Today I got no research done at all. This behavior is deprecated.
After I spoke at MIT CSR Science Lunch, Burles noted that if (a) the K+A events are triggered by mergers or interactions, and if (b) morphological merger evidence lasts for N dynamical times, where N is relatively independent of dynamical time, then (1) some significant fraction of K+As should show morphological evidence of interactions (duh!), and (2) the fraction should be a linear function of dynamical time, which can be estimated (at least crudely) from the SDSS data themselves. Smart idea, lets do it.