The part I find particularly most frustrating is when you're just messing around, and they have that "predictive" thing going on, that's supposed to show you what they think you're most likely wanting to see. (e.g. larger streets and landmarks have labels, whereas smaller side streets or whatever have been left blank for better overall legibility.)

However, there are times where I have simply wanted to see the street name of an unlabeled street, and the amount of zooming in, out, and panning around just to HOPE the renderer fills in the name is ridiculous.