He isn't remotely proposing a new kind of science. All he is proposing is to do regular science, but use cellular automata as a framework for new theories (rather than PDEs).