Maybe along with Kolmogorov complexity we could measure "usefulness" of a bit string, i.e. how much utility an agent would agree to pay for it. The idea is that

Maybe along with Kolmogorov complexity we could measure "usefulness" of a bit string, i.e. how much utility an agent would agree to pay for it. The idea is that "complex" things, like life, are likely to contain solutions to many problems we care about. Of course that approach depends on the agent, but maybe you could get a general definition by referring to a general agent, like AIXItl.