In living systems contrary to those of low level chemical systems (reactions) the failure of elements to bond is expressed as an improbability of the occurance coupled with the brevity of the survival of the new species (the atomic scientists word ...not mine!) what is not energy conservative ceases to exist after brief periods. Life on the otherhand persists through mutations and macroscopically results in a materially different system that can then propagate its particular mutation set to its progeny. Viruses do this, the most archaic Bacteria do this, plants and people do this. Stars do not do this, snow flakes do not do this, a catalyzing reaction between two elements or more complex chemicals does not do this. The evolution of what I propose as non life is rigid and fixed once the energy regime for a reaction is set, once oxygen and hydrogen meet under the right temperature and pressure constraints water WILL form. Abberant species of hydrogen/oxygen molecules are probabilistically denied to exist for any important (meaning allowing continued reaction) duration. Similarly, stars are not alive because they do not have the ability to continue their stellar burning beyond the limits of their acquired Hydrogen fuel, the initial mass value or addition of new material does lead to a change in the evolutionary trajectory of the star (will it be a white dwarf? a neutron star or a black hole? or will it go NOVA?) but these changes are precisely predicted ones. Unlike true life, who's evolved state under mutation holds the possibility for highly unpredictable changes in the final entity, particularly after generations of mutation and propagation.
So formerly my proposal for a new definition of "life" is:
Life, any entity that under the process of internal random mutation changes how it interacts with the outside environment from generation to generation despite having at any given moment the resources (even if fixed in quantity) required for its survival.