I would argue one should mainly focus on what you would use yourself, and only if you enjoy it and are running out of things to add, move onto other types (here

I would argue one should mainly focus on what you would use yourself, and only if you enjoy it and are running out of things to add, move onto other types (here it can be tricky to know what is really needed/used). The reason I recommend starting with your own priorities, is that this (when done by enough people) adds what is really being used the most by everyone. Mind that the existence of data also creates purposes for it. Let's say that features for blind people are mapped very well in your area and there is a developer involved with that community, s/he could create software to make use of it. For this example, StreetComplete asks for things like tactile paving at bus stops and crossings and auditory signals at traffic lights. One thing I don't see the use of at all but that is a ton of work in most areas is the "building kind" and "number of levels for this building". One could color-code the building type or render an approximate height for each building on 3d maps, but really, most maps don't do either of this, and these two quests are far more work than the dozens of other quests combined.