>uniform random strings’ K complexity is only _tightly concentrated around_ the strings’ length plus a machine-dependent constant

What is the distribution of the complexity of a string? Is there some Chernof-like bound?