[CamperBob2]

The key factor is the line width, or the range of frequencies over which the transition can be stimulated. The ratio of the stimulus frequency and line width is one way of expressing the resonator Q factor. In general, the lower the line width for a given transition, the higher the Q, the better the signal-to-noise ratio, and the more stable the resulting clock. (Imagine how much more precisely the frequency of a large bell could be measured compared to a cymbal or something else with a broader acoustical spectrum.)

Cs or Rb clocks give you a line width of a few hundred Hz at 9 GHz (Q=roughly 100 million), while quantum transitions in optical clocks can achieve line widths on the order of 1 Hz in the PHz region (equivalent Q in the quintillions.) There is a lot more to building a good clock than high Q, but it's a very important consideration ( http://www.leapsecond.com/pages/Q/ ).

What caught my eye is the ringdown time of the stimulated optical resonance, apparently in the hundreds of seconds. They talk about line widths in the GHz range, but that seems to refer to the laser rather than the underlying resonance being probed. It would have been interesting to hear more about what they expected regarding the actual transition line width. Probably the information is there but not in a form that I grokked, given insufficient background in that field.