The RM ( whether Codd's, Date's, Darwen's, Pascal's, McGoveran's or some
other authors whose names I can't remember) has for many years been
over-complicated by many of the same people who now push "nosql",
probably misunderstood too by most of them, with an industrial theory
that is yet not fully formed.
This seems very ironic to me because the nosql movement seems to be very
concerned with simplicity of implementation. The logical simplicity of
the RM as I understand it remains not well understood at all and some of
the famous popular thinkers and writers acknowledge this every so often.
They also get pilloried, wrongly, from time to time for miss-steps. I
suspect the difficulty of explaining it to people with an existing
computer science heritage makes it seem bigger than it really is.
Whereas I imagine that 10 year-old children without that baggage could
learn it easier if it were only shown to them, minus cosmetics such as
html. The right comparison is hardly between programmers and dba's.
The cruel fact is that people of both those ilks will be known far in
the future as a lost generation.
And, as is common, here we have 'sql' being conflated with 'relational'.
I think you are right about the 'sturdy construction'. I'd like to
see a coherent description of a 'logical nosql model'.