A ‘filter bubble’ – a term originally coined by Internet activist Eli Pariser – denotes the dynamic in which people are only exposed to a small subset of ideas and perspectives. More alarmingly, there is the threat of an extrapolated version of reality where individuals only see content that is already familiar and accepted by them. This paper seeks to analyse the creation and development of such filter bubbles in social platforms by conceptualising the phenomenon within the theories of encoding and computed sociality. Using the example of Facebook News Feed, it will be shown that personalisation can, through the algorithmic process of filtering, lead to a situation in which the bubble becomes reinforcing and ever narrowing. The filtering system, therefore, becomes a cycle that both shapes and is shaped by user behaviour. Further, it is argued that the starting point in this cycle is already a heavily mediated state.