At the time of using TIGER data, there was not a good ‘place=’ definition for US subdivisions and apartments, and they were assigned using hamlet tags. It’s not clear that TIGER even differentiated between them.
The preferred way to handle these in OSM is as a named landuse= polygon, although they can be changed to place=subdivision also. It seems that there is no convention to use place=apartments.
I noticed that too, Mark. I live in Salt Lake City. I changed quite a few of them to populated place. Many of these are historic names or not known locally at all. They should probably be removed altogether. I think they represent very little value.
This would be nice for a ‘mobile’ MapRoulette challenge. As you walk / bike / drive around, it would ask ‘Is this area locally known as X?’