Actually I see a lot of people investing quite some time in the database design - that was a motivation for doing the video. 🙂

I was mostly responding to the claim "data in isolation is pretty valueless if you don't know how to interpret the data". Normalisation is a very good tool to shape data into a form that makes interpretation quite obvious. So, perhaps, this video is mixing different topics?

Here is a customer. Here are his / her purchases. Which items would we recommend to him / her? Would we send her / him stuff without upfront payment? Only by credit card? <- that is what I was referring to.

Yes. All very interesting questions for a nice ERD.

I don't see how you can answer that question by looking at an ERD, sorry.

CREATE TABLE customer ( .., invoice_payment BOOLEAN );

I was referring to an algorithm that would take the purchase data and determine whether whether an upfront payment is required.

CREATE VIEW ... (...) 😉

Ah. I didn't know that algorithms implemented in DDL are data.... 😉

I get your point of course. To me, data and logic is all the same. The beauty of the relational model is that it already includes relational algebra to create and operate on derived data. You're doing the same, but you call it "API" and probably imply it is done elsewhere.