I totally agree, and there has been some discussion on twitter about this. The sample backgrounds, as constructed, lean into existing dnd tropes and conventions. I can see why they did this, as some people really like those conventions, but it undermines their attempts to open up if not do away with those same conventions. So whereas previous editions might describe Orcs as "brutish warriors," here we get a gladiator archtype that knows Orcish (and for bonus ickiness, they get "savage" attack and galdiators are historically associated with slavery).
There is huge potential in backgrounds doing collaborative worldbuilding work, but not if they just rely on the same tropes. They should emphasize that backgrounds are
specific and meant to correspond not just to general archtypes, but your character's origin in a particular setting. So the basic rule would be to craft your own, and then a sidebar they walk you through the process with setting-specific examples. You're not just an "urchin," you are urchin from waterdeep. What is waterdeep like, specifically? Why are there urchins in this city and how do urchins fit into that very specific setting? Or, you are an entertainer in Theros. What is the role of entertainment in that setting? What musical instruments are specifically popular there?
The way out of sterotypes is through
specificity and
worldbuilding. Admittedly, this is hard to present in an economical fashion and to standardize into a set of easy to pick up examples. But it's necessary if they want these aspects of the game to become less problematic. Background creation should be part of a session 0
procedure through which players contribute to defining aspects of the setting in these small ways. In the OSR, this is referred to as
"anti-canon" worldbuilding. Or, take a page from dungeon world, and allow players to fill in some of the "blanks" on the conceptual map of the world.