From what I gather it's the vintage ribbons that are most susceptible to damage.  I'm not sure there's any sort of rule of thumb or consistency in behaviour. You're brave man lol.

Provided you are careful about turning on the phantom power only after the ribbon microphone has been fully connected, there should be no problem. That is, until someone wheels a cabinet over the mic cable and one of the signal conductors gets shorted to the screen. Pop!

I've seen it happen. Not with my mics, luckily.

Turning on phantom is just as risky in many cases, if you believe the risk. Some designs put a double pole switch in the circuit to apply the voltage so that when there was no phantom - the two resistors were not padding the circuit while other designs put the resistors in circuit permanently and switch the volts to the centre point. The double pole switch offers the possibility of applying pin 2 or 3 voltage first.

With a transformer blocking DC paths, where are we getting this ribbon destroying current from. Realistically, with any fault condition, the only possible way it can happen is with we apply, but then importantly remove, a DC source across the transformer. Bending an loose fitting XLR to one side when shoving it in could apply power to pin 2 before pin 3 and 1, so could cause this too.

Normally the phantom will be applied across the transformer secondary, and with a centre tap, connected to ground - if one or the other section of the coil, referenced to the centre tap has DC on it that the other doesn't then if that then is removed, DC current flows and back EMF could induce a voltage in the primary, directly connected to the ribbon. The voltage could be high but the current will be low. The result is a jolt to the ribbon - it will physically 'twitch' - The same kind of twitch you might get if somebody bangs the case, or puts it near a drum - that kind of thing. Too much twitch and it could stretch beyond it's capacity to recover. This I think is the reason for vintage (expensive) ribbons being protected - not just from phantom but wind, idiots banging them and using them on the wrong sound sources. 

In suspect people running phantom through patch bays and re-patching hot with a ribbon could be the easiest way to do the damage - especially if it's done to the same mic repeatedly - but like all things ribbon, they get tired and sound different, and you are just speeding it up. 

I'm happy the occasional mistake won't be terminal.

It's the transient fault current that does the damage.

Most passive ribbon mics have an internal transformer (typically 1:55) that acts as both an impedance converter and a voltage multiplier between the ribbon (<1 Ohm) and the output cable. When the mic is connected normally to a pre-amp input with phantom power enabled, the transformer output floats happily at phantom power voltage. No d.c. current flows, so there is no drop through the 6K8 feed resistors from the 48V supply.

A simplified picture of what happens if a short-circuit causes one output conductor to be suddenly connected to ground is that the fault causes a 7mA current pulse to flow from the 48V supply through the 6K8 feed in that conductor. This current pulse is ratioed 1:55 by the transformer, resulting in a 390mA pulse being back-fed into the ribbon. Various winding inductances and return route impedances catch up with the pulse causing it to be short lived, but that magnitude of pulse is usually enough to blow the ribbon on most microphones.

Although different designs of passive ribbon microphone will have different transformer ratios, inductances and other parameters, the fault mechanism would be similar.