When is normalization too much?

[Dave Elo]

I had to make a new table for a new feature, where one selection was for the name of something in another table (say, phone). The phone table had an ID column and a name column, and the name values were stored in many other tables as references to their phone record... which is normally a bad idea, but it was made before my time.

Because the phone table is in another schema than the new feature table (I'll call it "ringtone"), I asked the legacy tech lead if I should be using the ID or name of the phone, because I'd seen that a phone name or two had changed before.

He said that I should use the name, because he said "use JOINs as least as possible" and "over-normalization is stupid"... I couldn't dispute this without being called argumentative, so I ultimately just said OK whatever.

I know the standard is to normalize to prevent changes from dereferencing data, but in this case I'd have to select from two different schemas... so which is worse in this case, JOINing from two different schemas or just using the phone name and hoping the rare case of it changing doesn't happen?