During the past century ever more sophisticated methods have been developed for constructing and programming computing and manufacturing machines. However, these methods are essentially forward processes that depend on intelligent human designers and programmers. They stand in stark contrast to Biology's methods of self-construction used to evoke the flexible information processor that is the mammalian neocortex. Understanding this radically different approach that uses algorithmic self-programming and construction could have enormous consequences for future computing and manufacturing technologies. In this talk we describe progress towards understanding these principles through detailed simulation of the development of the neocortex.