'The Life of the Cosmos' is an argument for a new way of looking at fundamental physics and cosmology. Its highlight is cosmological natural selection (CNS) but the underlying principle is 'relationalism', an idea derived from Leibniz (and basic to relativity theory), in which physical properties are relational rather than intrinsic. Space and time are principles relating physical things to each other, not absolute backgrounds within which physical things interact but do not themselves take part in interactions.
Lee Smolin argues that the universe is self-organised, a bit like an organism or an ecosystem (though nothing is gained by saying that the universe is actually 'alive'). The universe has evolved and possesses homoeostatic properties that keep many of its components in states far from thermal equilibrium. Another relational principle learnt from Leibniz is that a view of the whole universe as a far-from-equilibrium system does not imply a view-point from outside the universe.
CNS is a Darwinian solution to the 'special-tuning problem', which is the vast improbability that the universe should be set up precisely to suit life (as it seems to be). The answer is that a mechanism of natural selection can produce design without a designer or blueprint. In the case of cosmology, the key is the production of black holes. Assuming each universe is born as a black hole within another universe, then universes take part in a copying competition and the most typical universe (which we may assume ours to be) ought to belong to the lineage with the most fecund universes.
This prediction is testable: by changing any of the parameters of physics in our universe, one will produce a different universe with fewer black holes. It so happens that carbon chemistry (and, hence, carbon-based life-forms) is a natural by-product of maximising the number of black holes in a universe. Thus CNS is a rational alternative to the anthropic cosmological principle.
This is altogether a sensible and well-made argument and brilliantly original. Highly recommended.