The current process is error-prone too. Since no one mentioned this, I wanted to give a specific example. A recent name change from Trillium subgenus Sessilium to Trillium subgenus Sessilia withdrew a disagreeing ID from each of 682 observations. This introduced errors into the activity timelines of some of the observations.
Using a tool developed by @pisum:
I reviewed 430 observations and reissued 305 disagreeing IDs. Since I wase unable to use the Identify tool, this cleanup operation was expensive (in terms of tima and effort). In this case at least, the value of the name change turns out to be insignificantly small. In retrospect, I should not have asked for the name change.