I like these as examples of unknown unknowns because they remind us just how deeply buried some ideas are and how radical their reframing can be. They were unknown unknowns because they re-ground thought, vision, art, and the like: Newton’s discovery now lets us see gravity at work everywhere, and impressionism lets us see vision not in terms of connecting to standalone objects, but as an interaction between our eyes and light in all its colors; Monet’s paintings of the Cathedral at Rouen are, as every Intro to Impressionism course says, portraits not of the cathedral but of the impression made on us by the light bouncing off of it and onto us; it shows us vision itself.
Current Conceptual Model of Knowledge
Far more mysterious than Rumsfeld suggested, real unknown unknowns are unknown because they don’t fit into our current conceptual models. Middle Eastern regimes transferring weapons would fit neatly into our current structure of knowledge. The deeper unknown unknowns reshape that structure, sometimes fundamentally, as I think Impressionism did, and as quantum mechanics and deep learning are currently doing. They change the
architecture of knowledge itself. Rumsfeld’s example of an unknown unknown would be frightening in and of itself, but would skirt around the truly terrifying and awesome truth that the universe doesn’t care about knowledge or about us tiny creatures wandering the dark with our dim flashlights illuminating tiny circles that we then call knowledge.
AI’s ability to make predictions based on relationships among data that we often simply cannot understand casts a different light. AI may be making it easier for us to accept that, in the final analysis, the unknown unknown is the universe and everything in it.