A Code Without a Message: Photography's Topological Turn in the Age of Machine Learning

A media archaeological view of that other significant invention of Niépce – the internal combustion engine, or Pyréolophore – twenty years before the famous View from the Window at Le Gras might suggest that the self-driving car has one or two things to say about the future of photography. In the age of machine vision, we find the photograph retreating into vast visual datasets used in developing autonomous vehicles, where Barthes' 'message without a code' is seemingly inverted toward a total capture by the taxonomic logics of classificatory annotation. But a more fundamental rupture in photographic representation is under way, as visual meaning is mathematically reconstituted into a 'stochastic gestalt' within the semantic topology of high-dimensional vector space inherent to machine learning, demanding new critical methodologies capable of interrogating how images operate within computational assemblages.