist stance of the framework towards effective theories and their taxonomies is problematic. The main point is that, eMSC's strategy towards scientific realism has to be defended on two fronts, a local one and a global one, but the tools that the framework uses and develops can only focus on the local front, overlooking the global one. And this is something that allows for a multiplicity and relativity of ontologies that the realist stance of the framework wants to prima facie avoid. eMSC has two options (which can be combined). First, defend and/or develop a version of scientific realism that could be squared with a certain degree of multiplicity and relativity of ontologies. Second, explore some research avenues coming from the broader computational modelling field. By raising these concerns, some possible extensions for eMSC are noticed, while at the same time, those computational modelling research avenues are connected with the genuine philosophical questions that eMSC is concerned about.