I was looking at the exception_architecture documentation and noticed that the description for tassert  says:

Like uassert in that it checks for per-operation user errors

My understanding of tassert was that it was explicitly not used for user errors (that's what uassert is for), but for internal programming errors that indicate an error on our end. Can you please clarify that in the documentation?
 
Also - I might suggest expanding the top-level section of that document a tad with better summaries of when a developer should choose one over the other. For example, it gives little description for why you might choose uassert vs. massert. The overall document goes into more detail that I do find helpful, but I think most people tend to rely on the summaries at the top.