I agree on your view the modelling more or less remains the same (but with more topical-clusters interconnected). But I haven't seen such schemas at work before

I agree on your view the modelling more or less remains the same (but with more topical-clusters interconnected). But I haven't seen such schemas at work before (most with 15 tables). The (big) examples of such I believe adds to elasticity of what/how something is - in our perception - for example I view schemas as sort of a large tree where each branch grows on every (1:n, 1:1 ..) relations. Some branches converge. Some are isolated branches. This sort of made me curious to search for database schemas with large number of tables. Sorry if my answer doesnt make sense - a short answer would be - if I want to write good c code - I would be probably look into linux\kernel source. I vouch not much can wrong from there.