My main problems are:

1. Arguments frequently end up discussing the semantics of these concepts rather than using them to explore ideas and solve problems (see the thread on github with 20 differrent definitions of problem space and subdomain)

2. The concepts are vague/confusing so they can't be utilised in a useful way. i.e. a subdomain lives in the problem space but we can't define problem space so we're stuck in a stalemate and thus subdomains in DDD are the elephant in the room.

If DDD had clear definitions, maybe Uber wouldn't have created their own definitions.

I think Wardley's concept of landscape is much better. Subdomains are part of the landscape. There is a current landscape, which exists in a climate and we are seeking to change it to solve problems.