But, to state the obvious, anonymous items in computing (e.g. anonymous functions, or untitled/unaddressed draft emails etc.) generally need to be assigned an id for the software to keep track. The practice of avoiding natural keys and assigning unique ids is analogous. No paradigm is implied.
In programs, anonymous objects always have an unique ID automatically assigned - by the very virtue of being separate entities. Usually those IDs are what we call "references" in programming.
I'm talking about the level of e.g. programming language implementation. A lambda needs to be assigned an id, for example, but it's not a reference that the user can manipulate. It's an implementation detail, as are surrogate keys.
You don't manipulate the reference directly (i.e. you shouldn't learn its numeric value (though you can - useful for debugging), and you can't change it) - but you do use it indirectly all the times with ==, eq, or however the referential equality operator is called in a given programming language.
This is what, given (excuse the Java):
MyObject a = new String("choo choo");
MyObject b = a;
MyObject c = new String("choo choo");
allows you to observe that:
a == b;
a != c;
b != c;
even though:
a.equals(b);
a.equals(c);
b.equals(c);
under string comparison.
This is exactly like surrogate keys. An implementation detail that's crucial to the meaning of the whole thing.
