I wouldn't rely on this though unless using non-symmetric component values.
It is impossible to have anything other than non-symmetrical component values. All components even if marked with the same value have a tolerance. Even when you try and match components you only acheave a match to a certain number of places. Even if you acheave the impossibility of exactly the same component values, transistors with the same gain, and so on. They are physically diffrent things in physically different places. There willbe differences.
That's pretty much what they said in the Air Force, (plus they mentioned random circuit noise) would cause a preferential treatment such that the starting state on initial power up would always be the same, set or reset. But they also said don't worry about it too much, the circuit will oscillate, that's why it's called a astable multivibrator circuit.
When the circuit is first powered up, neither transistor will be switched on. However, this means that at this stage they will both have high base voltages and therefore a tendency to switch on, and inevitable slight asymmetries will mean that one of the transistors is first to switch on. This will quickly put the circuit into one of the above states, and oscillation will ensue. In practice, oscillation always occurs for practical values of R and C.