Our formulation of the problem is new, but it is based on existing theories of human intelligence. Firstly, inThe Analogical Mind (Gentner 2001), Hofstadter argues that analogy is the core of cognition. For example, we connect the concepts of a tree’s shadow, the dry region on the lee side of a mountain, and the metaphorical location beside an older, more accomplished sibling, and encode them using the word “shadow.” Moreover, analogy, that is, finding redundancy and creating abstractions, is at the core of data compression; in other words, it is the key to efficiently encoding information. Hutter has proposed that the problem of data compression--for instance, of the text on Wikipedia--is equivalent to strong AI (Hutter 2000). Relatedly, the long-standing “efficient coding hypothesis” in neuroscience posits that creating a maximally parsimonious model of information is a central task of the human brain.