Have just double checked, when in Full Space with user permission, more advanced features in ARKit APIs can be used like Plane Estimation, Scene Reconstruction, Image Anchoring, World Tracking, and Skeletal Hand Tracking.

Personally, I would like to explain Shared Space more like the Desktop, it allows multiple applications to be presented simultaneously. For example, Safari and a Messages are already opened in the Shared Space, launch your application will share the same view space along with those two (Safari and Messages), like WWDC demonstrated. 2. I think default and non-default refers to you can change the mode of your application presentation (Shared Space to Full Space or the other way around). There is no limitation in how your application is presented. But be aware, credential data like full hand skeleton is Only allowed to request in Full Space mode. 3. Passthrough means people can see their real surrounding environments around them (like the image showed above). 4. From the presentation, seems like Safari is running on Shared Space. 5. I think there is no limitation for what app can run on Full Space what so ever, maybe just need to create an immersive scene and handle the transition (I have not reach to that part yet, so I could be wrong...) I recommend go through these two sessions, which have a much better and detailed explanation: Session 10260: Get started with building apps for spatial computing and Session 10203: Develop your first immersive app. I just started going through the sessions and API documentations, please correct me if anything I explained is off. ^..^