Thank you. Again, I was mostly confused by this point not being stated as related to user talks only. (Is this only a temporary confusing state of the document or is there the risk that design decisions taken by having in mind only user talks will affect also completely different situations in later stages of the project?)
Back to the point: it's not broken, it's a wiki. Everything can always be improved, including talks, and I'm not aware of this ever causing any problem, it's just an unjustified assumption. Long threads become completely useless if you can't refactor them: think of the heated bugzilla discussions where it's impossible to distinguish noise, sub-threads, simple corrections, plain off-topic etc. from the meat. Mere chronological listing of posts as is is a completely dead concept which belongs to super old forum systems only, nowadays all discussion/comment systems allow a big degree of what we call refactoring and wikis are very modern in this; if you remove refactoring, you need to invent a shole new system to make discussions manageable and I'm not sure it's worth it.
Breaking refactoring might be a necessary evil at some point, but it mustn't be an aim.
I think the refactoring thing is going to be a longer term discussion but it is one that I'm thinking about heavily.
And yes, the document is in a state of flux. I've actually got a (much) larger version of it but it's a mess and this is the best that I could pull together. I'm going to be focusing on making some mockups to help explain the concepts better; I fear that much of this may make sense only with visuals (for example, how watchlist entries might work).