Yeah, I think .ic comes close to the way the real world works. The trouble with spice is that it calculates the steady-state DC condition before it "turns things loose", unless you add one or more initial conditions. I think you are also supposed to add UIC (use initial conditions) to the end of your .Tran statement, but LTspice seems to work without it.
Once the steady-state is established, many oscillators won't start in the absence of noise, which of course is present in all semiconductors (and resistors), but not in spice.